diff options
| author | Dave Airlie <airlied@redhat.com> | 2018-11-19 02:07:52 +0100 |
|---|---|---|
| committer | Dave Airlie <airlied@redhat.com> | 2018-11-19 02:07:52 +0100 |
| commit | 9235dd441af43599b9cdcce599a3da4083fcad3c (patch) | |
| tree | 5f8a79cc2d378f05e807c6c5e388394b8e86319d /drivers/gpu/drm | |
| parent | Merge tag 'drm-misc-next-2018-11-07' of git://anongit.freedesktop.org/drm/drm... (diff) | |
| parent | drm/amdgpu: log smu version (diff) | |
| download | linux-9235dd441af43599b9cdcce599a3da4083fcad3c.tar.xz linux-9235dd441af43599b9cdcce599a3da4083fcad3c.zip | |
Merge branch 'drm-next-4.21' of git://people.freedesktop.org/~agd5f/linux into drm-next
New features for 4.21:
amdgpu:
- Support for SDMA paging queue on vega
- Put compute EOP buffers into vram for better performance
- Share more code with amdkfd
- Support for scanout with DCC on gfx9
- Initial kerneldoc for DC
- Updated SMU firmware support for gfx8 chips
- Rework CSA handling for eventual support for preemption
- XGMI PSP support
- Clean up RLC handling
- Enable GPU reset by default on VI, SOC15 dGPUs
- Ring and IB test cleanups
amdkfd:
- Share more code with amdgpu
ttm:
- Move global init out of the drivers
scheduler:
- Track if schedulers are ready for work
- Timeout/fault handling changes to facilitate GPU recovery
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181114165113.3751-1-alexander.deucher@amd.com
Diffstat (limited to 'drivers/gpu/drm')
196 files changed, 5538 insertions, 5318 deletions
diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile index 576ba985e138..7f3be3506057 100644 --- a/drivers/gpu/drm/Makefile +++ b/drivers/gpu/drm/Makefile @@ -11,7 +11,7 @@ drm-y := drm_auth.o drm_bufs.o drm_cache.o \ drm_sysfs.o drm_hashtab.o drm_mm.o \ drm_crtc.o drm_fourcc.o drm_modes.o drm_edid.o \ drm_info.o drm_encoder_slave.o \ - drm_trace_points.o drm_global.o drm_prime.o \ + drm_trace_points.o drm_prime.o \ drm_rect.o drm_vma_manager.o drm_flip_work.o \ drm_modeset_lock.o drm_atomic.o drm_bridge.o \ drm_framebuffer.o drm_connector.o drm_blend.o \ diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 138cb787d27e..f76bcb9c45e4 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -53,7 +53,7 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \ amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \ amdgpu_gtt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o amdgpu_atomfirmware.o \ amdgpu_vf_error.o amdgpu_sched.o amdgpu_debugfs.o amdgpu_ids.o \ - amdgpu_gmc.o amdgpu_xgmi.o + amdgpu_gmc.o amdgpu_xgmi.o amdgpu_csa.o # add asic specific block amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \ @@ -105,6 +105,7 @@ amdgpu-y += \ # add GFX block amdgpu-y += \ amdgpu_gfx.o \ + amdgpu_rlc.o \ gfx_v8_0.o \ gfx_v9_0.o diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 104b2e0d893b..42f882c633ee 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -75,6 +75,7 @@ #include "amdgpu_sdma.h" #include "amdgpu_dm.h" #include "amdgpu_virt.h" +#include "amdgpu_csa.h" #include "amdgpu_gart.h" #include "amdgpu_debugfs.h" #include "amdgpu_job.h" @@ -433,7 +434,7 @@ typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT * default non-graphics QWORD index is 0xe0 - 0xFF inclusive */ - /* sDMA engines reserved from 0xe0 -oxef */ + /* sDMA engines reserved from 0xe0 -0xef */ AMDGPU_DOORBELL64_sDMA_ENGINE0 = 0xE0, AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0 = 0xE1, AMDGPU_DOORBELL64_sDMA_ENGINE1 = 0xE8, @@ -831,7 +832,6 @@ struct amdgpu_device { bool need_dma32; bool need_swiotlb; bool accel_working; - struct work_struct reset_work; struct notifier_block acpi_nb; struct amdgpu_i2c_chan *i2c_bus[AMDGPU_MAX_I2C_BUS]; struct amdgpu_debugfs debugfs[AMDGPU_DEBUGFS_MAX_COMPONENTS]; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index c31a8849e9f8..60f9a87e9c74 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -144,7 +144,7 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev) KGD_MAX_QUEUES); /* remove the KIQ bit as well */ - if (adev->gfx.kiq.ring.ready) + if (adev->gfx.kiq.ring.sched.ready) clear_bit(amdgpu_gfx_queue_to_bit(adev, adev->gfx.kiq.ring.me - 1, adev->gfx.kiq.ring.pipe, @@ -268,9 +268,9 @@ void amdgpu_amdkfd_gpu_reset(struct kgd_dev *kgd) amdgpu_device_gpu_recover(adev, NULL); } -int alloc_gtt_mem(struct kgd_dev *kgd, size_t size, - void **mem_obj, uint64_t *gpu_addr, - void **cpu_ptr, bool mqd_gfx9) +int amdgpu_amdkfd_alloc_gtt_mem(struct kgd_dev *kgd, size_t size, + void **mem_obj, uint64_t *gpu_addr, + void **cpu_ptr, bool mqd_gfx9) { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; struct amdgpu_bo *bo = NULL; @@ -340,7 +340,7 @@ allocate_mem_reserve_bo_failed: return r; } -void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj) +void amdgpu_amdkfd_free_gtt_mem(struct kgd_dev *kgd, void *mem_obj) { struct amdgpu_bo *bo = (struct amdgpu_bo *) mem_obj; @@ -351,8 +351,8 @@ void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj) amdgpu_bo_unref(&(bo)); } -void get_local_mem_info(struct kgd_dev *kgd, - struct kfd_local_mem_info *mem_info) +void amdgpu_amdkfd_get_local_mem_info(struct kgd_dev *kgd, + struct kfd_local_mem_info *mem_info) { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; uint64_t address_mask = adev->dev->dma_mask ? ~*adev->dev->dma_mask : @@ -383,7 +383,7 @@ void get_local_mem_info(struct kgd_dev *kgd, mem_info->mem_clk_max = 100; } -uint64_t get_gpu_clock_counter(struct kgd_dev *kgd) +uint64_t amdgpu_amdkfd_get_gpu_clock_counter(struct kgd_dev *kgd) { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; @@ -392,7 +392,7 @@ uint64_t get_gpu_clock_counter(struct kgd_dev *kgd) return 0; } -uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd) +uint32_t amdgpu_amdkfd_get_max_engine_clock_in_mhz(struct kgd_dev *kgd) { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; @@ -405,7 +405,7 @@ uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd) return 100; } -void get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info) +void amdgpu_amdkfd_get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info) { struct amdgpu_device *adev = (struct amdgpu_device *)kgd; struct amdgpu_cu_info acu_info = adev->gfx.cu_info; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index 8e0d4f7196b4..bcf587b4ba98 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -134,16 +134,16 @@ int amdgpu_amdkfd_post_reset(struct amdgpu_device *adev); void amdgpu_amdkfd_gpu_reset(struct kgd_dev *kgd); /* Shared API */ -int alloc_gtt_mem(struct kgd_dev *kgd, size_t size, - void **mem_obj, uint64_t *gpu_addr, - void **cpu_ptr, bool mqd_gfx9); -void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj); -void get_local_mem_info(struct kgd_dev *kgd, - struct kfd_local_mem_info *mem_info); -uint64_t get_gpu_clock_counter(struct kgd_dev *kgd); - -uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd); -void get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info); +int amdgpu_amdkfd_alloc_gtt_mem(struct kgd_dev *kgd, size_t size, + void **mem_obj, uint64_t *gpu_addr, + void **cpu_ptr, bool mqd_gfx9); +void amdgpu_amdkfd_free_gtt_mem(struct kgd_dev *kgd, void *mem_obj); +void amdgpu_amdkfd_get_local_mem_info(struct kgd_dev *kgd, + struct kfd_local_mem_info *mem_info); +uint64_t amdgpu_amdkfd_get_gpu_clock_counter(struct kgd_dev *kgd); + +uint32_t amdgpu_amdkfd_get_max_engine_clock_in_mhz(struct kgd_dev *kgd); +void amdgpu_amdkfd_get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info); uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd); uint64_t amdgpu_amdkfd_get_hive_id(struct kgd_dev *kgd); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c index 244d9834a381..72a357dae070 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c @@ -173,13 +173,6 @@ static int get_tile_config(struct kgd_dev *kgd, } static const struct kfd2kgd_calls kfd2kgd = { - .init_gtt_mem_allocation = alloc_gtt_mem, - .free_gtt_mem = free_gtt_mem, - .get_local_mem_info = get_local_mem_info, - .get_gpu_clock_counter = get_gpu_clock_counter, - .get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz, - .alloc_pasid = amdgpu_pasid_alloc, - .free_pasid = amdgpu_pasid_free, .program_sh_mem_settings = kgd_program_sh_mem_settings, .set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping, .init_interrupts = kgd_init_interrupts, @@ -200,28 +193,10 @@ static const struct kfd2kgd_calls kfd2kgd = { .get_fw_version = get_fw_version, .set_scratch_backing_va = set_scratch_backing_va, .get_tile_config = get_tile_config, - .get_cu_info = get_cu_info, - .get_vram_usage = amdgpu_amdkfd_get_vram_usage, - .create_process_vm = amdgpu_amdkfd_gpuvm_create_process_vm, - .acquire_process_vm = amdgpu_amdkfd_gpuvm_acquire_process_vm, - .destroy_process_vm = amdgpu_amdkfd_gpuvm_destroy_process_vm, - .release_process_vm = amdgpu_amdkfd_gpuvm_release_process_vm, - .get_process_page_dir = amdgpu_amdkfd_gpuvm_get_process_page_dir, .set_vm_context_page_table_base = set_vm_context_page_table_base, - .alloc_memory_of_gpu = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu, - .free_memory_of_gpu = amdgpu_amdkfd_gpuvm_free_memory_of_gpu, - .map_memory_to_gpu = amdgpu_amdkfd_gpuvm_map_memory_to_gpu, - .unmap_memory_to_gpu = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu, - .sync_memory = amdgpu_amdkfd_gpuvm_sync_memory, - .map_gtt_bo_to_kernel = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel, - .restore_process_bos = amdgpu_amdkfd_gpuvm_restore_process_bos, .invalidate_tlbs = invalidate_tlbs, .invalidate_tlbs_vmid = invalidate_tlbs_vmid, - .submit_ib = amdgpu_amdkfd_submit_ib, - .get_vm_fault_info = amdgpu_amdkfd_gpuvm_get_vm_fault_info, .read_vmid_from_vmfault_reg = read_vmid_from_vmfault_reg, - .gpu_recover = amdgpu_amdkfd_gpu_reset, - .set_compute_idle = amdgpu_amdkfd_set_compute_idle }; struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c index 9f149914ad6c..0e2a56b6a9b6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c @@ -128,13 +128,6 @@ static int get_tile_config(struct kgd_dev *kgd, } static const struct kfd2kgd_calls kfd2kgd = { - .init_gtt_mem_allocation = alloc_gtt_mem, - .free_gtt_mem = free_gtt_mem, - .get_local_mem_info = get_local_mem_info, - .get_gpu_clock_counter = get_gpu_clock_counter, - .get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz, - .alloc_pasid = amdgpu_pasid_alloc, - .free_pasid = amdgpu_pasid_free, .program_sh_mem_settings = kgd_program_sh_mem_settings, .set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping, .init_interrupts = kgd_init_interrupts, @@ -157,27 +150,9 @@ static const struct kfd2kgd_calls kfd2kgd = { .get_fw_version = get_fw_version, .set_scratch_backing_va = set_scratch_backing_va, .get_tile_config = get_tile_config, - .get_cu_info = get_cu_info, - .get_vram_usage = amdgpu_amdkfd_get_vram_usage, - .create_process_vm = amdgpu_amdkfd_gpuvm_create_process_vm, - .acquire_process_vm = amdgpu_amdkfd_gpuvm_acquire_process_vm, - .destroy_process_vm = amdgpu_amdkfd_gpuvm_destroy_process_vm, - .release_process_vm = amdgpu_amdkfd_gpuvm_release_process_vm, - .get_process_page_dir = amdgpu_amdkfd_gpuvm_get_process_page_dir, .set_vm_context_page_table_base = set_vm_context_page_table_base, - .alloc_memory_of_gpu = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu, - .free_memory_of_gpu = amdgpu_amdkfd_gpuvm_free_memory_of_gpu, - .map_memory_to_gpu = amdgpu_amdkfd_gpuvm_map_memory_to_gpu, - .unmap_memory_to_gpu = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu, - .sync_memory = amdgpu_amdkfd_gpuvm_sync_memory, - .map_gtt_bo_to_kernel = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel, - .restore_process_bos = amdgpu_amdkfd_gpuvm_restore_process_bos, .invalidate_tlbs = invalidate_tlbs, .invalidate_tlbs_vmid = invalidate_tlbs_vmid, - .submit_ib = amdgpu_amdkfd_submit_ib, - .get_vm_fault_info = amdgpu_amdkfd_gpuvm_get_vm_fault_info, - .gpu_recover = amdgpu_amdkfd_gpu_reset, - .set_compute_idle = amdgpu_amdkfd_set_compute_idle }; struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c index 42cb4c4e0929..03b604c96d94 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c @@ -46,38 +46,9 @@ #include "v9_structs.h" #include "soc15.h" #include "soc15d.h" +#include "mmhub_v1_0.h" +#include "gfxhub_v1_0.h" -/* HACK: MMHUB and GC both have VM-related register with the same - * names but different offsets. Define the MMHUB register we need here - * with a prefix. A proper solution would be to move the functions - * programming these registers into gfx_v9_0.c and mmhub_v1_0.c - * respectively. - */ -#define mmMMHUB_VM_INVALIDATE_ENG16_REQ 0x06f3 -#define mmMMHUB_VM_INVALIDATE_ENG16_REQ_BASE_IDX 0 - -#define mmMMHUB_VM_INVALIDATE_ENG16_ACK 0x0705 -#define mmMMHUB_VM_INVALIDATE_ENG16_ACK_BASE_IDX 0 - -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32 0x072b -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32_BASE_IDX 0 -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32 0x072c -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32_BASE_IDX 0 - -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32 0x074b -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32_BASE_IDX 0 -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32 0x074c -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32_BASE_IDX 0 - -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32 0x076b -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32_BASE_IDX 0 -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32 0x076c -#define mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32_BASE_IDX 0 - -#define mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_LO32 0x0727 -#define mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_LO32_BASE_IDX 0 -#define mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_HI32 0x0728 -#define mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_HI32_BASE_IDX 0 #define V9_PIPE_PER_MEC (4) #define V9_QUEUES_PER_PIPE_MEC (8) @@ -167,13 +138,6 @@ static int amdgpu_amdkfd_get_tile_config(struct kgd_dev *kgd, } static const struct kfd2kgd_calls kfd2kgd = { - .init_gtt_mem_allocation = alloc_gtt_mem, - .free_gtt_mem = free_gtt_mem, - .get_local_mem_info = get_local_mem_info, - .get_gpu_clock_counter = get_gpu_clock_counter, - .get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz, - .alloc_pasid = amdgpu_pasid_alloc, - .free_pasid = amdgpu_pasid_free, .program_sh_mem_settings = kgd_program_sh_mem_settings, .set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping, .init_interrupts = kgd_init_interrupts, @@ -196,26 +160,9 @@ static const struct kfd2kgd_calls kfd2kgd = { .get_fw_version = get_fw_version, .set_scratch_backing_va = set_scratch_backing_va, .get_tile_config = amdgpu_amdkfd_get_tile_config, - .get_cu_info = get_cu_info, - .get_vram_usage = amdgpu_amdkfd_get_vram_usage, - .create_process_vm = amdgpu_amdkfd_gpuvm_create_process_vm, - .acquire_process_vm = amdgpu_amdkfd_gpuvm_acquire_process_vm, - .destroy_process_vm = amdgpu_amdkfd_gpuvm_destroy_process_vm, - .release_process_vm = amdgpu_amdkfd_gpuvm_release_process_vm, - .get_process_page_dir = amdgpu_amdkfd_gpuvm_get_process_page_dir, .set_vm_context_page_table_base = set_vm_context_page_table_base, - .alloc_memory_of_gpu = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu, - .free_memory_of_gpu = amdgpu_amdkfd_gpuvm_free_memory_of_gpu, - .map_memory_to_gpu = amdgpu_amdkfd_gpuvm_map_memory_to_gpu, - .unmap_memory_to_gpu = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu, - .sync_memory = amdgpu_amdkfd_gpuvm_sync_memory, - .map_gtt_bo_to_kernel = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel, - .restore_process_bos = amdgpu_amdkfd_gpuvm_restore_process_bos, .invalidate_tlbs = invalidate_tlbs, .invalidate_tlbs_vmid = invalidate_tlbs_vmid, - .submit_ib = amdgpu_amdkfd_submit_ib, - .gpu_recover = amdgpu_amdkfd_gpu_reset, - .set_compute_idle = amdgpu_amdkfd_set_compute_idle, .get_hive_id = amdgpu_amdkfd_get_hive_id, }; @@ -785,15 +732,6 @@ static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd, static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid) { struct amdgpu_device *adev = (struct amdgpu_device *) kgd; - uint32_t req = (1 << vmid) | - (0 << VM_INVALIDATE_ENG16_REQ__FLUSH_TYPE__SHIFT) | /* legacy */ - VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PTES_MASK | - VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE0_MASK | - VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE1_MASK | - VM_INVALIDATE_ENG16_REQ__INVALIDATE_L2_PDE2_MASK | - VM_INVALIDATE_ENG16_REQ__INVALIDATE_L1_PTES_MASK; - - mutex_lock(&adev->srbm_mutex); /* Use legacy mode tlb invalidation. * @@ -810,34 +748,7 @@ static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid) * TODO 2: support range-based invalidation, requires kfg2kgd * interface change */ - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_ADDR_RANGE_LO32), - 0xffffffff); - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_ADDR_RANGE_HI32), - 0x0000001f); - - WREG32(SOC15_REG_OFFSET(MMHUB, 0, - mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_LO32), - 0xffffffff); - WREG32(SOC15_REG_OFFSET(MMHUB, 0, - mmMMHUB_VM_INVALIDATE_ENG16_ADDR_RANGE_HI32), - 0x0000001f); - - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_REQ), req); - - WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_INVALIDATE_ENG16_REQ), - req); - - while (!(RREG32(SOC15_REG_OFFSET(GC, 0, mmVM_INVALIDATE_ENG16_ACK)) & - (1 << vmid))) - cpu_relax(); - - while (!(RREG32(SOC15_REG_OFFSET(MMHUB, 0, - mmMMHUB_VM_INVALIDATE_ENG16_ACK)) & - (1 << vmid))) - cpu_relax(); - - mutex_unlock(&adev->srbm_mutex); - + amdgpu_gmc_flush_gpu_tlb(adev, vmid, 0); } static int invalidate_tlbs_with_kiq(struct amdgpu_device *adev, uint16_t pasid) @@ -876,7 +787,7 @@ static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid) if (adev->in_gpu_reset) return -EIO; - if (ring->ready) + if (ring->sched.ready) return invalidate_tlbs_with_kiq(adev, pasid); for (vmid = 0; vmid < 16; vmid++) { @@ -1016,7 +927,6 @@ static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid, uint64_t page_table_base) { struct amdgpu_device *adev = get_amdgpu_device(kgd); - uint64_t base = page_table_base | AMDGPU_PTE_VALID; if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) { pr_err("trying to set page table base for wrong VMID %u\n", @@ -1028,25 +938,7 @@ static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid, * now, all processes share the same address space size, like * on GFX8 and older. */ - WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32) + (vmid*2), 0); - WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32) + (vmid*2), 0); - - WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32) + (vmid*2), - lower_32_bits(adev->vm_manager.max_pfn - 1)); - WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32) + (vmid*2), - upper_32_bits(adev->vm_manager.max_pfn - 1)); - - WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32) + (vmid*2), lower_32_bits(base)); - WREG32(SOC15_REG_OFFSET(MMHUB, 0, mmMMHUB_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32) + (vmid*2), upper_32_bits(base)); - - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32) + (vmid*2), 0); - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32) + (vmid*2), 0); - - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32) + (vmid*2), - lower_32_bits(adev->vm_manager.max_pfn - 1)); - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32) + (vmid*2), - upper_32_bits(adev->vm_manager.max_pfn - 1)); + mmhub_v1_0_setup_vm_pt_regs(adev, vmid, page_table_base); - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32) + (vmid*2), lower_32_bits(base)); - WREG32(SOC15_REG_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32) + (vmid*2), upper_32_bits(base)); + gfxhub_v1_0_setup_vm_pt_regs(adev, vmid, page_table_base); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index 8816c697b205..ceadeeadfa56 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -330,7 +330,9 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device, case CHIP_TOPAZ: if (((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0x81)) || ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0x83)) || - ((adev->pdev->device == 0x6907) && (adev->pdev->revision == 0x87))) { + ((adev->pdev->device == 0x6907) && (adev->pdev->revision == 0x87)) || + ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0xD1)) || + ((adev->pdev->device == 0x6900) && (adev->pdev->revision == 0xD3))) { info->is_kicker = true; strcpy(fw_name, "amdgpu/topaz_k_smc.bin"); } else @@ -351,7 +353,6 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device, if (type == CGS_UCODE_ID_SMU) { if (((adev->pdev->device == 0x67ef) && ((adev->pdev->revision == 0xe0) || - (adev->pdev->revision == 0xe2) || (adev->pdev->revision == 0xe5))) || ((adev->pdev->device == 0x67ff) && ((adev->pdev->revision == 0xcf) || @@ -359,8 +360,13 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device, (adev->pdev->revision == 0xff)))) { info->is_kicker = true; strcpy(fw_name, "amdgpu/polaris11_k_smc.bin"); - } else + } else if ((adev->pdev->device == 0x67ef) && + (adev->pdev->revision == 0xe2)) { + info->is_kicker = true; + strcpy(fw_name, "amdgpu/polaris11_k2_smc.bin"); + } else { strcpy(fw_name, "amdgpu/polaris11_smc.bin"); + } } else if (type == CGS_UCODE_ID_SMU_SK) { strcpy(fw_name, "amdgpu/polaris11_smc_sk.bin"); } @@ -378,14 +384,31 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device, (adev->pdev->revision == 0xef))) { info->is_kicker = true; strcpy(fw_name, "amdgpu/polaris10_k_smc.bin"); - } else + } else if ((adev->pdev->device == 0x67df) && + ((adev->pdev->revision == 0xe1) || + (adev->pdev->revision == 0xf7))) { + info->is_kicker = true; + strcpy(fw_name, "amdgpu/polaris10_k2_smc.bin"); + } else { strcpy(fw_name, "amdgpu/polaris10_smc.bin"); + } } else if (type == CGS_UCODE_ID_SMU_SK) { strcpy(fw_name, "amdgpu/polaris10_smc_sk.bin"); } break; case CHIP_POLARIS12: - strcpy(fw_name, "amdgpu/polaris12_smc.bin"); + if (((adev->pdev->device == 0x6987) && + ((adev->pdev->revision == 0xc0) || + (adev->pdev->revision == 0xc3))) || + ((adev->pdev->device == 0x6981) && + ((adev->pdev->revision == 0x00) || + (adev->pdev->revision == 0x01) || + (adev->pdev->revision == 0x10)))) { + info->is_kicker = true; + strcpy(fw_name, "amdgpu/polaris12_k_smc.bin"); + } else { + strcpy(fw_name, "amdgpu/polaris12_smc.bin"); + } break; case CHIP_VEGAM: strcpy(fw_name, "amdgpu/vegam_smc.bin"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 35bc8fc3bc70..024dfbd87f11 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -1260,8 +1260,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p, return 0; error_abort: - dma_fence_put(&job->base.s_fence->finished); - job->base.s_fence = NULL; + drm_sched_job_cleanup(&job->base); amdgpu_mn_unlock(p->mn); error_unlock: @@ -1285,7 +1284,7 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) r = amdgpu_cs_parser_init(&parser, data); if (r) { - DRM_ERROR("Failed to initialize parser !\n"); + DRM_ERROR("Failed to initialize parser %d!\n", r); goto out; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c new file mode 100644 index 000000000000..0c590ddf250a --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c @@ -0,0 +1,117 @@ +/* + * 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. + + * * Author: Monk.liu@amd.com + */ + +#include "amdgpu.h" + +uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev) +{ + uint64_t addr = adev->vm_manager.max_pfn << AMDGPU_GPU_PAGE_SHIFT; + + addr -= AMDGPU_VA_RESERVED_SIZE; + addr = amdgpu_gmc_sign_extend(addr); + + return addr; +} + +int amdgpu_allocate_static_csa(struct amdgpu_device *adev, struct amdgpu_bo **bo, + u32 domain, uint32_t size) +{ + int r; + void *ptr; + + r = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE, + domain, bo, + NULL, &ptr); + if (!bo) + return -ENOMEM; + + memset(ptr, 0, size); + return 0; +} + +void amdgpu_free_static_csa(struct amdgpu_bo **bo) +{ + amdgpu_bo_free_kernel(bo, NULL, NULL); +} + +/* + * amdgpu_map_static_csa should be called during amdgpu_vm_init + * it maps virtual address amdgpu_csa_vaddr() to this VM, and each command + * submission of GFX should use this virtual address within META_DATA init + * package to support SRIOV gfx preemption. + */ +int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, + struct amdgpu_bo *bo, struct amdgpu_bo_va **bo_va, + uint64_t csa_addr, uint32_t size) +{ + struct ww_acquire_ctx ticket; + struct list_head list; + struct amdgpu_bo_list_entry pd; + struct ttm_validate_buffer csa_tv; + int r; + + INIT_LIST_HEAD(&list); + INIT_LIST_HEAD(&csa_tv.head); + csa_tv.bo = &bo->tbo; + csa_tv.shared = true; + + list_add(&csa_tv.head, &list); + amdgpu_vm_get_pd_bo(vm, &list, &pd); + + r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL); + if (r) { + DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r); + return r; + } + + *bo_va = amdgpu_vm_bo_add(adev, vm, bo); + if (!*bo_va) { + ttm_eu_backoff_reservation(&ticket, &list); + DRM_ERROR("failed to create bo_va for static CSA\n"); + return -ENOMEM; + } + + r = amdgpu_vm_alloc_pts(adev, (*bo_va)->base.vm, csa_addr, + size); + if (r) { + DRM_ERROR("failed to allocate pts for static CSA, err=%d\n", r); + amdgpu_vm_bo_rmv(adev, *bo_va); + ttm_eu_backoff_reservation(&ticket, &list); + return r; + } + + r = amdgpu_vm_bo_map(adev, *bo_va, csa_addr, 0, size, + AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE | + AMDGPU_PTE_EXECUTABLE); + + if (r) { + DRM_ERROR("failed to do bo_map on static CSA, err=%d\n", r); + amdgpu_vm_bo_rmv(adev, *bo_va); + ttm_eu_backoff_reservation(&ticket, &list); + return r; + } + + ttm_eu_backoff_reservation(&ticket, &list); + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.h new file mode 100644 index 000000000000..524b4437a021 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.h @@ -0,0 +1,39 @@ +/* + * 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. + * + * Author: Monk.liu@amd.com + */ + +#ifndef AMDGPU_CSA_MANAGER_H +#define AMDGPU_CSA_MANAGER_H + +#define AMDGPU_CSA_SIZE (128 * 1024) + +uint32_t amdgpu_get_total_csa_size(struct amdgpu_device *adev); +uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev); +int amdgpu_allocate_static_csa(struct amdgpu_device *adev, struct amdgpu_bo **bo, + u32 domain, uint32_t size); +int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, + struct amdgpu_bo *bo, struct amdgpu_bo_va **bo_va, + uint64_t csa_addr, uint32_t size); +void amdgpu_free_static_csa(struct amdgpu_bo **bo); + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 30bc345d6fdf..590588a82471 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -1656,7 +1656,9 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) /* right after GMC hw init, we create CSA */ if (amdgpu_sriov_vf(adev)) { - r = amdgpu_allocate_static_csa(adev); + r = amdgpu_allocate_static_csa(adev, &adev->virt.csa_obj, + AMDGPU_GEM_DOMAIN_VRAM, + AMDGPU_CSA_SIZE); if (r) { DRM_ERROR("allocate CSA failed %d\n", r); return r; @@ -1681,7 +1683,8 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) if (r) return r; - amdgpu_xgmi_add_device(adev); + if (adev->gmc.xgmi.num_physical_nodes > 1) + amdgpu_xgmi_add_device(adev); amdgpu_amdkfd_device_init(adev); if (amdgpu_sriov_vf(adev)) @@ -1890,7 +1893,7 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev) if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC) { amdgpu_ucode_free_bo(adev); - amdgpu_free_static_csa(adev); + amdgpu_free_static_csa(&adev->virt.csa_obj); amdgpu_device_wb_fini(adev); amdgpu_device_vram_scratch_fini(adev); } @@ -3295,13 +3298,35 @@ bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev) return false; } - if (amdgpu_gpu_recovery == 0 || (amdgpu_gpu_recovery == -1 && - !amdgpu_sriov_vf(adev))) { - DRM_INFO("GPU recovery disabled.\n"); - return false; + if (amdgpu_gpu_recovery == 0) + goto disabled; + + if (amdgpu_sriov_vf(adev)) + return true; + + if (amdgpu_gpu_recovery == -1) { + switch (adev->asic_type) { + case CHIP_TOPAZ: + case CHIP_TONGA: + case CHIP_FIJI: + case CHIP_POLARIS10: + case CHIP_POLARIS11: + case CHIP_POLARIS12: + case CHIP_VEGAM: + case CHIP_VEGA20: + case CHIP_VEGA10: + case CHIP_VEGA12: + break; + default: + goto disabled; + } } return true; + +disabled: + DRM_INFO("GPU recovery disabled.\n"); + return false; } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index 5448cf27654e..ee47c11e92ce 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -398,9 +398,9 @@ int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring, ring->fence_drv.irq_type = irq_type; ring->fence_drv.initialized = true; - dev_dbg(adev->dev, "fence driver on ring %d use gpu addr 0x%016llx, " - "cpu addr 0x%p\n", ring->idx, - ring->fence_drv.gpu_addr, ring->fence_drv.cpu_addr); + DRM_DEV_DEBUG(adev->dev, "fence driver on ring %s use gpu addr " + "0x%016llx, cpu addr 0x%p\n", ring->name, + ring->fence_drv.gpu_addr, ring->fence_drv.cpu_addr); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c index 11fea28f8ad3..6d11e1721147 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c @@ -248,7 +248,7 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, } mb(); amdgpu_asic_flush_hdp(adev, NULL); - amdgpu_gmc_flush_gpu_tlb(adev, 0); + amdgpu_gmc_flush_gpu_tlb(adev, 0, 0); return 0; } @@ -259,6 +259,8 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, * @offset: offset into the GPU's gart aperture * @pages: number of pages to bind * @dma_addr: DMA addresses of pages + * @flags: page table entry flags + * @dst: CPU address of the gart table * * Map the dma_addresses into GART entries (all asics). * Returns 0 for success, -EINVAL for failure. @@ -331,7 +333,7 @@ int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset, mb(); amdgpu_asic_flush_hdp(adev, NULL); - amdgpu_gmc_flush_gpu_tlb(adev, 0); + amdgpu_gmc_flush_gpu_tlb(adev, 0, 0); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h index 9ff62887e4e3..afa2e2877d87 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h @@ -41,6 +41,7 @@ struct amdgpu_bo; struct amdgpu_gart { struct amdgpu_bo *bo; + /* CPU kmapped address of gart table */ void *ptr; unsigned num_gpu_pages; unsigned num_cpu_pages; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 1a656b8657f7..6a70c0b7105f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -25,6 +25,7 @@ #include <drm/drmP.h> #include "amdgpu.h" #include "amdgpu_gfx.h" +#include "amdgpu_rlc.h" /* delay 0.1 second to enable gfx off feature */ #define GFX_OFF_DELAY_ENABLE msecs_to_jiffies(100) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index b61b5c11aead..f790e15bcd08 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -29,6 +29,7 @@ */ #include "clearstate_defs.h" #include "amdgpu_ring.h" +#include "amdgpu_rlc.h" /* GFX current status */ #define AMDGPU_GFX_NORMAL_MODE 0x00000000L @@ -37,59 +38,6 @@ #define AMDGPU_GFX_CG_DISABLED_MODE 0x00000004L #define AMDGPU_GFX_LBPW_DISABLED_MODE 0x00000008L - -struct amdgpu_rlc_funcs { - void (*enter_safe_mode)(struct amdgpu_device *adev); - void (*exit_safe_mode)(struct amdgpu_device *adev); -}; - -struct amdgpu_rlc { - /* for power gating */ - struct amdgpu_bo *save_restore_obj; - uint64_t save_restore_gpu_addr; - volatile uint32_t *sr_ptr; - const u32 *reg_list; - u32 reg_list_size; - /* for clear state */ - struct amdgpu_bo *clear_state_obj; - uint64_t clear_state_gpu_addr; - volatile uint32_t *cs_ptr; - const struct cs_section_def *cs_data; - u32 clear_state_size; - /* for cp tables */ - struct amdgpu_bo *cp_table_obj; - uint64_t cp_table_gpu_addr; - volatile uint32_t *cp_table_ptr; - u32 cp_table_size; - - /* safe mode for updating CG/PG state */ - bool in_safe_mode; - const struct amdgpu_rlc_funcs *funcs; - - /* for firmware data */ - u32 save_and_restore_offset; - u32 clear_state_descriptor_offset; - u32 avail_scratch_ram_locations; - u32 reg_restore_list_size; - u32 reg_list_format_start; - u32 reg_list_format_separate_start; - u32 starting_offsets_start; - u32 reg_list_format_size_bytes; - u32 reg_list_size_bytes; - u32 reg_list_format_direct_reg_list_length; - u32 save_restore_list_cntl_size_bytes; - u32 save_restore_list_gpm_size_bytes; - u32 save_restore_list_srm_size_bytes; - - u32 *register_list_format; - u32 *register_restore; - u8 *save_restore_list_cntl; - u8 *save_restore_list_gpm; - u8 *save_restore_list_srm; - - bool is_rlc_v2_1; -}; - #define AMDGPU_MAX_COMPUTE_QUEUES KGD_MAX_QUEUES struct amdgpu_mec { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h index 6fa7ef446e46..8c57924c075f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h @@ -64,7 +64,7 @@ struct amdgpu_vmhub { struct amdgpu_gmc_funcs { /* flush the vm tlb via mmio */ void (*flush_gpu_tlb)(struct amdgpu_device *adev, - uint32_t vmid); + uint32_t vmid, uint32_t flush_type); /* flush the vm tlb via ring */ uint64_t (*emit_flush_gpu_tlb)(struct amdgpu_ring *ring, unsigned vmid, uint64_t pd_addr); @@ -89,7 +89,7 @@ struct amdgpu_gmc_funcs { struct amdgpu_xgmi { /* from psp */ - u64 device_id; + u64 node_id; u64 hive_id; /* fixed per family */ u64 node_segment_size; @@ -151,7 +151,7 @@ struct amdgpu_gmc { struct amdgpu_xgmi xgmi; }; -#define amdgpu_gmc_flush_gpu_tlb(adev, vmid) (adev)->gmc.gmc_funcs->flush_gpu_tlb((adev), (vmid)) +#define amdgpu_gmc_flush_gpu_tlb(adev, vmid, type) (adev)->gmc.gmc_funcs->flush_gpu_tlb((adev), (vmid), (type)) #define amdgpu_gmc_emit_flush_gpu_tlb(r, vmid, addr) (r)->adev->gmc.gmc_funcs->emit_flush_gpu_tlb((r), (vmid), (addr)) #define amdgpu_gmc_emit_pasid_mapping(r, vmid, pasid) (r)->adev->gmc.gmc_funcs->emit_pasid_mapping((r), (vmid), (pasid)) #define amdgpu_gmc_set_pte_pde(adev, pt, idx, addr, flags) (adev)->gmc.gmc_funcs->set_pte_pde((adev), (pt), (idx), (addr), (flags)) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index b8963b725dfa..c48207b377bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -146,7 +146,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, fence_ctx = 0; } - if (!ring->ready) { + if (!ring->sched.ready) { dev_err(adev->dev, "couldn't schedule ib on ring <%s>\n", ring->name); return -EINVAL; } @@ -221,8 +221,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, !amdgpu_sriov_vf(adev)) /* for SRIOV preemption, Preamble CE ib must be inserted anyway */ continue; - amdgpu_ring_emit_ib(ring, ib, job ? job->vmid : 0, - need_ctx_switch); + amdgpu_ring_emit_ib(ring, job, ib, need_ctx_switch); need_ctx_switch = false; } @@ -347,19 +346,14 @@ int amdgpu_ib_ring_tests(struct amdgpu_device *adev) tmo_gfx = 8 * AMDGPU_IB_TEST_TIMEOUT; } - for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { + for (i = 0; i < adev->num_rings; ++i) { struct amdgpu_ring *ring = adev->rings[i]; long tmo; - if (!ring || !ring->ready) - continue; - - /* skip IB tests for KIQ in general for the below reasons: - * 1. We never submit IBs to the KIQ - * 2. KIQ doesn't use the EOP interrupts, - * we use some other CP interrupt. + /* KIQ rings don't have an IB test because we never submit IBs + * to them and they have no interrupt support. */ - if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ) + if (!ring->sched.ready || !ring->funcs->test_ib) continue; /* MM engine need more time */ @@ -374,20 +368,23 @@ int amdgpu_ib_ring_tests(struct amdgpu_device *adev) tmo = tmo_gfx; r = amdgpu_ring_test_ib(ring, tmo); - if (r) { - ring->ready = false; - - if (ring == &adev->gfx.gfx_ring[0]) { - /* oh, oh, that's really bad */ - DRM_ERROR("amdgpu: failed testing IB on GFX ring (%d).\n", r); - adev->accel_working = false; - return r; - - } else { - /* still not good, but we can live with it */ - DRM_ERROR("amdgpu: failed testing IB on ring %d (%d).\n", i, r); - ret = r; - } + if (!r) { + DRM_DEV_DEBUG(adev->dev, "ib test on %s succeeded\n", + ring->name); + continue; + } + + ring->sched.ready = false; + DRM_DEV_ERROR(adev->dev, "IB test failed on %s (%d).\n", + ring->name, r); + + if (ring == &adev->gfx.gfx_ring[0]) { + /* oh, oh, that's really bad */ + adev->accel_working = false; + return r; + + } else { + ret = r; } } return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 52c17f6219a7..6b6524f04ce0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -94,23 +94,6 @@ static void amdgpu_hotplug_work_func(struct work_struct *work) } /** - * amdgpu_irq_reset_work_func - execute GPU reset - * - * @work: work struct pointer - * - * Execute scheduled GPU reset (Cayman+). - * This function is called when the IRQ handler thinks we need a GPU reset. - */ -static void amdgpu_irq_reset_work_func(struct work_struct *work) -{ - struct amdgpu_device *adev = container_of(work, struct amdgpu_device, - reset_work); - - if (!amdgpu_sriov_vf(adev) && amdgpu_device_should_recover_gpu(adev)) - amdgpu_device_gpu_recover(adev, NULL); -} - -/** * amdgpu_irq_disable_all - disable *all* interrupts * * @adev: amdgpu device pointer @@ -262,15 +245,12 @@ int amdgpu_irq_init(struct amdgpu_device *adev) amdgpu_hotplug_work_func); } - INIT_WORK(&adev->reset_work, amdgpu_irq_reset_work_func); - adev->irq.installed = true; r = drm_irq_install(adev->ddev, adev->ddev->pdev->irq); if (r) { adev->irq.installed = false; if (!amdgpu_device_has_dc_support(adev)) flush_work(&adev->hotplug_work); - cancel_work_sync(&adev->reset_work); return r; } adev->ddev->max_vblank_count = 0x00ffffff; @@ -299,7 +279,6 @@ void amdgpu_irq_fini(struct amdgpu_device *adev) pci_disable_msi(adev->pdev); if (!amdgpu_device_has_dc_support(adev)) flush_work(&adev->hotplug_work); - cancel_work_sync(&adev->reset_work); } for (i = 0; i < AMDGPU_IRQ_CLIENTID_MAX; ++i) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index 755f733bf0d9..e0af44fd6a0c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -112,6 +112,8 @@ static void amdgpu_job_free_cb(struct drm_sched_job *s_job) struct amdgpu_ring *ring = to_amdgpu_ring(s_job->sched); struct amdgpu_job *job = to_amdgpu_job(s_job); + drm_sched_job_cleanup(s_job); + amdgpu_ring_priority_put(ring, s_job->s_priority); dma_fence_put(job->fence); amdgpu_sync_free(&job->sync); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h index 57cfe78a262b..e1b46a6703de 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h @@ -33,6 +33,8 @@ #define to_amdgpu_job(sched_job) \ container_of((sched_job), struct amdgpu_job, base) +#define AMDGPU_JOB_GET_VMID(job) ((job) ? (job)->vmid : 0) + struct amdgpu_fence; struct amdgpu_job { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index 81732a84c2ab..9b3164c0f861 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -336,7 +336,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, case AMDGPU_HW_IP_GFX: type = AMD_IP_BLOCK_TYPE_GFX; for (i = 0; i < adev->gfx.num_gfx_rings; i++) - if (adev->gfx.gfx_ring[i].ready) + if (adev->gfx.gfx_ring[i].sched.ready) ++num_rings; ib_start_alignment = 32; ib_size_alignment = 32; @@ -344,7 +344,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, case AMDGPU_HW_IP_COMPUTE: type = AMD_IP_BLOCK_TYPE_GFX; for (i = 0; i < adev->gfx.num_compute_rings; i++) - if (adev->gfx.compute_ring[i].ready) + if (adev->gfx.compute_ring[i].sched.ready) ++num_rings; ib_start_alignment = 32; ib_size_alignment = 32; @@ -352,7 +352,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, case AMDGPU_HW_IP_DMA: type = AMD_IP_BLOCK_TYPE_SDMA; for (i = 0; i < adev->sdma.num_instances; i++) - if (adev->sdma.instance[i].ring.ready) + if (adev->sdma.instance[i].ring.sched.ready) ++num_rings; ib_start_alignment = 256; ib_size_alignment = 4; @@ -363,7 +363,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, if (adev->uvd.harvest_config & (1 << i)) continue; - if (adev->uvd.inst[i].ring.ready) + if (adev->uvd.inst[i].ring.sched.ready) ++num_rings; } ib_start_alignment = 64; @@ -372,7 +372,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, case AMDGPU_HW_IP_VCE: type = AMD_IP_BLOCK_TYPE_VCE; for (i = 0; i < adev->vce.num_rings; i++) - if (adev->vce.ring[i].ready) + if (adev->vce.ring[i].sched.ready) ++num_rings; ib_start_alignment = 4; ib_size_alignment = 1; @@ -384,7 +384,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, continue; for (j = 0; j < adev->uvd.num_enc_rings; j++) - if (adev->uvd.inst[i].ring_enc[j].ready) + if (adev->uvd.inst[i].ring_enc[j].sched.ready) ++num_rings; } ib_start_alignment = 64; @@ -392,7 +392,7 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, break; case AMDGPU_HW_IP_VCN_DEC: type = AMD_IP_BLOCK_TYPE_VCN; - if (adev->vcn.ring_dec.ready) + if (adev->vcn.ring_dec.sched.ready) ++num_rings; ib_start_alignment = 16; ib_size_alignment = 16; @@ -400,14 +400,14 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, case AMDGPU_HW_IP_VCN_ENC: type = AMD_IP_BLOCK_TYPE_VCN; for (i = 0; i < adev->vcn.num_enc_rings; i++) - if (adev->vcn.ring_enc[i].ready) + if (adev->vcn.ring_enc[i].sched.ready) ++num_rings; ib_start_alignment = 64; ib_size_alignment = 1; break; case AMDGPU_HW_IP_VCN_JPEG: type = AMD_IP_BLOCK_TYPE_VCN; - if (adev->vcn.ring_jpeg.ready) + if (adev->vcn.ring_jpeg.sched.ready) ++num_rings; ib_start_alignment = 16; ib_size_alignment = 16; @@ -978,7 +978,10 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv) } if (amdgpu_sriov_vf(adev)) { - r = amdgpu_map_static_csa(adev, &fpriv->vm, &fpriv->csa_va); + uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_GMC_HOLE_MASK; + + r = amdgpu_map_static_csa(adev, &fpriv->vm, adev->virt.csa_obj, + &fpriv->csa_va, csa_addr, AMDGPU_CSA_SIZE); if (r) goto error_vm; } @@ -1048,8 +1051,8 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, pasid = fpriv->vm.pasid; pd = amdgpu_bo_ref(fpriv->vm.root.base.bo); - amdgpu_vm_fini(adev, &fpriv->vm); amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr); + amdgpu_vm_fini(adev, &fpriv->vm); if (pasid) amdgpu_pasid_free_delayed(pd->tbo.resv, pasid); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index b9e9e8b02fb7..11723d8fffbd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -57,7 +57,6 @@ struct amdgpu_hpd; #define to_amdgpu_connector(x) container_of(x, struct amdgpu_connector, base) #define to_amdgpu_encoder(x) container_of(x, struct amdgpu_encoder, base) #define to_amdgpu_framebuffer(x) container_of(x, struct amdgpu_framebuffer, base) -#define to_amdgpu_plane(x) container_of(x, struct amdgpu_plane, base) #define to_dm_plane_state(x) container_of(x, struct dm_plane_state, base); @@ -325,7 +324,7 @@ struct amdgpu_mode_info { struct card_info *atom_card_info; bool mode_config_initialized; struct amdgpu_crtc *crtcs[AMDGPU_MAX_CRTCS]; - struct amdgpu_plane *planes[AMDGPU_MAX_PLANES]; + struct drm_plane *planes[AMDGPU_MAX_PLANES]; struct amdgpu_afmt *afmt[AMDGPU_MAX_AFMT_BLOCKS]; /* DVI-I properties */ struct drm_property *coherent_mode_property; @@ -434,11 +433,6 @@ struct amdgpu_crtc { struct drm_pending_vblank_event *event; }; -struct amdgpu_plane { - struct drm_plane base; - enum drm_plane_type plane_type; -}; - struct amdgpu_encoder_atom_dig { bool linkb; /* atom dig */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index 59cc678de8c1..7235cd0b0fa9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -2129,7 +2129,7 @@ void amdgpu_pm_compute_clocks(struct amdgpu_device *adev) for (i = 0; i < AMDGPU_MAX_RINGS; i++) { struct amdgpu_ring *ring = adev->rings[i]; - if (ring && ring->ready) + if (ring && ring->sched.ready) amdgpu_fence_wait_empty(ring); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 25d2f3e757f1..e05dc66b1090 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -90,6 +90,8 @@ static int psp_sw_fini(void *handle) adev->psp.sos_fw = NULL; release_firmware(adev->psp.asd_fw); adev->psp.asd_fw = NULL; + release_firmware(adev->psp.ta_fw); + adev->psp.ta_fw = NULL; return 0; } @@ -118,21 +120,25 @@ int psp_wait_for(struct psp_context *psp, uint32_t reg_index, static int psp_cmd_submit_buf(struct psp_context *psp, struct amdgpu_firmware_info *ucode, - struct psp_gfx_cmd_resp *cmd, uint64_t fence_mc_addr, - int index) + struct psp_gfx_cmd_resp *cmd, uint64_t fence_mc_addr) { int ret; + int index; memset(psp->cmd_buf_mem, 0, PSP_CMD_BUFFER_SIZE); memcpy(psp->cmd_buf_mem, cmd, sizeof(struct psp_gfx_cmd_resp)); + index = atomic_inc_return(&psp->fence_value); ret = psp_cmd_submit(psp, ucode, psp->cmd_buf_mc_addr, fence_mc_addr, index); + if (ret) { + atomic_dec(&psp->fence_value); + return ret; + } - while (*((unsigned int *)psp->fence_buf) != index) { + while (*((unsigned int *)psp->fence_buf) != index) msleep(1); - } /* the status field must be 0 after FW is loaded */ if (ucode && psp->cmd_buf_mem->resp.status) { @@ -191,7 +197,7 @@ static int psp_tmr_load(struct psp_context *psp) PSP_TMR_SIZE, psp->tmr_mc_addr); ret = psp_cmd_submit_buf(psp, NULL, cmd, - psp->fence_buf_mc_addr, 1); + psp->fence_buf_mc_addr); if (ret) goto failed; @@ -258,13 +264,194 @@ static int psp_asd_load(struct psp_context *psp) psp->asd_ucode_size, PSP_ASD_SHARED_MEM_SIZE); ret = psp_cmd_submit_buf(psp, NULL, cmd, - psp->fence_buf_mc_addr, 2); + psp->fence_buf_mc_addr); + + kfree(cmd); + + return ret; +} + +static void psp_prep_xgmi_ta_load_cmd_buf(struct psp_gfx_cmd_resp *cmd, + uint64_t xgmi_ta_mc, uint64_t xgmi_mc_shared, + uint32_t xgmi_ta_size, uint32_t shared_size) +{ + cmd->cmd_id = GFX_CMD_ID_LOAD_TA; + cmd->cmd.cmd_load_ta.app_phy_addr_lo = lower_32_bits(xgmi_ta_mc); + cmd->cmd.cmd_load_ta.app_phy_addr_hi = upper_32_bits(xgmi_ta_mc); + cmd->cmd.cmd_load_ta.app_len = xgmi_ta_size; + + cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_lo = lower_32_bits(xgmi_mc_shared); + cmd->cmd.cmd_load_ta.cmd_buf_phy_addr_hi = upper_32_bits(xgmi_mc_shared); + cmd->cmd.cmd_load_ta.cmd_buf_len = shared_size; +} + +static int psp_xgmi_init_shared_buf(struct psp_context *psp) +{ + int ret; + + /* + * Allocate 16k memory aligned to 4k from Frame Buffer (local + * physical) for xgmi ta <-> Driver + */ + ret = amdgpu_bo_create_kernel(psp->adev, PSP_XGMI_SHARED_MEM_SIZE, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM, + &psp->xgmi_context.xgmi_shared_bo, + &psp->xgmi_context.xgmi_shared_mc_addr, + &psp->xgmi_context.xgmi_shared_buf); + + return ret; +} + +static int psp_xgmi_load(struct psp_context *psp) +{ + int ret; + struct psp_gfx_cmd_resp *cmd; + + /* + * TODO: bypass the loading in sriov for now + */ + if (amdgpu_sriov_vf(psp->adev)) + return 0; + + cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL); + if (!cmd) + return -ENOMEM; + + memset(psp->fw_pri_buf, 0, PSP_1_MEG); + memcpy(psp->fw_pri_buf, psp->ta_xgmi_start_addr, psp->ta_xgmi_ucode_size); + + psp_prep_xgmi_ta_load_cmd_buf(cmd, psp->fw_pri_mc_addr, + psp->xgmi_context.xgmi_shared_mc_addr, + psp->ta_xgmi_ucode_size, PSP_XGMI_SHARED_MEM_SIZE); + + ret = psp_cmd_submit_buf(psp, NULL, cmd, + psp->fence_buf_mc_addr); + + if (!ret) { + psp->xgmi_context.initialized = 1; + psp->xgmi_context.session_id = cmd->resp.session_id; + } + + kfree(cmd); + + return ret; +} + +static void psp_prep_xgmi_ta_unload_cmd_buf(struct psp_gfx_cmd_resp *cmd, + uint32_t xgmi_session_id) +{ + cmd->cmd_id = GFX_CMD_ID_UNLOAD_TA; + cmd->cmd.cmd_unload_ta.session_id = xgmi_session_id; +} + +static int psp_xgmi_unload(struct psp_context *psp) +{ + int ret; + struct psp_gfx_cmd_resp *cmd; + + /* + * TODO: bypass the unloading in sriov for now + */ + if (amdgpu_sriov_vf(psp->adev)) + return 0; + + cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL); + if (!cmd) + return -ENOMEM; + + psp_prep_xgmi_ta_unload_cmd_buf(cmd, psp->xgmi_context.session_id); + + ret = psp_cmd_submit_buf(psp, NULL, cmd, + psp->fence_buf_mc_addr); kfree(cmd); return ret; } +static void psp_prep_xgmi_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd, + uint32_t ta_cmd_id, + uint32_t xgmi_session_id) +{ + cmd->cmd_id = GFX_CMD_ID_INVOKE_CMD; + cmd->cmd.cmd_invoke_cmd.session_id = xgmi_session_id; + cmd->cmd.cmd_invoke_cmd.ta_cmd_id = ta_cmd_id; + /* Note: cmd_invoke_cmd.buf is not used for now */ +} + +int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id) +{ + int ret; + struct psp_gfx_cmd_resp *cmd; + + /* + * TODO: bypass the loading in sriov for now + */ + if (amdgpu_sriov_vf(psp->adev)) + return 0; + + cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL); + if (!cmd) + return -ENOMEM; + + psp_prep_xgmi_ta_invoke_cmd_buf(cmd, ta_cmd_id, + psp->xgmi_context.session_id); + + ret = psp_cmd_submit_buf(psp, NULL, cmd, + psp->fence_buf_mc_addr); + + kfree(cmd); + + return ret; +} + +static int psp_xgmi_terminate(struct psp_context *psp) +{ + int ret; + + if (!psp->xgmi_context.initialized) + return 0; + + ret = psp_xgmi_unload(psp); + if (ret) + return ret; + + psp->xgmi_context.initialized = 0; + + /* free xgmi shared memory */ + amdgpu_bo_free_kernel(&psp->xgmi_context.xgmi_shared_bo, + &psp->xgmi_context.xgmi_shared_mc_addr, + &psp->xgmi_context.xgmi_shared_buf); + + return 0; +} + +static int psp_xgmi_initialize(struct psp_context *psp) +{ + struct ta_xgmi_shared_memory *xgmi_cmd; + int ret; + + if (!psp->xgmi_context.initialized) { + ret = psp_xgmi_init_shared_buf(psp); + if (ret) + return ret; + } + + /* Load XGMI TA */ + ret = psp_xgmi_load(psp); + if (ret) + return ret; + + /* Initialize XGMI session */ + xgmi_cmd = (struct ta_xgmi_shared_memory *)(psp->xgmi_context.xgmi_shared_buf); + memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); + xgmi_cmd->cmd_id = TA_COMMAND_XGMI__INITIALIZE; + + ret = psp_xgmi_invoke(psp, xgmi_cmd->cmd_id); + + return ret; +} + static int psp_hw_start(struct psp_context *psp) { struct amdgpu_device *adev = psp->adev; @@ -292,6 +479,15 @@ static int psp_hw_start(struct psp_context *psp) if (ret) return ret; + if (adev->gmc.xgmi.num_physical_nodes > 1) { + ret = psp_xgmi_initialize(psp); + /* Warning the XGMI seesion initialize failure + * Instead of stop driver initialization + */ + if (ret) + dev_err(psp->adev->dev, + "XGMI: Failed to initialize XGMI session\n"); + } return 0; } @@ -321,7 +517,7 @@ static int psp_np_fw_load(struct psp_context *psp) return ret; ret = psp_cmd_submit_buf(psp, ucode, psp->cmd, - psp->fence_buf_mc_addr, i + 3); + psp->fence_buf_mc_addr); if (ret) return ret; @@ -452,6 +648,10 @@ static int psp_hw_fini(void *handle) if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) return 0; + if (adev->gmc.xgmi.num_physical_nodes > 1 && + psp->xgmi_context.initialized == 1) + psp_xgmi_terminate(psp); + psp_ring_destroy(psp, PSP_RING_TYPE__KM); amdgpu_bo_free_kernel(&psp->tmr_bo, &psp->tmr_mc_addr, &psp->tmr_buf); @@ -479,6 +679,15 @@ static int psp_suspend(void *handle) if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) return 0; + if (adev->gmc.xgmi.num_physical_nodes > 1 && + psp->xgmi_context.initialized == 1) { + ret = psp_xgmi_terminate(psp); + if (ret) { + DRM_ERROR("Failed to terminate xgmi ta\n"); + return ret; + } + } + ret = psp_ring_stop(psp, PSP_RING_TYPE__KM); if (ret) { DRM_ERROR("PSP ring stop failed\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index 8b8720e9c3f0..9ec5d1a666a6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -27,14 +27,17 @@ #include "amdgpu.h" #include "psp_gfx_if.h" +#include "ta_xgmi_if.h" #define PSP_FENCE_BUFFER_SIZE 0x1000 #define PSP_CMD_BUFFER_SIZE 0x1000 -#define PSP_ASD_SHARED_MEM_SIZE 0x4000 +#define PSP_ASD_SHARED_MEM_SIZE 0x4000 +#define PSP_XGMI_SHARED_MEM_SIZE 0x4000 #define PSP_1_MEG 0x100000 #define PSP_TMR_SIZE 0x400000 struct psp_context; +struct psp_xgmi_node_info; struct psp_xgmi_topology_info; enum psp_ring_type @@ -80,12 +83,20 @@ struct psp_funcs enum AMDGPU_UCODE_ID ucode_type); bool (*smu_reload_quirk)(struct psp_context *psp); int (*mode1_reset)(struct psp_context *psp); - uint64_t (*xgmi_get_device_id)(struct psp_context *psp); + uint64_t (*xgmi_get_node_id)(struct psp_context *psp); uint64_t (*xgmi_get_hive_id)(struct psp_context *psp); int (*xgmi_get_topology_info)(struct psp_context *psp, int number_devices, - struct psp_xgmi_topology_info *topology); + struct psp_xgmi_topology_info *topology); int (*xgmi_set_topology_info)(struct psp_context *psp, int number_devices, - struct psp_xgmi_topology_info *topology); + struct psp_xgmi_topology_info *topology); +}; + +struct psp_xgmi_context { + uint8_t initialized; + uint32_t session_id; + struct amdgpu_bo *xgmi_shared_bo; + uint64_t xgmi_shared_mc_addr; + void *xgmi_shared_buf; }; struct psp_context @@ -96,7 +107,7 @@ struct psp_context const struct psp_funcs *funcs; - /* fence buffer */ + /* firmware buffer */ struct amdgpu_bo *fw_pri_bo; uint64_t fw_pri_mc_addr; void *fw_pri_buf; @@ -134,6 +145,16 @@ struct psp_context struct amdgpu_bo *cmd_buf_bo; uint64_t cmd_buf_mc_addr; struct psp_gfx_cmd_resp *cmd_buf_mem; + + /* fence value associated with cmd buffer */ + atomic_t fence_value; + + /* xgmi ta firmware and buffer */ + const struct firmware *ta_fw; + uint32_t ta_xgmi_ucode_version; + uint32_t ta_xgmi_ucode_size; + uint8_t *ta_xgmi_start_addr; + struct psp_xgmi_context xgmi_context; }; struct amdgpu_psp_funcs { @@ -141,21 +162,17 @@ struct amdgpu_psp_funcs { enum AMDGPU_UCODE_ID); }; +#define AMDGPU_XGMI_MAX_CONNECTED_NODES 64 +struct psp_xgmi_node_info { + uint64_t node_id; + uint8_t num_hops; + uint8_t is_sharing_enabled; + enum ta_xgmi_assigned_sdma_engine sdma_engine; +}; + struct psp_xgmi_topology_info { - /* Generated by PSP to identify the GPU instance within xgmi connection */ - uint64_t device_id; - /* - * If all bits set to 0 , driver indicates it wants to retrieve the xgmi - * connection vector topology, but not access enable the connections - * if some or all bits are set to 1, driver indicates it want to retrieve the - * current xgmi topology and access enable the link to GPU[i] associated - * with the bit position in the vector. - * On return,: bits indicated which xgmi links are present/active depending - * on the value passed in. The relative bit offset for the relative GPU index - * within the hive is always marked active. - */ - uint32_t connection_mask; - uint32_t reserved; /* must be 0 */ + uint32_t num_nodes; + struct psp_xgmi_node_info nodes[AMDGPU_XGMI_MAX_CONNECTED_NODES]; }; #define psp_prep_cmd_buf(ucode, type) (psp)->funcs->prep_cmd_buf((ucode), (type)) @@ -177,8 +194,8 @@ struct psp_xgmi_topology_info { ((psp)->funcs->smu_reload_quirk ? (psp)->funcs->smu_reload_quirk((psp)) : false) #define psp_mode1_reset(psp) \ ((psp)->funcs->mode1_reset ? (psp)->funcs->mode1_reset((psp)) : false) -#define psp_xgmi_get_device_id(psp) \ - ((psp)->funcs->xgmi_get_device_id ? (psp)->funcs->xgmi_get_device_id((psp)) : 0) +#define psp_xgmi_get_node_id(psp) \ + ((psp)->funcs->xgmi_get_node_id ? (psp)->funcs->xgmi_get_node_id((psp)) : 0) #define psp_xgmi_get_hive_id(psp) \ ((psp)->funcs->xgmi_get_hive_id ? (psp)->funcs->xgmi_get_hive_id((psp)) : 0) #define psp_xgmi_get_topology_info(psp, num_device, topology) \ @@ -199,6 +216,8 @@ extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index, extern const struct amdgpu_ip_block_version psp_v10_0_ip_block; int psp_gpu_reset(struct amdgpu_device *adev); +int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id); + extern const struct amdgpu_ip_block_version psp_v11_0_ip_block; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index b70e85ec147d..5b75bdc8dc28 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -338,7 +338,7 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring, */ void amdgpu_ring_fini(struct amdgpu_ring *ring) { - ring->ready = false; + ring->sched.ready = false; /* Not to finish a ring which is not initialized */ if (!(ring->adev) || !(ring->adev->rings[ring->idx])) @@ -500,3 +500,29 @@ static void amdgpu_debugfs_ring_fini(struct amdgpu_ring *ring) debugfs_remove(ring->ent); #endif } + +/** + * amdgpu_ring_test_helper - tests ring and set sched readiness status + * + * @ring: ring to try the recovery on + * + * Tests ring and set sched readiness status + * + * Returns 0 on success, error on failure. + */ +int amdgpu_ring_test_helper(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + int r; + + r = amdgpu_ring_test_ring(ring); + if (r) + DRM_DEV_ERROR(adev->dev, "ring %s test failed (%d)\n", + ring->name, r); + else + DRM_DEV_DEBUG(adev->dev, "ring test on %s succeeded\n", + ring->name); + + ring->sched.ready = !r; + return r; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index 4caa301ce454..0beb01fef83f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h @@ -129,8 +129,9 @@ struct amdgpu_ring_funcs { unsigned emit_ib_size; /* command emit functions */ void (*emit_ib)(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch); + bool ctx_switch); void (*emit_fence)(struct amdgpu_ring *ring, uint64_t addr, uint64_t seq, unsigned flags); void (*emit_pipeline_sync)(struct amdgpu_ring *ring); @@ -189,7 +190,6 @@ struct amdgpu_ring { uint64_t gpu_addr; uint64_t ptr_mask; uint32_t buf_mask; - bool ready; u32 idx; u32 me; u32 pipe; @@ -229,7 +229,7 @@ struct amdgpu_ring { #define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r)) #define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r)) #define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r)) -#define amdgpu_ring_emit_ib(r, ib, vmid, c) (r)->funcs->emit_ib((r), (ib), (vmid), (c)) +#define amdgpu_ring_emit_ib(r, job, ib, c) ((r)->funcs->emit_ib((r), (job), (ib), (c))) #define amdgpu_ring_emit_pipeline_sync(r) (r)->funcs->emit_pipeline_sync((r)) #define amdgpu_ring_emit_vm_flush(r, vmid, addr) (r)->funcs->emit_vm_flush((r), (vmid), (addr)) #define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags)) @@ -313,4 +313,6 @@ static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring, ring->count_dw -= count_dw; } +int amdgpu_ring_test_helper(struct amdgpu_ring *ring); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c new file mode 100644 index 000000000000..c8793e6cc3c5 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c @@ -0,0 +1,282 @@ +/* + * Copyright 2014 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * 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 <linux/firmware.h> +#include "amdgpu.h" +#include "amdgpu_gfx.h" +#include "amdgpu_rlc.h" + +/** + * amdgpu_gfx_rlc_enter_safe_mode - Set RLC into safe mode + * + * @adev: amdgpu_device pointer + * + * Set RLC enter into safe mode if RLC is enabled and haven't in safe mode. + */ +void amdgpu_gfx_rlc_enter_safe_mode(struct amdgpu_device *adev) +{ + if (adev->gfx.rlc.in_safe_mode) + return; + + /* if RLC is not enabled, do nothing */ + if (!adev->gfx.rlc.funcs->is_rlc_enabled(adev)) + return; + + if (adev->cg_flags & + (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG | + AMD_CG_SUPPORT_GFX_3D_CGCG)) { + adev->gfx.rlc.funcs->set_safe_mode(adev); + adev->gfx.rlc.in_safe_mode = true; + } +} + +/** + * amdgpu_gfx_rlc_exit_safe_mode - Set RLC out of safe mode + * + * @adev: amdgpu_device pointer + * + * Set RLC exit safe mode if RLC is enabled and have entered into safe mode. + */ +void amdgpu_gfx_rlc_exit_safe_mode(struct amdgpu_device *adev) +{ + if (!(adev->gfx.rlc.in_safe_mode)) + return; + + /* if RLC is not enabled, do nothing */ + if (!adev->gfx.rlc.funcs->is_rlc_enabled(adev)) + return; + + if (adev->cg_flags & + (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG | + AMD_CG_SUPPORT_GFX_3D_CGCG)) { + adev->gfx.rlc.funcs->unset_safe_mode(adev); + adev->gfx.rlc.in_safe_mode = false; + } +} + +/** + * amdgpu_gfx_rlc_init_sr - Init save restore block + * + * @adev: amdgpu_device pointer + * @dws: the size of save restore block + * + * Allocate and setup value to save restore block of rlc. + * Returns 0 on succeess or negative error code if allocate failed. + */ +int amdgpu_gfx_rlc_init_sr(struct amdgpu_device *adev, u32 dws) +{ + const u32 *src_ptr; + volatile u32 *dst_ptr; + u32 i; + int r; + + /* allocate save restore block */ + r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + &adev->gfx.rlc.save_restore_obj, + &adev->gfx.rlc.save_restore_gpu_addr, + (void **)&adev->gfx.rlc.sr_ptr); + if (r) { + dev_warn(adev->dev, "(%d) create RLC sr bo failed\n", r); + amdgpu_gfx_rlc_fini(adev); + return r; + } + + /* write the sr buffer */ + src_ptr = adev->gfx.rlc.reg_list; + dst_ptr = adev->gfx.rlc.sr_ptr; + for (i = 0; i < adev->gfx.rlc.reg_list_size; i++) + dst_ptr[i] = cpu_to_le32(src_ptr[i]); + amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj); + amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj); + + return 0; +} + +/** + * amdgpu_gfx_rlc_init_csb - Init clear state block + * + * @adev: amdgpu_device pointer + * + * Allocate and setup value to clear state block of rlc. + * Returns 0 on succeess or negative error code if allocate failed. + */ +int amdgpu_gfx_rlc_init_csb(struct amdgpu_device *adev) +{ + volatile u32 *dst_ptr; + u32 dws; + int r; + + /* allocate clear state block */ + adev->gfx.rlc.clear_state_size = dws = adev->gfx.rlc.funcs->get_csb_size(adev); + r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + &adev->gfx.rlc.clear_state_obj, + &adev->gfx.rlc.clear_state_gpu_addr, + (void **)&adev->gfx.rlc.cs_ptr); + if (r) { + dev_err(adev->dev, "(%d) failed to create rlc csb bo\n", r); + amdgpu_gfx_rlc_fini(adev); + return r; + } + + /* set up the cs buffer */ + dst_ptr = adev->gfx.rlc.cs_ptr; + adev->gfx.rlc.funcs->get_csb_buffer(adev, dst_ptr); + amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj); + amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj); + amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj); + + return 0; +} + +/** + * amdgpu_gfx_rlc_init_cpt - Init cp table + * + * @adev: amdgpu_device pointer + * + * Allocate and setup value to cp table of rlc. + * Returns 0 on succeess or negative error code if allocate failed. + */ +int amdgpu_gfx_rlc_init_cpt(struct amdgpu_device *adev) +{ + int r; + + r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM, + &adev->gfx.rlc.cp_table_obj, + &adev->gfx.rlc.cp_table_gpu_addr, + (void **)&adev->gfx.rlc.cp_table_ptr); + if (r) { + dev_err(adev->dev, "(%d) failed to create cp table bo\n", r); + amdgpu_gfx_rlc_fini(adev); + return r; + } + + /* set up the cp table */ + amdgpu_gfx_rlc_setup_cp_table(adev); + amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj); + amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj); + + return 0; +} + +/** + * amdgpu_gfx_rlc_setup_cp_table - setup cp the buffer of cp table + * + * @adev: amdgpu_device pointer + * + * Write cp firmware data into cp table. + */ +void amdgpu_gfx_rlc_setup_cp_table(struct amdgpu_device *adev) +{ + const __le32 *fw_data; + volatile u32 *dst_ptr; + int me, i, max_me; + u32 bo_offset = 0; + u32 table_offset, table_size; + + max_me = adev->gfx.rlc.funcs->get_cp_table_num(adev); + + /* write the cp table buffer */ + dst_ptr = adev->gfx.rlc.cp_table_ptr; + for (me = 0; me < max_me; me++) { + if (me == 0) { + const struct gfx_firmware_header_v1_0 *hdr = + (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data; + fw_data = (const __le32 *) + (adev->gfx.ce_fw->data + + le32_to_cpu(hdr->header.ucode_array_offset_bytes)); + table_offset = le32_to_cpu(hdr->jt_offset); + table_size = le32_to_cpu(hdr->jt_size); + } else if (me == 1) { + const struct gfx_firmware_header_v1_0 *hdr = + (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data; + fw_data = (const __le32 *) + (adev->gfx.pfp_fw->data + + le32_to_cpu(hdr->header.ucode_array_offset_bytes)); + table_offset = le32_to_cpu(hdr->jt_offset); + table_size = le32_to_cpu(hdr->jt_size); + } else if (me == 2) { + const struct gfx_firmware_header_v1_0 *hdr = + (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data; + fw_data = (const __le32 *) + (adev->gfx.me_fw->data + + le32_to_cpu(hdr->header.ucode_array_offset_bytes)); + table_offset = le32_to_cpu(hdr->jt_offset); + table_size = le32_to_cpu(hdr->jt_size); + } else if (me == 3) { + const struct gfx_firmware_header_v1_0 *hdr = + (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data; + fw_data = (const __le32 *) + (adev->gfx.mec_fw->data + + le32_to_cpu(hdr->header.ucode_array_offset_bytes)); + table_offset = le32_to_cpu(hdr->jt_offset); + table_size = le32_to_cpu(hdr->jt_size); + } else if (me == 4) { + const struct gfx_firmware_header_v1_0 *hdr = + (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data; + fw_data = (const __le32 *) + (adev->gfx.mec2_fw->data + + le32_to_cpu(hdr->header.ucode_array_offset_bytes)); + table_offset = le32_to_cpu(hdr->jt_offset); + table_size = le32_to_cpu(hdr->jt_size); + } + + for (i = 0; i < table_size; i ++) { + dst_ptr[bo_offset + i] = + cpu_to_le32(le32_to_cpu(fw_data[table_offset + i])); + } + + bo_offset += table_size; + } +} + +/** + * amdgpu_gfx_rlc_fini - Free BO which used for RLC + * + * @adev: amdgpu_device pointer + * + * Free three BO which is used for rlc_save_restore_block, rlc_clear_state_block + * and rlc_jump_table_block. + */ +void amdgpu_gfx_rlc_fini(struct amdgpu_device *adev) +{ + /* save restore block */ + if (adev->gfx.rlc.save_restore_obj) { + amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, + &adev->gfx.rlc.save_restore_gpu_addr, + (void **)&adev->gfx.rlc.sr_ptr); + } + + /* clear state block */ + amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, + &adev->gfx.rlc.clear_state_gpu_addr, + (void **)&adev->gfx.rlc.cs_ptr); + + /* jump table block */ + amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, + &adev->gfx.rlc.cp_table_gpu_addr, + (void **)&adev->gfx.rlc.cp_table_ptr); +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h new file mode 100644 index 000000000000..49a8ab52113b --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h @@ -0,0 +1,98 @@ +/* + * 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 __AMDGPU_RLC_H__ +#define __AMDGPU_RLC_H__ + +#include "clearstate_defs.h" + +struct amdgpu_rlc_funcs { + bool (*is_rlc_enabled)(struct amdgpu_device *adev); + void (*set_safe_mode)(struct amdgpu_device *adev); + void (*unset_safe_mode)(struct amdgpu_device *adev); + int (*init)(struct amdgpu_device *adev); + u32 (*get_csb_size)(struct amdgpu_device *adev); + void (*get_csb_buffer)(struct amdgpu_device *adev, volatile u32 *buffer); + int (*get_cp_table_num)(struct amdgpu_device *adev); + int (*resume)(struct amdgpu_device *adev); + void (*stop)(struct amdgpu_device *adev); + void (*reset)(struct amdgpu_device *adev); + void (*start)(struct amdgpu_device *adev); +}; + +struct amdgpu_rlc { + /* for power gating */ + struct amdgpu_bo *save_restore_obj; + uint64_t save_restore_gpu_addr; + volatile uint32_t *sr_ptr; + const u32 *reg_list; + u32 reg_list_size; + /* for clear state */ + struct amdgpu_bo *clear_state_obj; + uint64_t clear_state_gpu_addr; + volatile uint32_t *cs_ptr; + const struct cs_section_def *cs_data; + u32 clear_state_size; + /* for cp tables */ + struct amdgpu_bo *cp_table_obj; + uint64_t cp_table_gpu_addr; + volatile uint32_t *cp_table_ptr; + u32 cp_table_size; + + /* safe mode for updating CG/PG state */ + bool in_safe_mode; + const struct amdgpu_rlc_funcs *funcs; + + /* for firmware data */ + u32 save_and_restore_offset; + u32 clear_state_descriptor_offset; + u32 avail_scratch_ram_locations; + u32 reg_restore_list_size; + u32 reg_list_format_start; + u32 reg_list_format_separate_start; + u32 starting_offsets_start; + u32 reg_list_format_size_bytes; + u32 reg_list_size_bytes; + u32 reg_list_format_direct_reg_list_length; + u32 save_restore_list_cntl_size_bytes; + u32 save_restore_list_gpm_size_bytes; + u32 save_restore_list_srm_size_bytes; + + u32 *register_list_format; + u32 *register_restore; + u8 *save_restore_list_cntl; + u8 *save_restore_list_gpm; + u8 *save_restore_list_srm; + + bool is_rlc_v2_1; +}; + +void amdgpu_gfx_rlc_enter_safe_mode(struct amdgpu_device *adev); +void amdgpu_gfx_rlc_exit_safe_mode(struct amdgpu_device *adev); +int amdgpu_gfx_rlc_init_sr(struct amdgpu_device *adev, u32 dws); +int amdgpu_gfx_rlc_init_csb(struct amdgpu_device *adev); +int amdgpu_gfx_rlc_init_cpt(struct amdgpu_device *adev); +void amdgpu_gfx_rlc_setup_cp_table(struct amdgpu_device *adev); +void amdgpu_gfx_rlc_fini(struct amdgpu_device *adev); + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c index bc9244b429ef..115bb0c99b0f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c @@ -28,17 +28,31 @@ * GPU SDMA IP block helpers function. */ -struct amdgpu_sdma_instance * amdgpu_get_sdma_instance(struct amdgpu_ring *ring) +struct amdgpu_sdma_instance *amdgpu_sdma_get_instance_from_ring(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; int i; for (i = 0; i < adev->sdma.num_instances; i++) - if (&adev->sdma.instance[i].ring == ring) - break; + if (ring == &adev->sdma.instance[i].ring || + ring == &adev->sdma.instance[i].page) + return &adev->sdma.instance[i]; - if (i < AMDGPU_MAX_SDMA_INSTANCES) - return &adev->sdma.instance[i]; - else - return NULL; + return NULL; +} + +int amdgpu_sdma_get_index_from_ring(struct amdgpu_ring *ring, uint32_t *index) +{ + struct amdgpu_device *adev = ring->adev; + int i; + + for (i = 0; i < adev->sdma.num_instances; i++) { + if (ring == &adev->sdma.instance[i].ring || + ring == &adev->sdma.instance[i].page) { + *index = i; + return 0; + } + } + + return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h index 500113ec65ca..16b1a6ae5ba6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h @@ -41,6 +41,7 @@ struct amdgpu_sdma_instance { uint32_t feature_version; struct amdgpu_ring ring; + struct amdgpu_ring page; bool burst_nop; }; @@ -50,6 +51,7 @@ struct amdgpu_sdma { struct amdgpu_irq_src illegal_inst_irq; int num_instances; uint32_t srbm_soft_reset; + bool has_page_queue; }; /* @@ -92,6 +94,7 @@ struct amdgpu_buffer_funcs { #define amdgpu_emit_fill_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_fill_buffer((ib), (s), (d), (b)) struct amdgpu_sdma_instance * -amdgpu_get_sdma_instance(struct amdgpu_ring *ring); +amdgpu_sdma_get_instance_from_ring(struct amdgpu_ring *ring); +int amdgpu_sdma_get_index_from_ring(struct amdgpu_ring *ring, uint32_t *index); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h index e9bf70e2ac51..626abca770a0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h @@ -218,6 +218,7 @@ TRACE_EVENT(amdgpu_vm_grab_id, TP_ARGS(vm, ring, job), TP_STRUCT__entry( __field(u32, pasid) + __string(ring, ring->name) __field(u32, ring) __field(u32, vmid) __field(u32, vm_hub) @@ -227,14 +228,14 @@ TRACE_EVENT(amdgpu_vm_grab_id, TP_fast_assign( __entry->pasid = vm->pasid; - __entry->ring = ring->idx; + __assign_str(ring, ring->name) __entry->vmid = job->vmid; __entry->vm_hub = ring->funcs->vmhub, __entry->pd_addr = job->vm_pd_addr; __entry->needs_flush = job->vm_needs_flush; ), - TP_printk("pasid=%d, ring=%u, id=%u, hub=%u, pd_addr=%010Lx needs_flush=%u", - __entry->pasid, __entry->ring, __entry->vmid, + TP_printk("pasid=%d, ring=%s, id=%u, hub=%u, pd_addr=%010Lx needs_flush=%u", + __entry->pasid, __get_str(ring), __entry->vmid, __entry->vm_hub, __entry->pd_addr, __entry->needs_flush) ); @@ -366,20 +367,20 @@ TRACE_EVENT(amdgpu_vm_flush, uint64_t pd_addr), TP_ARGS(ring, vmid, pd_addr), TP_STRUCT__entry( - __field(u32, ring) + __string(ring, ring->name) __field(u32, vmid) __field(u32, vm_hub) __field(u64, pd_addr) ), TP_fast_assign( - __entry->ring = ring->idx; + __assign_str(ring, ring->name) __entry->vmid = vmid; __entry->vm_hub = ring->funcs->vmhub; __entry->pd_addr = pd_addr; ), - TP_printk("ring=%u, id=%u, hub=%u, pd_addr=%010Lx", - __entry->ring, __entry->vmid, + TP_printk("ring=%s, id=%u, hub=%u, pd_addr=%010Lx", + __get_str(ring), __entry->vmid, __entry->vm_hub,__entry->pd_addr) ); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index a44fc12ae1f9..c91ec3101d00 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -61,100 +61,6 @@ static int amdgpu_map_buffer(struct ttm_buffer_object *bo, static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev); static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev); -/* - * Global memory. - */ - -/** - * amdgpu_ttm_mem_global_init - Initialize and acquire reference to - * memory object - * - * @ref: Object for initialization. - * - * This is called by drm_global_item_ref() when an object is being - * initialized. - */ -static int amdgpu_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -/** - * amdgpu_ttm_mem_global_release - Drop reference to a memory object - * - * @ref: Object being removed - * - * This is called by drm_global_item_unref() when an object is being - * released. - */ -static void amdgpu_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -/** - * amdgpu_ttm_global_init - Initialize global TTM memory reference structures. - * - * @adev: AMDGPU device for which the global structures need to be registered. - * - * This is called as part of the AMDGPU ttm init from amdgpu_ttm_init() - * during bring up. - */ -static int amdgpu_ttm_global_init(struct amdgpu_device *adev) -{ - struct drm_global_reference *global_ref; - int r; - - /* ensure reference is false in case init fails */ - adev->mman.mem_global_referenced = false; - - global_ref = &adev->mman.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &amdgpu_ttm_mem_global_init; - global_ref->release = &amdgpu_ttm_mem_global_release; - r = drm_global_item_ref(global_ref); - if (r) { - DRM_ERROR("Failed setting up TTM memory accounting " - "subsystem.\n"); - goto error_mem; - } - - adev->mman.bo_global_ref.mem_glob = - adev->mman.mem_global_ref.object; - global_ref = &adev->mman.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - r = drm_global_item_ref(global_ref); - if (r) { - DRM_ERROR("Failed setting up TTM BO subsystem.\n"); - goto error_bo; - } - - mutex_init(&adev->mman.gtt_window_lock); - - adev->mman.mem_global_referenced = true; - - return 0; - -error_bo: - drm_global_item_unref(&adev->mman.mem_global_ref); -error_mem: - return r; -} - -static void amdgpu_ttm_global_fini(struct amdgpu_device *adev) -{ - if (adev->mman.mem_global_referenced) { - mutex_destroy(&adev->mman.gtt_window_lock); - drm_global_item_unref(&adev->mman.bo_global_ref.ref); - drm_global_item_unref(&adev->mman.mem_global_ref); - adev->mman.mem_global_referenced = false; - } -} - static int amdgpu_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags) { return 0; @@ -1758,14 +1664,10 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) int r; u64 vis_vram_limit; - /* initialize global references for vram/gtt */ - r = amdgpu_ttm_global_init(adev); - if (r) { - return r; - } + mutex_init(&adev->mman.gtt_window_lock); + /* 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, &amdgpu_bo_driver, adev->ddev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, @@ -1922,7 +1824,6 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev) ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_GWS); ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_OA); ttm_bo_device_release(&adev->mman.bdev); - amdgpu_ttm_global_fini(adev); adev->mman.initialized = false; DRM_INFO("amdgpu: ttm finalized\n"); } @@ -2069,7 +1970,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, unsigned i; int r; - if (direct_submit && !ring->ready) { + if (direct_submit && !ring->sched.ready) { DRM_ERROR("Trying to move memory with ring turned off.\n"); return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index fe8f276e9811..b5b2d101f7db 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -39,8 +39,6 @@ #define AMDGPU_GTT_NUM_TRANSFER_WINDOWS 2 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; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h index aa6641b944a0..7ac25a1c7853 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h @@ -58,6 +58,17 @@ struct psp_firmware_header_v1_0 { }; /* version_major=1, version_minor=0 */ +struct ta_firmware_header_v1_0 { + struct common_firmware_header header; + uint32_t ta_xgmi_ucode_version; + uint32_t ta_xgmi_offset_bytes; + uint32_t ta_xgmi_size_bytes; + uint32_t ta_ras_ucode_version; + uint32_t ta_ras_offset_bytes; + uint32_t ta_ras_size_bytes; +}; + +/* version_major=1, version_minor=0 */ struct gfx_firmware_header_v1_0 { struct common_firmware_header header; uint32_t ucode_feature_version; @@ -170,6 +181,7 @@ union amdgpu_firmware_header { struct mc_firmware_header_v1_0 mc; struct smc_firmware_header_v1_0 smc; struct psp_firmware_header_v1_0 psp; + struct ta_firmware_header_v1_0 ta; struct gfx_firmware_header_v1_0 gfx; struct rlc_firmware_header_v1_0 rlc; struct rlc_firmware_header_v2_0 rlc_v2_0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c index e5a6db6beab7..69896f451e8a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c @@ -1243,30 +1243,20 @@ int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout) { struct dma_fence *fence; long r; - uint32_t ip_instance = ring->me; r = amdgpu_uvd_get_create_msg(ring, 1, NULL); - if (r) { - DRM_ERROR("amdgpu: (%d)failed to get create msg (%ld).\n", ip_instance, r); + if (r) goto error; - } r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence); - if (r) { - DRM_ERROR("amdgpu: (%d)failed to get destroy ib (%ld).\n", ip_instance, r); + if (r) goto error; - } r = dma_fence_wait_timeout(fence, false, timeout); - if (r == 0) { - DRM_ERROR("amdgpu: (%d)IB test timed out.\n", ip_instance); + if (r == 0) r = -ETIMEDOUT; - } else if (r < 0) { - DRM_ERROR("amdgpu: (%d)fence wait failed (%ld).\n", ip_instance, r); - } else { - DRM_DEBUG("ib test on (%d)ring %d succeeded\n", ip_instance, ring->idx); + else if (r > 0) r = 0; - } dma_fence_put(fence); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index 5f3f54073818..98a1b2ce2b9d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -1032,8 +1032,10 @@ out: * @ib: the IB to execute * */ -void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) +void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { amdgpu_ring_write(ring, VCE_CMD_IB); amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); @@ -1079,11 +1081,9 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring) return 0; r = amdgpu_ring_alloc(ring, 16); - if (r) { - DRM_ERROR("amdgpu: vce failed to lock ring %d (%d).\n", - ring->idx, r); + if (r) return r; - } + amdgpu_ring_write(ring, VCE_CMD_END); amdgpu_ring_commit(ring); @@ -1093,14 +1093,8 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed\n", - ring->idx); + if (i >= timeout) r = -ETIMEDOUT; - } return r; } @@ -1121,27 +1115,19 @@ int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout) return 0; r = amdgpu_vce_get_create_msg(ring, 1, NULL); - if (r) { - DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r); + if (r) goto error; - } r = amdgpu_vce_get_destroy_msg(ring, 1, true, &fence); - if (r) { - DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r); + if (r) goto error; - } r = dma_fence_wait_timeout(fence, false, timeout); - if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); + if (r == 0) r = -ETIMEDOUT; - } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); - } else { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + else if (r > 0) r = 0; - } + error: dma_fence_put(fence); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h index a1f209eed4c4..50293652af14 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h @@ -65,8 +65,8 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle, void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp); int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx); int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx); -void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch); +void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job, + struct amdgpu_ib *ib, bool ctx_switch); void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq, unsigned flags); int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index 27da13df2f11..e2e42e3fbcf3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -425,11 +425,9 @@ int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring) WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", - ring->idx, r); + if (r) return r; - } + amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0)); amdgpu_ring_write(ring, 0xDEADBEEF); @@ -441,14 +439,9 @@ int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + return r; } @@ -570,30 +563,20 @@ int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_vcn_dec_get_create_msg(ring, 1, NULL); - if (r) { - DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r); + if (r) goto error; - } r = amdgpu_vcn_dec_get_destroy_msg(ring, 1, &fence); - if (r) { - DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r); + if (r) goto error; - } r = dma_fence_wait_timeout(fence, false, timeout); - if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); + if (r == 0) r = -ETIMEDOUT; - } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); - } else { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + else if (r > 0) r = 0; - } dma_fence_put(fence); - error: return r; } @@ -606,11 +589,9 @@ int amdgpu_vcn_enc_ring_test_ring(struct amdgpu_ring *ring) int r; r = amdgpu_ring_alloc(ring, 16); - if (r) { - DRM_ERROR("amdgpu: vcn enc failed to lock ring %d (%d).\n", - ring->idx, r); + if (r) return r; - } + amdgpu_ring_write(ring, VCN_ENC_CMD_END); amdgpu_ring_commit(ring); @@ -620,14 +601,8 @@ int amdgpu_vcn_enc_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed\n", - ring->idx); + if (i >= adev->usec_timeout) r = -ETIMEDOUT; - } return r; } @@ -742,27 +717,19 @@ int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_vcn_enc_get_create_msg(ring, 1, NULL); - if (r) { - DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r); + if (r) goto error; - } r = amdgpu_vcn_enc_get_destroy_msg(ring, 1, &fence); - if (r) { - DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r); + if (r) goto error; - } r = dma_fence_wait_timeout(fence, false, timeout); - if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); + if (r == 0) r = -ETIMEDOUT; - } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); - } else { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + else if (r > 0) r = 0; - } + error: dma_fence_put(fence); return r; @@ -778,11 +745,8 @@ int amdgpu_vcn_jpeg_ring_test_ring(struct amdgpu_ring *ring) WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", - ring->idx, r); + if (r) return r; - } amdgpu_ring_write(ring, PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0, 0, 0)); @@ -796,14 +760,8 @@ int amdgpu_vcn_jpeg_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; return r; } @@ -856,21 +814,18 @@ int amdgpu_vcn_jpeg_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r = 0; r = amdgpu_vcn_jpeg_set_reg(ring, 1, &fence); - if (r) { - DRM_ERROR("amdgpu: failed to set jpeg register (%ld).\n", r); + if (r) goto error; - } r = dma_fence_wait_timeout(fence, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); r = -ETIMEDOUT; goto error; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto error; - } else + } else { r = 0; + } for (i = 0; i < adev->usec_timeout; i++) { tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9)); @@ -879,15 +834,10 @@ int amdgpu_vcn_jpeg_ring_test_ib(struct amdgpu_ring *ring, long timeout) DRM_UDELAY(1); } - if (i < adev->usec_timeout) - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); - else { - DRM_ERROR("ib test failed (0x%08X)\n", tmp); - r = -EINVAL; - } + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; dma_fence_put(fence); - error: return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index f2f358aa0597..cfee74732edb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -23,16 +23,6 @@ #include "amdgpu.h" -uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev) -{ - uint64_t addr = adev->vm_manager.max_pfn << AMDGPU_GPU_PAGE_SHIFT; - - addr -= AMDGPU_VA_RESERVED_SIZE; - addr = amdgpu_gmc_sign_extend(addr); - - return addr; -} - bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev) { /* By now all MMIO pages except mailbox are blocked */ @@ -41,88 +31,6 @@ bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev) return RREG32_NO_KIQ(0xc040) == 0xffffffff; } -int amdgpu_allocate_static_csa(struct amdgpu_device *adev) -{ - int r; - void *ptr; - - r = amdgpu_bo_create_kernel(adev, AMDGPU_CSA_SIZE, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, &adev->virt.csa_obj, - &adev->virt.csa_vmid0_addr, &ptr); - if (r) - return r; - - memset(ptr, 0, AMDGPU_CSA_SIZE); - return 0; -} - -void amdgpu_free_static_csa(struct amdgpu_device *adev) { - amdgpu_bo_free_kernel(&adev->virt.csa_obj, - &adev->virt.csa_vmid0_addr, - NULL); -} - -/* - * amdgpu_map_static_csa should be called during amdgpu_vm_init - * it maps virtual address amdgpu_csa_vaddr() to this VM, and each command - * submission of GFX should use this virtual address within META_DATA init - * package to support SRIOV gfx preemption. - */ -int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, - struct amdgpu_bo_va **bo_va) -{ - uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_GMC_HOLE_MASK; - struct ww_acquire_ctx ticket; - struct list_head list; - struct amdgpu_bo_list_entry pd; - struct ttm_validate_buffer csa_tv; - int r; - - INIT_LIST_HEAD(&list); - INIT_LIST_HEAD(&csa_tv.head); - csa_tv.bo = &adev->virt.csa_obj->tbo; - csa_tv.shared = true; - - list_add(&csa_tv.head, &list); - amdgpu_vm_get_pd_bo(vm, &list, &pd); - - r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL); - if (r) { - DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r); - return r; - } - - *bo_va = amdgpu_vm_bo_add(adev, vm, adev->virt.csa_obj); - if (!*bo_va) { - ttm_eu_backoff_reservation(&ticket, &list); - DRM_ERROR("failed to create bo_va for static CSA\n"); - return -ENOMEM; - } - - r = amdgpu_vm_alloc_pts(adev, (*bo_va)->base.vm, csa_addr, - AMDGPU_CSA_SIZE); - if (r) { - DRM_ERROR("failed to allocate pts for static CSA, err=%d\n", r); - amdgpu_vm_bo_rmv(adev, *bo_va); - ttm_eu_backoff_reservation(&ticket, &list); - return r; - } - - r = amdgpu_vm_bo_map(adev, *bo_va, csa_addr, 0, AMDGPU_CSA_SIZE, - AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE | - AMDGPU_PTE_EXECUTABLE); - - if (r) { - DRM_ERROR("failed to do bo_map on static CSA, err=%d\n", r); - amdgpu_vm_bo_rmv(adev, *bo_va); - ttm_eu_backoff_reservation(&ticket, &list); - return r; - } - - ttm_eu_backoff_reservation(&ticket, &list); - return 0; -} - void amdgpu_virt_init_setting(struct amdgpu_device *adev) { /* enable virtual display */ @@ -162,9 +70,7 @@ uint32_t amdgpu_virt_kiq_rreg(struct amdgpu_device *adev, uint32_t reg) if (r < 1 && (adev->in_gpu_reset || in_interrupt())) goto failed_kiq_read; - if (in_interrupt()) - might_sleep(); - + might_sleep(); while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) { msleep(MAX_KIQ_REG_BAILOUT_INTERVAL); r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); @@ -210,9 +116,7 @@ void amdgpu_virt_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v) if (r < 1 && (adev->in_gpu_reset || in_interrupt())) goto failed_kiq_write; - if (in_interrupt()) - might_sleep(); - + might_sleep(); while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) { msleep(MAX_KIQ_REG_BAILOUT_INTERVAL); @@ -228,6 +132,46 @@ failed_kiq_write: pr_err("failed to write reg:%x\n", reg); } +void amdgpu_virt_kiq_reg_write_reg_wait(struct amdgpu_device *adev, + uint32_t reg0, uint32_t reg1, + uint32_t ref, uint32_t mask) +{ + struct amdgpu_kiq *kiq = &adev->gfx.kiq; + struct amdgpu_ring *ring = &kiq->ring; + signed long r, cnt = 0; + unsigned long flags; + uint32_t seq; + + spin_lock_irqsave(&kiq->ring_lock, flags); + amdgpu_ring_alloc(ring, 32); + amdgpu_ring_emit_reg_write_reg_wait(ring, reg0, reg1, + ref, mask); + amdgpu_fence_emit_polling(ring, &seq); + amdgpu_ring_commit(ring); + spin_unlock_irqrestore(&kiq->ring_lock, flags); + + r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); + + /* don't wait anymore for IRQ context */ + if (r < 1 && in_interrupt()) + goto failed_kiq; + + might_sleep(); + while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) { + + msleep(MAX_KIQ_REG_BAILOUT_INTERVAL); + r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); + } + + if (cnt > MAX_KIQ_REG_TRY) + goto failed_kiq; + + return; + +failed_kiq: + pr_err("failed to write reg %x wait reg %x\n", reg0, reg1); +} + /** * amdgpu_virt_request_full_gpu() - request full gpu access * @amdgpu: amdgpu device. diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 880ac113a3a9..0728fbc9a692 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -238,7 +238,6 @@ typedef struct amdgim_vf2pf_info_v2 amdgim_vf2pf_info ; struct amdgpu_virt { uint32_t caps; struct amdgpu_bo *csa_obj; - uint64_t csa_vmid0_addr; bool chained_ib_support; uint32_t reg_val_offs; struct amdgpu_irq_src ack_irq; @@ -251,8 +250,6 @@ struct amdgpu_virt { uint32_t gim_feature; }; -#define AMDGPU_CSA_SIZE (8 * 1024) - #define amdgpu_sriov_enabled(adev) \ ((adev)->virt.caps & AMDGPU_SRIOV_CAPS_ENABLE_IOV) @@ -277,17 +274,13 @@ static inline bool is_virtual_machine(void) #endif } -struct amdgpu_vm; - -uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev); bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev); -int amdgpu_allocate_static_csa(struct amdgpu_device *adev); -int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, - struct amdgpu_bo_va **bo_va); -void amdgpu_free_static_csa(struct amdgpu_device *adev); void amdgpu_virt_init_setting(struct amdgpu_device *adev); uint32_t amdgpu_virt_kiq_rreg(struct amdgpu_device *adev, uint32_t reg); void amdgpu_virt_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v); +void amdgpu_virt_kiq_reg_write_reg_wait(struct amdgpu_device *adev, + uint32_t reg0, uint32_t rreg1, + uint32_t ref, uint32_t mask); int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init); int amdgpu_virt_release_full_gpu(struct amdgpu_device *adev, bool init); int amdgpu_virt_reset_gpu(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 897afbb348c1..909216a9b447 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -63,7 +63,7 @@ static struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev) int amdgpu_xgmi_add_device(struct amdgpu_device *adev) { - struct psp_xgmi_topology_info tmp_topology[AMDGPU_MAX_XGMI_DEVICE_PER_HIVE]; + struct psp_xgmi_topology_info *tmp_topology; struct amdgpu_hive_info *hive; struct amdgpu_xgmi *entry; struct amdgpu_device *tmp_adev; @@ -73,10 +73,12 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) if ((adev->asic_type < CHIP_VEGA20) || (adev->flags & AMD_IS_APU) ) return 0; - adev->gmc.xgmi.device_id = psp_xgmi_get_device_id(&adev->psp); + adev->gmc.xgmi.node_id = psp_xgmi_get_node_id(&adev->psp); adev->gmc.xgmi.hive_id = psp_xgmi_get_hive_id(&adev->psp); - memset(&tmp_topology[0], 0, sizeof(tmp_topology)); + tmp_topology = kzalloc(sizeof(struct psp_xgmi_topology_info), GFP_KERNEL); + if (!tmp_topology) + return -ENOMEM; mutex_lock(&xgmi_mutex); hive = amdgpu_get_xgmi_hive(adev); if (!hive) @@ -84,23 +86,28 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) list_add_tail(&adev->gmc.xgmi.head, &hive->device_list); list_for_each_entry(entry, &hive->device_list, head) - tmp_topology[count++].device_id = entry->device_id; - - ret = psp_xgmi_get_topology_info(&adev->psp, count, tmp_topology); - if (ret) { - dev_err(adev->dev, - "XGMI: Get topology failure on device %llx, hive %llx, ret %d", - adev->gmc.xgmi.device_id, - adev->gmc.xgmi.hive_id, ret); - goto exit; + tmp_topology->nodes[count++].node_id = entry->node_id; + + /* Each psp need to get the latest topology */ + list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) { + ret = psp_xgmi_get_topology_info(&tmp_adev->psp, count, tmp_topology); + if (ret) { + dev_err(tmp_adev->dev, + "XGMI: Get topology failure on device %llx, hive %llx, ret %d", + tmp_adev->gmc.xgmi.node_id, + tmp_adev->gmc.xgmi.hive_id, ret); + /* To do : continue with some node failed or disable the whole hive */ + break; + } } + /* Each psp need to set the latest topology */ list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) { ret = psp_xgmi_set_topology_info(&tmp_adev->psp, count, tmp_topology); if (ret) { dev_err(tmp_adev->dev, "XGMI: Set topology failure on device %llx, hive %llx, ret %d", - tmp_adev->gmc.xgmi.device_id, + tmp_adev->gmc.xgmi.node_id, tmp_adev->gmc.xgmi.hive_id, ret); /* To do : continue with some node failed or disable the whole hive */ break; @@ -113,7 +120,6 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) exit: mutex_unlock(&xgmi_mutex); + kfree(tmp_topology); return ret; } - - diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c index 79220a91abe3..86e14c754dd4 100644 --- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c @@ -743,19 +743,19 @@ static int ci_enable_didt(struct amdgpu_device *adev, bool enable) if (pi->caps_sq_ramping || pi->caps_db_ramping || pi->caps_td_ramping || pi->caps_tcp_ramping) { - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); if (enable) { ret = ci_program_pt_config_registers(adev, didt_config_ci); if (ret) { - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return ret; } } ci_do_enable_didt(adev, enable); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c index b918c8886b75..45795191de1f 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c @@ -198,7 +198,7 @@ static void cik_sdma_ring_set_wptr(struct amdgpu_ring *ring) static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) { - struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring); + struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring); int i; for (i = 0; i < count; i++) @@ -218,9 +218,11 @@ static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) * Schedule an IB in the DMA ring (CIK). */ static void cik_sdma_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 extra_bits = vmid & 0xf; /* IB packet must end on a 8 DW boundary */ @@ -316,8 +318,8 @@ static void cik_sdma_gfx_stop(struct amdgpu_device *adev) WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl); WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], 0); } - sdma0->ready = false; - sdma1->ready = false; + sdma0->sched.ready = false; + sdma1->sched.ready = false; } /** @@ -494,18 +496,16 @@ static int cik_sdma_gfx_resume(struct amdgpu_device *adev) /* enable DMA IBs */ WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl); - ring->ready = true; + ring->sched.ready = true; } cik_sdma_enable(adev, true); for (i = 0; i < adev->sdma.num_instances; i++) { ring = &adev->sdma.instance[i].ring; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) return r; - } if (adev->mman.buffer_funcs_ring == ring) amdgpu_ttm_set_buffer_funcs_status(adev, true); @@ -618,21 +618,17 @@ static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring) u64 gpu_addr; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); r = amdgpu_ring_alloc(ring, 5); - if (r) { - DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_device_wb_free(adev, index); - return r; - } + if (r) + goto error_free_wb; + amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0)); amdgpu_ring_write(ring, lower_32_bits(gpu_addr)); amdgpu_ring_write(ring, upper_32_bits(gpu_addr)); @@ -647,15 +643,11 @@ static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } - amdgpu_device_wb_free(adev, index); + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; +error_free_wb: + amdgpu_device_wb_free(adev, index); return r; } @@ -678,20 +670,16 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 256, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err0; - } ib.ptr[0] = SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0); @@ -706,21 +694,16 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out\n"); r = -ETIMEDOUT; goto err1; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto err1; } tmp = le32_to_cpu(adev->wb.wb[index]); - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + if (tmp == 0xDEADBEEF) r = 0; - } else { - DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp); + else r = -EINVAL; - } err1: amdgpu_ib_free(adev, &ib, NULL); @@ -822,7 +805,7 @@ static void cik_sdma_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe, */ static void cik_sdma_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib) { - struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring); + struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring); u32 pad_count; int i; @@ -1214,8 +1197,11 @@ static int cik_sdma_process_illegal_inst_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { + u8 instance_id; + DRM_ERROR("Illegal instruction in SDMA command stream\n"); - schedule_work(&adev->reset_work); + instance_id = (entry->ring_id & 0x3) >> 0; + drm_sched_fault(&adev->sdma.instance[instance_id].ring.sched); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c index d76eb27945dc..1dc3013ea1d5 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c @@ -1775,18 +1775,15 @@ static int gfx_v6_0_ring_test_ring(struct amdgpu_ring *ring) int r; r = amdgpu_gfx_scratch_get(adev, &scratch); - if (r) { - DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r); + if (r) return r; - } + WREG32(scratch, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_gfx_scratch_free(adev, scratch); - return r; - } + if (r) + goto error_free_scratch; + amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1)); amdgpu_ring_write(ring, (scratch - PACKET3_SET_CONFIG_REG_START)); amdgpu_ring_write(ring, 0xDEADBEEF); @@ -1798,13 +1795,11 @@ static int gfx_v6_0_ring_test_ring(struct amdgpu_ring *ring) break; DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n", - ring->idx, scratch, tmp); - r = -EINVAL; - } + + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + +error_free_scratch: amdgpu_gfx_scratch_free(adev, scratch); return r; } @@ -1845,9 +1840,11 @@ static void gfx_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, } static void gfx_v6_0_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 header, control = 0; /* insert SWITCH_BUFFER packet before first IB in the ring frame */ @@ -1892,17 +1889,15 @@ static int gfx_v6_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_gfx_scratch_get(adev, &scratch); - if (r) { - DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r); + if (r) return r; - } + WREG32(scratch, 0xCAFEDEAD); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 256, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err1; - } + ib.ptr[0] = PACKET3(PACKET3_SET_CONFIG_REG, 1); ib.ptr[1] = ((scratch - PACKET3_SET_CONFIG_REG_START)); ib.ptr[2] = 0xDEADBEEF; @@ -1914,22 +1909,16 @@ static int gfx_v6_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out\n"); r = -ETIMEDOUT; goto err2; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto err2; } tmp = RREG32(scratch); - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + if (tmp == 0xDEADBEEF) r = 0; - } else { - DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n", - scratch, tmp); + else r = -EINVAL; - } err2: amdgpu_ib_free(adev, &ib, NULL); @@ -1950,9 +1939,9 @@ static void gfx_v6_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable) CP_ME_CNTL__CE_HALT_MASK)); WREG32(mmSCRATCH_UMSK, 0); for (i = 0; i < adev->gfx.num_gfx_rings; i++) - adev->gfx.gfx_ring[i].ready = false; + adev->gfx.gfx_ring[i].sched.ready = false; for (i = 0; i < adev->gfx.num_compute_rings; i++) - adev->gfx.compute_ring[i].ready = false; + adev->gfx.compute_ring[i].sched.ready = false; } udelay(50); } @@ -2124,12 +2113,9 @@ static int gfx_v6_0_cp_gfx_resume(struct amdgpu_device *adev) /* start the rings */ gfx_v6_0_cp_gfx_start(adev); - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) return r; - } return 0; } @@ -2227,14 +2213,11 @@ static int gfx_v6_0_cp_compute_resume(struct amdgpu_device *adev) WREG32(mmCP_RB2_CNTL, tmp); WREG32(mmCP_RB2_BASE, ring->gpu_addr >> 8); - adev->gfx.compute_ring[0].ready = false; - adev->gfx.compute_ring[1].ready = false; for (i = 0; i < 2; i++) { - r = amdgpu_ring_test_ring(&adev->gfx.compute_ring[i]); + r = amdgpu_ring_test_helper(&adev->gfx.compute_ring[i]); if (r) return r; - adev->gfx.compute_ring[i].ready = true; } return 0; @@ -2368,18 +2351,11 @@ static void gfx_v6_0_ring_emit_wreg(struct amdgpu_ring *ring, amdgpu_ring_write(ring, val); } -static void gfx_v6_0_rlc_fini(struct amdgpu_device *adev) -{ - amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL); - amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL); - amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL); -} - static int gfx_v6_0_rlc_init(struct amdgpu_device *adev) { const u32 *src_ptr; volatile u32 *dst_ptr; - u32 dws, i; + u32 dws; u64 reg_list_mc_addr; const struct cs_section_def *cs_data; int r; @@ -2394,26 +2370,10 @@ static int gfx_v6_0_rlc_init(struct amdgpu_device *adev) cs_data = adev->gfx.rlc.cs_data; if (src_ptr) { - /* save restore block */ - r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->gfx.rlc.save_restore_obj, - &adev->gfx.rlc.save_restore_gpu_addr, - (void **)&adev->gfx.rlc.sr_ptr); - if (r) { - dev_warn(adev->dev, "(%d) create RLC sr bo failed\n", - r); - gfx_v6_0_rlc_fini(adev); + /* init save restore block */ + r = amdgpu_gfx_rlc_init_sr(adev, dws); + if (r) return r; - } - - /* write the sr buffer */ - dst_ptr = adev->gfx.rlc.sr_ptr; - for (i = 0; i < adev->gfx.rlc.reg_list_size; i++) - dst_ptr[i] = cpu_to_le32(src_ptr[i]); - - amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj); - amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj); } if (cs_data) { @@ -2428,7 +2388,7 @@ static int gfx_v6_0_rlc_init(struct amdgpu_device *adev) (void **)&adev->gfx.rlc.cs_ptr); if (r) { dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r); - gfx_v6_0_rlc_fini(adev); + amdgpu_gfx_rlc_fini(adev); return r; } @@ -2549,8 +2509,8 @@ static int gfx_v6_0_rlc_resume(struct amdgpu_device *adev) if (!adev->gfx.rlc_fw) return -EINVAL; - gfx_v6_0_rlc_stop(adev); - gfx_v6_0_rlc_reset(adev); + adev->gfx.rlc.funcs->stop(adev); + adev->gfx.rlc.funcs->reset(adev); gfx_v6_0_init_pg(adev); gfx_v6_0_init_cg(adev); @@ -2578,7 +2538,7 @@ static int gfx_v6_0_rlc_resume(struct amdgpu_device *adev) WREG32(mmRLC_UCODE_ADDR, 0); gfx_v6_0_enable_lbpw(adev, gfx_v6_0_lbpw_supported(adev)); - gfx_v6_0_rlc_start(adev); + adev->gfx.rlc.funcs->start(adev); return 0; } @@ -3075,6 +3035,14 @@ static const struct amdgpu_gfx_funcs gfx_v6_0_gfx_funcs = { .select_me_pipe_q = &gfx_v6_0_select_me_pipe_q }; +static const struct amdgpu_rlc_funcs gfx_v6_0_rlc_funcs = { + .init = gfx_v6_0_rlc_init, + .resume = gfx_v6_0_rlc_resume, + .stop = gfx_v6_0_rlc_stop, + .reset = gfx_v6_0_rlc_reset, + .start = gfx_v6_0_rlc_start +}; + static int gfx_v6_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -3082,6 +3050,7 @@ static int gfx_v6_0_early_init(void *handle) adev->gfx.num_gfx_rings = GFX6_NUM_GFX_RINGS; adev->gfx.num_compute_rings = GFX6_NUM_COMPUTE_RINGS; adev->gfx.funcs = &gfx_v6_0_gfx_funcs; + adev->gfx.rlc.funcs = &gfx_v6_0_rlc_funcs; gfx_v6_0_set_ring_funcs(adev); gfx_v6_0_set_irq_funcs(adev); @@ -3114,7 +3083,7 @@ static int gfx_v6_0_sw_init(void *handle) return r; } - r = gfx_v6_0_rlc_init(adev); + r = adev->gfx.rlc.funcs->init(adev); if (r) { DRM_ERROR("Failed to init rlc BOs!\n"); return r; @@ -3165,7 +3134,7 @@ static int gfx_v6_0_sw_fini(void *handle) for (i = 0; i < adev->gfx.num_compute_rings; i++) amdgpu_ring_fini(&adev->gfx.compute_ring[i]); - gfx_v6_0_rlc_fini(adev); + amdgpu_gfx_rlc_fini(adev); return 0; } @@ -3177,7 +3146,7 @@ static int gfx_v6_0_hw_init(void *handle) gfx_v6_0_constants_init(adev); - r = gfx_v6_0_rlc_resume(adev); + r = adev->gfx.rlc.funcs->resume(adev); if (r) return r; @@ -3195,7 +3164,7 @@ static int gfx_v6_0_hw_fini(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; gfx_v6_0_cp_enable(adev, false); - gfx_v6_0_rlc_stop(adev); + adev->gfx.rlc.funcs->stop(adev); gfx_v6_0_fini_pg(adev); return 0; @@ -3393,12 +3362,31 @@ static int gfx_v6_0_eop_irq(struct amdgpu_device *adev, return 0; } +static void gfx_v6_0_fault(struct amdgpu_device *adev, + struct amdgpu_iv_entry *entry) +{ + struct amdgpu_ring *ring; + + switch (entry->ring_id) { + case 0: + ring = &adev->gfx.gfx_ring[0]; + break; + case 1: + case 2: + ring = &adev->gfx.compute_ring[entry->ring_id - 1]; + break; + default: + return; + } + drm_sched_fault(&ring->sched); +} + static int gfx_v6_0_priv_reg_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { DRM_ERROR("Illegal register access in command stream\n"); - schedule_work(&adev->reset_work); + gfx_v6_0_fault(adev, entry); return 0; } @@ -3407,7 +3395,7 @@ static int gfx_v6_0_priv_inst_irq(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { DRM_ERROR("Illegal instruction in command stream\n"); - schedule_work(&adev->reset_work); + gfx_v6_0_fault(adev, entry); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index 0e72bc09939a..f467b9bd090d 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -882,7 +882,6 @@ static const u32 kalindi_rlc_save_restore_register_list[] = static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev); static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer); -static void gfx_v7_0_init_cp_pg_table(struct amdgpu_device *adev); static void gfx_v7_0_init_pg(struct amdgpu_device *adev); static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev); @@ -2064,17 +2063,14 @@ static int gfx_v7_0_ring_test_ring(struct amdgpu_ring *ring) int r; r = amdgpu_gfx_scratch_get(adev, &scratch); - if (r) { - DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r); + if (r) return r; - } + WREG32(scratch, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_gfx_scratch_free(adev, scratch); - return r; - } + if (r) + goto error_free_scratch; + amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1)); amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START)); amdgpu_ring_write(ring, 0xDEADBEEF); @@ -2086,13 +2082,10 @@ static int gfx_v7_0_ring_test_ring(struct amdgpu_ring *ring) break; DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n", - ring->idx, scratch, tmp); - r = -EINVAL; - } + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + +error_free_scratch: amdgpu_gfx_scratch_free(adev, scratch); return r; } @@ -2233,9 +2226,11 @@ static void gfx_v7_0_ring_emit_fence_compute(struct amdgpu_ring *ring, * on the gfx ring for execution by the GPU. */ static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 header, control = 0; /* insert SWITCH_BUFFER packet before first IB in the ring frame */ @@ -2262,9 +2257,11 @@ static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, } static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24); amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); @@ -2316,17 +2313,15 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_gfx_scratch_get(adev, &scratch); - if (r) { - DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r); + if (r) return r; - } + WREG32(scratch, 0xCAFEDEAD); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 256, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err1; - } + ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1); ib.ptr[1] = ((scratch - PACKET3_SET_UCONFIG_REG_START)); ib.ptr[2] = 0xDEADBEEF; @@ -2338,22 +2333,16 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out\n"); r = -ETIMEDOUT; goto err2; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto err2; } tmp = RREG32(scratch); - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + if (tmp == 0xDEADBEEF) r = 0; - } else { - DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n", - scratch, tmp); + else r = -EINVAL; - } err2: amdgpu_ib_free(adev, &ib, NULL); @@ -2403,7 +2392,7 @@ static void gfx_v7_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable) } else { WREG32(mmCP_ME_CNTL, (CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK)); for (i = 0; i < adev->gfx.num_gfx_rings; i++) - adev->gfx.gfx_ring[i].ready = false; + adev->gfx.gfx_ring[i].sched.ready = false; } udelay(50); } @@ -2613,12 +2602,9 @@ static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev) /* start the ring */ gfx_v7_0_cp_gfx_start(adev); - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) return r; - } return 0; } @@ -2675,7 +2661,7 @@ static void gfx_v7_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) } else { WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK)); for (i = 0; i < adev->gfx.num_compute_rings; i++) - adev->gfx.compute_ring[i].ready = false; + adev->gfx.compute_ring[i].sched.ready = false; } udelay(50); } @@ -2781,7 +2767,7 @@ static int gfx_v7_0_mec_init(struct amdgpu_device *adev) * GFX7_MEC_HPD_SIZE * 2; r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_GTT, + AMDGPU_GEM_DOMAIN_VRAM, &adev->gfx.mec.hpd_eop_obj, &adev->gfx.mec.hpd_eop_gpu_addr, (void **)&hpd); @@ -3106,10 +3092,7 @@ static int gfx_v7_0_cp_compute_resume(struct amdgpu_device *adev) for (i = 0; i < adev->gfx.num_compute_rings; i++) { ring = &adev->gfx.compute_ring[i]; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) - ring->ready = false; + amdgpu_ring_test_helper(ring); } return 0; @@ -3268,18 +3251,10 @@ static void gfx_v7_0_ring_emit_wreg(struct amdgpu_ring *ring, * The RLC is a multi-purpose microengine that handles a * variety of functions. */ -static void gfx_v7_0_rlc_fini(struct amdgpu_device *adev) -{ - amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL); - amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL); - amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL); -} - static int gfx_v7_0_rlc_init(struct amdgpu_device *adev) { const u32 *src_ptr; - volatile u32 *dst_ptr; - u32 dws, i; + u32 dws; const struct cs_section_def *cs_data; int r; @@ -3306,66 +3281,23 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev) cs_data = adev->gfx.rlc.cs_data; if (src_ptr) { - /* save restore block */ - r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->gfx.rlc.save_restore_obj, - &adev->gfx.rlc.save_restore_gpu_addr, - (void **)&adev->gfx.rlc.sr_ptr); - if (r) { - dev_warn(adev->dev, "(%d) create, pin or map of RLC sr bo failed\n", r); - gfx_v7_0_rlc_fini(adev); + /* init save restore block */ + r = amdgpu_gfx_rlc_init_sr(adev, dws); + if (r) return r; - } - - /* write the sr buffer */ - dst_ptr = adev->gfx.rlc.sr_ptr; - for (i = 0; i < adev->gfx.rlc.reg_list_size; i++) - dst_ptr[i] = cpu_to_le32(src_ptr[i]); - amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj); - amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj); } if (cs_data) { - /* clear state block */ - adev->gfx.rlc.clear_state_size = dws = gfx_v7_0_get_csb_size(adev); - - r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->gfx.rlc.clear_state_obj, - &adev->gfx.rlc.clear_state_gpu_addr, - (void **)&adev->gfx.rlc.cs_ptr); - if (r) { - dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r); - gfx_v7_0_rlc_fini(adev); + /* init clear state block */ + r = amdgpu_gfx_rlc_init_csb(adev); + if (r) return r; - } - - /* set up the cs buffer */ - dst_ptr = adev->gfx.rlc.cs_ptr; - gfx_v7_0_get_csb_buffer(adev, dst_ptr); - amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj); - amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj); } if (adev->gfx.rlc.cp_table_size) { - - r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size, - PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM, - &adev->gfx.rlc.cp_table_obj, - &adev->gfx.rlc.cp_table_gpu_addr, - (void **)&adev->gfx.rlc.cp_table_ptr); - if (r) { - dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r); - gfx_v7_0_rlc_fini(adev); + r = amdgpu_gfx_rlc_init_cpt(adev); + if (r) return r; - } - - gfx_v7_0_init_cp_pg_table(adev); - - amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj); - amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj); - } return 0; @@ -3446,7 +3378,12 @@ static u32 gfx_v7_0_halt_rlc(struct amdgpu_device *adev) return orig; } -static void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev) +static bool gfx_v7_0_is_rlc_enabled(struct amdgpu_device *adev) +{ + return true; +} + +static void gfx_v7_0_set_safe_mode(struct amdgpu_device *adev) { u32 tmp, i, mask; @@ -3468,7 +3405,7 @@ static void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev) } } -static void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev) +static void gfx_v7_0_unset_safe_mode(struct amdgpu_device *adev) { u32 tmp; @@ -3545,13 +3482,13 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev) adev->gfx.rlc_feature_version = le32_to_cpu( hdr->ucode_feature_version); - gfx_v7_0_rlc_stop(adev); + adev->gfx.rlc.funcs->stop(adev); /* disable CG */ tmp = RREG32(mmRLC_CGCG_CGLS_CTRL) & 0xfffffffc; WREG32(mmRLC_CGCG_CGLS_CTRL, tmp); - gfx_v7_0_rlc_reset(adev); + adev->gfx.rlc.funcs->reset(adev); gfx_v7_0_init_pg(adev); @@ -3582,7 +3519,7 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev) if (adev->asic_type == CHIP_BONAIRE) WREG32(mmRLC_DRIVER_CPDMA_STATUS, 0); - gfx_v7_0_rlc_start(adev); + adev->gfx.rlc.funcs->start(adev); return 0; } @@ -3784,72 +3721,12 @@ static void gfx_v7_0_enable_gds_pg(struct amdgpu_device *adev, bool enable) WREG32(mmRLC_PG_CNTL, data); } -static void gfx_v7_0_init_cp_pg_table(struct amdgpu_device *adev) +static int gfx_v7_0_cp_pg_table_num(struct amdgpu_device *adev) { - const __le32 *fw_data; - volatile u32 *dst_ptr; - int me, i, max_me = 4; - u32 bo_offset = 0; - u32 table_offset, table_size; - if (adev->asic_type == CHIP_KAVERI) - max_me = 5; - - if (adev->gfx.rlc.cp_table_ptr == NULL) - return; - - /* write the cp table buffer */ - dst_ptr = adev->gfx.rlc.cp_table_ptr; - for (me = 0; me < max_me; me++) { - if (me == 0) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data; - fw_data = (const __le32 *) - (adev->gfx.ce_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 1) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data; - fw_data = (const __le32 *) - (adev->gfx.pfp_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 2) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data; - fw_data = (const __le32 *) - (adev->gfx.me_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 3) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data; - fw_data = (const __le32 *) - (adev->gfx.mec_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data; - fw_data = (const __le32 *) - (adev->gfx.mec2_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } - - for (i = 0; i < table_size; i ++) { - dst_ptr[bo_offset + i] = - cpu_to_le32(le32_to_cpu(fw_data[table_offset + i])); - } - - bo_offset += table_size; - } + return 5; + else + return 4; } static void gfx_v7_0_enable_gfx_cgpg(struct amdgpu_device *adev, @@ -4288,8 +4165,17 @@ static const struct amdgpu_gfx_funcs gfx_v7_0_gfx_funcs = { }; static const struct amdgpu_rlc_funcs gfx_v7_0_rlc_funcs = { - .enter_safe_mode = gfx_v7_0_enter_rlc_safe_mode, - .exit_safe_mode = gfx_v7_0_exit_rlc_safe_mode + .is_rlc_enabled = gfx_v7_0_is_rlc_enabled, + .set_safe_mode = gfx_v7_0_set_safe_mode, + .unset_safe_mode = gfx_v7_0_unset_safe_mode, + .init = gfx_v7_0_rlc_init, + .get_csb_size = gfx_v7_0_get_csb_size, + .get_csb_buffer = gfx_v7_0_get_csb_buffer, + .get_cp_table_num = gfx_v7_0_cp_pg_table_num, + .resume = gfx_v7_0_rlc_resume, + .stop = gfx_v7_0_rlc_stop, + .reset = gfx_v7_0_rlc_reset, + .start = gfx_v7_0_rlc_start }; static int gfx_v7_0_early_init(void *handle) @@ -4540,7 +4426,7 @@ static int gfx_v7_0_sw_init(void *handle) return r; } - r = gfx_v7_0_rlc_init(adev); + r = adev->gfx.rlc.funcs->init(adev); if (r) { DRM_ERROR("Failed to init rlc BOs!\n"); return r; @@ -4604,7 +4490,7 @@ static int gfx_v7_0_sw_fini(void *handle) amdgpu_ring_fini(&adev->gfx.compute_ring[i]); gfx_v7_0_cp_compute_fini(adev); - gfx_v7_0_rlc_fini(adev); + amdgpu_gfx_rlc_fini(adev); gfx_v7_0_mec_fini(adev); amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, &adev->gfx.rlc.clear_state_gpu_addr, @@ -4627,7 +4513,7 @@ static int gfx_v7_0_hw_init(void *handle) gfx_v7_0_constants_init(adev); /* init rlc */ - r = gfx_v7_0_rlc_resume(adev); + r = adev->gfx.rlc.funcs->resume(adev); if (r) return r; @@ -4645,7 +4531,7 @@ static int gfx_v7_0_hw_fini(void *handle) amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0); amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0); gfx_v7_0_cp_enable(adev, false); - gfx_v7_0_rlc_stop(adev); + adev->gfx.rlc.funcs->stop(adev); gfx_v7_0_fini_pg(adev); return 0; @@ -4730,7 +4616,7 @@ static int gfx_v7_0_soft_reset(void *handle) gfx_v7_0_update_cg(adev, false); /* stop the rlc */ - gfx_v7_0_rlc_stop(adev); + adev->gfx.rlc.funcs->stop(adev); /* Disable GFX parsing/prefetching */ WREG32(mmCP_ME_CNTL, CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK); @@ -4959,12 +4845,36 @@ static int gfx_v7_0_eop_irq(struct amdgpu_device *adev, return 0; } +static void gfx_v7_0_fault(struct amdgpu_device *adev, + struct amdgpu_iv_entry *entry) +{ + struct amdgpu_ring *ring; + u8 me_id, pipe_id; + int i; + + me_id = (entry->ring_id & 0x0c) >> 2; + pipe_id = (entry->ring_id & 0x03) >> 0; + switch (me_id) { + case 0: + drm_sched_fault(&adev->gfx.gfx_ring[0].sched); + break; + case 1: + case 2: + for (i = 0; i < adev->gfx.num_compute_rings; i++) { + ring = &adev->gfx.compute_ring[i]; + if ((ring->me == me_id) && (ring->pipe == pipe_id)) + drm_sched_fault(&ring->sched); + } + break; + } +} + static int gfx_v7_0_priv_reg_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { DRM_ERROR("Illegal register access in command stream\n"); - schedule_work(&adev->reset_work); + gfx_v7_0_fault(adev, entry); return 0; } @@ -4974,7 +4884,7 @@ static int gfx_v7_0_priv_inst_irq(struct amdgpu_device *adev, { DRM_ERROR("Illegal instruction in command stream\n"); // XXX soft reset the gfx block only - schedule_work(&adev->reset_work); + gfx_v7_0_fault(adev, entry); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 617b0c8908a3..cb066a8dccd7 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -54,7 +54,7 @@ #include "ivsrcid/ivsrcid_vislands30.h" #define GFX8_NUM_GFX_RINGS 1 -#define GFX8_MEC_HPD_SIZE 2048 +#define GFX8_MEC_HPD_SIZE 4096 #define TOPAZ_GB_ADDR_CONFIG_GOLDEN 0x22010001 #define CARRIZO_GB_ADDR_CONFIG_GOLDEN 0x22010001 @@ -839,18 +839,14 @@ static int gfx_v8_0_ring_test_ring(struct amdgpu_ring *ring) int r; r = amdgpu_gfx_scratch_get(adev, &scratch); - if (r) { - DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r); + if (r) return r; - } + WREG32(scratch, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", - ring->idx, r); - amdgpu_gfx_scratch_free(adev, scratch); - return r; - } + if (r) + goto error_free_scratch; + amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1)); amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START)); amdgpu_ring_write(ring, 0xDEADBEEF); @@ -862,14 +858,11 @@ static int gfx_v8_0_ring_test_ring(struct amdgpu_ring *ring) break; DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n", - ring->idx, scratch, tmp); - r = -EINVAL; - } + + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + +error_free_scratch: amdgpu_gfx_scratch_free(adev, scratch); return r; } @@ -886,19 +879,16 @@ static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 16, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err1; - } + ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3); ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM; ib.ptr[2] = lower_32_bits(gpu_addr); @@ -912,22 +902,17 @@ static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); r = -ETIMEDOUT; goto err2; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto err2; } tmp = adev->wb.wb[index]; - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + if (tmp == 0xDEADBEEF) r = 0; - } else { - DRM_ERROR("ib test on ring %d failed\n", ring->idx); + else r = -EINVAL; - } err2: amdgpu_ib_free(adev, &ib, NULL); @@ -1298,81 +1283,16 @@ static void gfx_v8_0_get_csb_buffer(struct amdgpu_device *adev, buffer[count++] = cpu_to_le32(0); } -static void cz_init_cp_jump_table(struct amdgpu_device *adev) +static int gfx_v8_0_cp_jump_table_num(struct amdgpu_device *adev) { - const __le32 *fw_data; - volatile u32 *dst_ptr; - int me, i, max_me = 4; - u32 bo_offset = 0; - u32 table_offset, table_size; - if (adev->asic_type == CHIP_CARRIZO) - max_me = 5; - - /* write the cp table buffer */ - dst_ptr = adev->gfx.rlc.cp_table_ptr; - for (me = 0; me < max_me; me++) { - if (me == 0) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data; - fw_data = (const __le32 *) - (adev->gfx.ce_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 1) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data; - fw_data = (const __le32 *) - (adev->gfx.pfp_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 2) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data; - fw_data = (const __le32 *) - (adev->gfx.me_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 3) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data; - fw_data = (const __le32 *) - (adev->gfx.mec_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 4) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data; - fw_data = (const __le32 *) - (adev->gfx.mec2_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } - - for (i = 0; i < table_size; i ++) { - dst_ptr[bo_offset + i] = - cpu_to_le32(le32_to_cpu(fw_data[table_offset + i])); - } - - bo_offset += table_size; - } -} - -static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev) -{ - amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL); - amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL); + return 5; + else + return 4; } static int gfx_v8_0_rlc_init(struct amdgpu_device *adev) { - volatile u32 *dst_ptr; - u32 dws; const struct cs_section_def *cs_data; int r; @@ -1381,44 +1301,18 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev) cs_data = adev->gfx.rlc.cs_data; if (cs_data) { - /* clear state block */ - adev->gfx.rlc.clear_state_size = dws = gfx_v8_0_get_csb_size(adev); - - r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->gfx.rlc.clear_state_obj, - &adev->gfx.rlc.clear_state_gpu_addr, - (void **)&adev->gfx.rlc.cs_ptr); - if (r) { - dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r); - gfx_v8_0_rlc_fini(adev); + /* init clear state block */ + r = amdgpu_gfx_rlc_init_csb(adev); + if (r) return r; - } - - /* set up the cs buffer */ - dst_ptr = adev->gfx.rlc.cs_ptr; - gfx_v8_0_get_csb_buffer(adev, dst_ptr); - amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj); - amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj); } if ((adev->asic_type == CHIP_CARRIZO) || (adev->asic_type == CHIP_STONEY)) { adev->gfx.rlc.cp_table_size = ALIGN(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */ - r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size, - PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM, - &adev->gfx.rlc.cp_table_obj, - &adev->gfx.rlc.cp_table_gpu_addr, - (void **)&adev->gfx.rlc.cp_table_ptr); - if (r) { - dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r); + r = amdgpu_gfx_rlc_init_cpt(adev); + if (r) return r; - } - - cz_init_cp_jump_table(adev); - - amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj); - amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj); } return 0; @@ -1443,7 +1337,7 @@ static int gfx_v8_0_mec_init(struct amdgpu_device *adev) mec_hpd_size = adev->gfx.num_compute_rings * GFX8_MEC_HPD_SIZE; r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_GTT, + AMDGPU_GEM_DOMAIN_VRAM, &adev->gfx.mec.hpd_eop_obj, &adev->gfx.mec.hpd_eop_gpu_addr, (void **)&hpd); @@ -1629,7 +1523,7 @@ static int gfx_v8_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) return 0; /* bail if the compute ring is not ready */ - if (!ring->ready) + if (!ring->sched.ready) return 0; tmp = RREG32(mmGB_EDC_MODE); @@ -2088,7 +1982,7 @@ static int gfx_v8_0_sw_init(void *handle) return r; } - r = gfx_v8_0_rlc_init(adev); + r = adev->gfx.rlc.funcs->init(adev); if (r) { DRM_ERROR("Failed to init rlc BOs!\n"); return r; @@ -2181,7 +2075,7 @@ static int gfx_v8_0_sw_fini(void *handle) amdgpu_gfx_kiq_fini(adev); gfx_v8_0_mec_fini(adev); - gfx_v8_0_rlc_fini(adev); + amdgpu_gfx_rlc_fini(adev); amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, &adev->gfx.rlc.clear_state_gpu_addr, (void **)&adev->gfx.rlc.cs_ptr); @@ -4175,10 +4069,10 @@ static void gfx_v8_0_rlc_start(struct amdgpu_device *adev) static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev) { - gfx_v8_0_rlc_stop(adev); - gfx_v8_0_rlc_reset(adev); + adev->gfx.rlc.funcs->stop(adev); + adev->gfx.rlc.funcs->reset(adev); gfx_v8_0_init_pg(adev); - gfx_v8_0_rlc_start(adev); + adev->gfx.rlc.funcs->start(adev); return 0; } @@ -4197,7 +4091,7 @@ static void gfx_v8_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable) tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, 1); tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, 1); for (i = 0; i < adev->gfx.num_gfx_rings; i++) - adev->gfx.gfx_ring[i].ready = false; + adev->gfx.gfx_ring[i].sched.ready = false; } WREG32(mmCP_ME_CNTL, tmp); udelay(50); @@ -4379,10 +4273,8 @@ static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev) /* start the ring */ amdgpu_ring_clear_ring(ring); gfx_v8_0_cp_gfx_start(adev); - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) - ring->ready = false; + ring->sched.ready = true; + r = amdgpu_ring_test_helper(ring); return r; } @@ -4396,8 +4288,8 @@ static void gfx_v8_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) } else { WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK)); for (i = 0; i < adev->gfx.num_compute_rings; i++) - adev->gfx.compute_ring[i].ready = false; - adev->gfx.kiq.ring.ready = false; + adev->gfx.compute_ring[i].sched.ready = false; + adev->gfx.kiq.ring.sched.ready = false; } udelay(50); } @@ -4473,11 +4365,9 @@ static int gfx_v8_0_kiq_kcq_enable(struct amdgpu_device *adev) amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr)); } - r = amdgpu_ring_test_ring(kiq_ring); - if (r) { + r = amdgpu_ring_test_helper(kiq_ring); + if (r) DRM_ERROR("KCQ enable failed\n"); - kiq_ring->ready = false; - } return r; } @@ -4781,7 +4671,7 @@ static int gfx_v8_0_kiq_resume(struct amdgpu_device *adev) amdgpu_bo_kunmap(ring->mqd_obj); ring->mqd_ptr = NULL; amdgpu_bo_unreserve(ring->mqd_obj); - ring->ready = true; + ring->sched.ready = true; return 0; } @@ -4820,10 +4710,7 @@ static int gfx_v8_0_kcq_resume(struct amdgpu_device *adev) */ for (i = adev->gfx.num_compute_rings - 1; i >= 0; i--) { ring = &adev->gfx.compute_ring[i]; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) - ring->ready = false; + r = amdgpu_ring_test_helper(ring); } done: @@ -4867,7 +4754,7 @@ static int gfx_v8_0_hw_init(void *handle) gfx_v8_0_init_golden_registers(adev); gfx_v8_0_constants_init(adev); - r = gfx_v8_0_rlc_resume(adev); + r = adev->gfx.rlc.funcs->resume(adev); if (r) return r; @@ -4899,7 +4786,7 @@ static int gfx_v8_0_kcq_disable(struct amdgpu_device *adev) amdgpu_ring_write(kiq_ring, 0); amdgpu_ring_write(kiq_ring, 0); } - r = amdgpu_ring_test_ring(kiq_ring); + r = amdgpu_ring_test_helper(kiq_ring); if (r) DRM_ERROR("KCQ disable failed\n"); @@ -4973,16 +4860,16 @@ static int gfx_v8_0_hw_fini(void *handle) pr_debug("For SRIOV client, shouldn't do anything.\n"); return 0; } - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); if (!gfx_v8_0_wait_for_idle(adev)) gfx_v8_0_cp_enable(adev, false); else pr_err("cp is busy, skip halt cp\n"); if (!gfx_v8_0_wait_for_rlc_idle(adev)) - gfx_v8_0_rlc_stop(adev); + adev->gfx.rlc.funcs->stop(adev); else pr_err("rlc is busy, skip halt rlc\n"); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return 0; } @@ -5071,7 +4958,7 @@ static int gfx_v8_0_pre_soft_reset(void *handle) srbm_soft_reset = adev->gfx.srbm_soft_reset; /* stop the rlc */ - gfx_v8_0_rlc_stop(adev); + adev->gfx.rlc.funcs->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)) @@ -5197,7 +5084,7 @@ static int gfx_v8_0_post_soft_reset(void *handle) REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX)) gfx_v8_0_cp_gfx_resume(adev); - gfx_v8_0_rlc_start(adev); + adev->gfx.rlc.funcs->start(adev); return 0; } @@ -5445,7 +5332,7 @@ static int gfx_v8_0_set_powergating_state(void *handle, AMD_PG_SUPPORT_RLC_SMU_HS | AMD_PG_SUPPORT_CP | AMD_PG_SUPPORT_GFX_DMG)) - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); switch (adev->asic_type) { case CHIP_CARRIZO: case CHIP_STONEY: @@ -5499,7 +5386,7 @@ static int gfx_v8_0_set_powergating_state(void *handle, AMD_PG_SUPPORT_RLC_SMU_HS | AMD_PG_SUPPORT_CP | AMD_PG_SUPPORT_GFX_DMG)) - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return 0; } @@ -5593,57 +5480,53 @@ static void gfx_v8_0_send_serdes_cmd(struct amdgpu_device *adev, #define RLC_GPR_REG2__MESSAGE__SHIFT 0x00000001 #define RLC_GPR_REG2__MESSAGE_MASK 0x0000001e -static void iceland_enter_rlc_safe_mode(struct amdgpu_device *adev) +static bool gfx_v8_0_is_rlc_enabled(struct amdgpu_device *adev) { - u32 data; - unsigned i; + uint32_t rlc_setting; - data = RREG32(mmRLC_CNTL); - if (!(data & RLC_CNTL__RLC_ENABLE_F32_MASK)) - return; + rlc_setting = RREG32(mmRLC_CNTL); + if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK)) + return false; - if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) { - data |= RLC_SAFE_MODE__CMD_MASK; - data &= ~RLC_SAFE_MODE__MESSAGE_MASK; - data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT); - WREG32(mmRLC_SAFE_MODE, data); + return true; +} - for (i = 0; i < adev->usec_timeout; i++) { - if ((RREG32(mmRLC_GPM_STAT) & - (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK | - RLC_GPM_STAT__GFX_POWER_STATUS_MASK)) == - (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK | - RLC_GPM_STAT__GFX_POWER_STATUS_MASK)) - break; - udelay(1); - } +static void gfx_v8_0_set_safe_mode(struct amdgpu_device *adev) +{ + uint32_t data; + unsigned i; + data = RREG32(mmRLC_CNTL); + data |= RLC_SAFE_MODE__CMD_MASK; + data &= ~RLC_SAFE_MODE__MESSAGE_MASK; + data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT); + WREG32(mmRLC_SAFE_MODE, data); - for (i = 0; i < adev->usec_timeout; i++) { - if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD)) - break; - udelay(1); - } - adev->gfx.rlc.in_safe_mode = true; + /* wait for RLC_SAFE_MODE */ + for (i = 0; i < adev->usec_timeout; i++) { + if ((RREG32(mmRLC_GPM_STAT) & + (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK | + RLC_GPM_STAT__GFX_POWER_STATUS_MASK)) == + (RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK | + RLC_GPM_STAT__GFX_POWER_STATUS_MASK)) + break; + udelay(1); + } + for (i = 0; i < adev->usec_timeout; i++) { + if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD)) + break; + udelay(1); } } -static void iceland_exit_rlc_safe_mode(struct amdgpu_device *adev) +static void gfx_v8_0_unset_safe_mode(struct amdgpu_device *adev) { - u32 data = 0; + uint32_t data; unsigned i; data = RREG32(mmRLC_CNTL); - if (!(data & RLC_CNTL__RLC_ENABLE_F32_MASK)) - return; - - if (adev->cg_flags & (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) { - if (adev->gfx.rlc.in_safe_mode) { - data |= RLC_SAFE_MODE__CMD_MASK; - data &= ~RLC_SAFE_MODE__MESSAGE_MASK; - WREG32(mmRLC_SAFE_MODE, data); - adev->gfx.rlc.in_safe_mode = false; - } - } + data |= RLC_SAFE_MODE__CMD_MASK; + data &= ~RLC_SAFE_MODE__MESSAGE_MASK; + WREG32(mmRLC_SAFE_MODE, data); for (i = 0; i < adev->usec_timeout; i++) { if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD)) @@ -5653,8 +5536,17 @@ static void iceland_exit_rlc_safe_mode(struct amdgpu_device *adev) } static const struct amdgpu_rlc_funcs iceland_rlc_funcs = { - .enter_safe_mode = iceland_enter_rlc_safe_mode, - .exit_safe_mode = iceland_exit_rlc_safe_mode + .is_rlc_enabled = gfx_v8_0_is_rlc_enabled, + .set_safe_mode = gfx_v8_0_set_safe_mode, + .unset_safe_mode = gfx_v8_0_unset_safe_mode, + .init = gfx_v8_0_rlc_init, + .get_csb_size = gfx_v8_0_get_csb_size, + .get_csb_buffer = gfx_v8_0_get_csb_buffer, + .get_cp_table_num = gfx_v8_0_cp_jump_table_num, + .resume = gfx_v8_0_rlc_resume, + .stop = gfx_v8_0_rlc_stop, + .reset = gfx_v8_0_rlc_reset, + .start = gfx_v8_0_rlc_start }; static void gfx_v8_0_update_medium_grain_clock_gating(struct amdgpu_device *adev, @@ -5662,7 +5554,7 @@ static void gfx_v8_0_update_medium_grain_clock_gating(struct amdgpu_device *adev { uint32_t temp, data; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); /* It is disabled by HW by default */ if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) { @@ -5758,7 +5650,7 @@ static void gfx_v8_0_update_medium_grain_clock_gating(struct amdgpu_device *adev gfx_v8_0_wait_for_rlc_serdes(adev); } - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev, @@ -5768,7 +5660,7 @@ static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev temp = data = RREG32(mmRLC_CGCG_CGLS_CTRL); - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) { temp1 = data1 = RREG32(mmRLC_CGTT_MGCG_OVERRIDE); @@ -5851,7 +5743,7 @@ static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev gfx_v8_0_wait_for_rlc_serdes(adev); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } static int gfx_v8_0_update_gfx_clock_gating(struct amdgpu_device *adev, bool enable) @@ -6131,9 +6023,11 @@ static void gfx_v8_0_ring_emit_vgt_flush(struct amdgpu_ring *ring) } static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 header, control = 0; if (ib->flags & AMDGPU_IB_FLAG_CE) @@ -6161,9 +6055,11 @@ static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, } static void gfx_v8_0_ring_emit_ib_compute(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24); amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); @@ -6738,12 +6634,39 @@ static int gfx_v8_0_eop_irq(struct amdgpu_device *adev, return 0; } +static void gfx_v8_0_fault(struct amdgpu_device *adev, + struct amdgpu_iv_entry *entry) +{ + u8 me_id, pipe_id, queue_id; + struct amdgpu_ring *ring; + int i; + + me_id = (entry->ring_id & 0x0c) >> 2; + pipe_id = (entry->ring_id & 0x03) >> 0; + queue_id = (entry->ring_id & 0x70) >> 4; + + switch (me_id) { + case 0: + drm_sched_fault(&adev->gfx.gfx_ring[0].sched); + break; + case 1: + case 2: + for (i = 0; i < adev->gfx.num_compute_rings; i++) { + ring = &adev->gfx.compute_ring[i]; + if (ring->me == me_id && ring->pipe == pipe_id && + ring->queue == queue_id) + drm_sched_fault(&ring->sched); + } + break; + } +} + static int gfx_v8_0_priv_reg_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { DRM_ERROR("Illegal register access in command stream\n"); - schedule_work(&adev->reset_work); + gfx_v8_0_fault(adev, entry); return 0; } @@ -6752,7 +6675,7 @@ static int gfx_v8_0_priv_inst_irq(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { DRM_ERROR("Illegal instruction in command stream\n"); - schedule_work(&adev->reset_work); + gfx_v8_0_fault(adev, entry); return 0; } @@ -6976,10 +6899,8 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_kiq = { 17 + /* gfx_v8_0_ring_emit_vm_flush */ 7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_kiq x3 for user fence, vm fence */ .emit_ib_size = 4, /* gfx_v8_0_ring_emit_ib_compute */ - .emit_ib = gfx_v8_0_ring_emit_ib_compute, .emit_fence = gfx_v8_0_ring_emit_fence_kiq, .test_ring = gfx_v8_0_ring_test_ring, - .test_ib = gfx_v8_0_ring_test_ib, .insert_nop = amdgpu_ring_insert_nop, .pad_ib = amdgpu_ring_generic_pad_ib, .emit_rreg = gfx_v8_0_ring_emit_rreg, diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 6d7baf59d6e1..c27caa144c57 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -41,7 +41,7 @@ #include "ivsrcid/gfx/irqsrcs_gfx_9_0.h" #define GFX9_NUM_GFX_RINGS 1 -#define GFX9_MEC_HPD_SIZE 2048 +#define GFX9_MEC_HPD_SIZE 4096 #define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L #define RLC_SAVE_RESTORE_ADDR_STARTING_OFFSET 0x00000000L @@ -396,18 +396,14 @@ static int gfx_v9_0_ring_test_ring(struct amdgpu_ring *ring) int r; r = amdgpu_gfx_scratch_get(adev, &scratch); - if (r) { - DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r); + if (r) return r; - } + WREG32(scratch, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", - ring->idx, r); - amdgpu_gfx_scratch_free(adev, scratch); - return r; - } + if (r) + goto error_free_scratch; + amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1)); amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START)); amdgpu_ring_write(ring, 0xDEADBEEF); @@ -419,14 +415,11 @@ static int gfx_v9_0_ring_test_ring(struct amdgpu_ring *ring) break; DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n", - ring->idx, scratch, tmp); - r = -EINVAL; - } + + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + +error_free_scratch: amdgpu_gfx_scratch_free(adev, scratch); return r; } @@ -443,19 +436,16 @@ static int gfx_v9_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); adev->wb.wb[index] = cpu_to_le32(0xCAFEDEAD); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 16, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err1; - } + ib.ptr[0] = PACKET3(PACKET3_WRITE_DATA, 3); ib.ptr[1] = WRITE_DATA_DST_SEL(5) | WR_CONFIRM; ib.ptr[2] = lower_32_bits(gpu_addr); @@ -469,22 +459,17 @@ static int gfx_v9_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); - r = -ETIMEDOUT; - goto err2; + r = -ETIMEDOUT; + goto err2; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); - goto err2; + goto err2; } tmp = adev->wb.wb[index]; - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); - r = 0; - } else { - DRM_ERROR("ib test on ring %d failed\n", ring->idx); - r = -EINVAL; - } + if (tmp == 0xDEADBEEF) + r = 0; + else + r = -EINVAL; err2: amdgpu_ib_free(adev, &ib, NULL); @@ -1065,85 +1050,13 @@ static void gfx_v9_0_enable_lbpw(struct amdgpu_device *adev, bool enable) WREG32_FIELD15(GC, 0, RLC_LB_CNTL, LOAD_BALANCE_ENABLE, enable ? 1 : 0); } -static void rv_init_cp_jump_table(struct amdgpu_device *adev) -{ - const __le32 *fw_data; - volatile u32 *dst_ptr; - int me, i, max_me = 5; - u32 bo_offset = 0; - u32 table_offset, table_size; - - /* write the cp table buffer */ - dst_ptr = adev->gfx.rlc.cp_table_ptr; - for (me = 0; me < max_me; me++) { - if (me == 0) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data; - fw_data = (const __le32 *) - (adev->gfx.ce_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 1) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data; - fw_data = (const __le32 *) - (adev->gfx.pfp_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 2) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data; - fw_data = (const __le32 *) - (adev->gfx.me_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 3) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data; - fw_data = (const __le32 *) - (adev->gfx.mec_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } else if (me == 4) { - const struct gfx_firmware_header_v1_0 *hdr = - (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data; - fw_data = (const __le32 *) - (adev->gfx.mec2_fw->data + - le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - table_offset = le32_to_cpu(hdr->jt_offset); - table_size = le32_to_cpu(hdr->jt_size); - } - - for (i = 0; i < table_size; i ++) { - dst_ptr[bo_offset + i] = - cpu_to_le32(le32_to_cpu(fw_data[table_offset + i])); - } - - bo_offset += table_size; - } -} - -static void gfx_v9_0_rlc_fini(struct amdgpu_device *adev) +static int gfx_v9_0_cp_jump_table_num(struct amdgpu_device *adev) { - /* clear state block */ - amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, - &adev->gfx.rlc.clear_state_gpu_addr, - (void **)&adev->gfx.rlc.cs_ptr); - - /* jump table block */ - amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, - &adev->gfx.rlc.cp_table_gpu_addr, - (void **)&adev->gfx.rlc.cp_table_ptr); + return 5; } static int gfx_v9_0_rlc_init(struct amdgpu_device *adev) { - volatile u32 *dst_ptr; - u32 dws; const struct cs_section_def *cs_data; int r; @@ -1152,45 +1065,18 @@ static int gfx_v9_0_rlc_init(struct amdgpu_device *adev) cs_data = adev->gfx.rlc.cs_data; if (cs_data) { - /* clear state block */ - adev->gfx.rlc.clear_state_size = dws = gfx_v9_0_get_csb_size(adev); - r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->gfx.rlc.clear_state_obj, - &adev->gfx.rlc.clear_state_gpu_addr, - (void **)&adev->gfx.rlc.cs_ptr); - if (r) { - dev_err(adev->dev, "(%d) failed to create rlc csb bo\n", - r); - gfx_v9_0_rlc_fini(adev); + /* init clear state block */ + r = amdgpu_gfx_rlc_init_csb(adev); + if (r) return r; - } - /* set up the cs buffer */ - dst_ptr = adev->gfx.rlc.cs_ptr; - gfx_v9_0_get_csb_buffer(adev, dst_ptr); - amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj); - amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj); - amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj); } if (adev->asic_type == CHIP_RAVEN) { /* TODO: double check the cp_table_size for RV */ adev->gfx.rlc.cp_table_size = ALIGN(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */ - r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size, - PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM, - &adev->gfx.rlc.cp_table_obj, - &adev->gfx.rlc.cp_table_gpu_addr, - (void **)&adev->gfx.rlc.cp_table_ptr); - if (r) { - dev_err(adev->dev, - "(%d) failed to create cp table bo\n", r); - gfx_v9_0_rlc_fini(adev); + r = amdgpu_gfx_rlc_init_cpt(adev); + if (r) return r; - } - - rv_init_cp_jump_table(adev); - amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj); - amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj); } switch (adev->asic_type) { @@ -1264,7 +1150,7 @@ static int gfx_v9_0_mec_init(struct amdgpu_device *adev) mec_hpd_size = adev->gfx.num_compute_rings * GFX9_MEC_HPD_SIZE; r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_GTT, + AMDGPU_GEM_DOMAIN_VRAM, &adev->gfx.mec.hpd_eop_obj, &adev->gfx.mec.hpd_eop_gpu_addr, (void **)&hpd); @@ -1635,8 +1521,8 @@ static int gfx_v9_0_ngg_en(struct amdgpu_device *adev) /* Clear GDS reserved memory */ r = amdgpu_ring_alloc(ring, 17); if (r) { - DRM_ERROR("amdgpu: NGG failed to lock ring %d (%d).\n", - ring->idx, r); + DRM_ERROR("amdgpu: NGG failed to lock ring %s (%d).\n", + ring->name, r); return r; } @@ -1748,7 +1634,7 @@ static int gfx_v9_0_sw_init(void *handle) return r; } - r = gfx_v9_0_rlc_init(adev); + r = adev->gfx.rlc.funcs->init(adev); if (r) { DRM_ERROR("Failed to init rlc BOs!\n"); return r; @@ -2498,12 +2384,12 @@ static int gfx_v9_0_rlc_resume(struct amdgpu_device *adev) return 0; } - gfx_v9_0_rlc_stop(adev); + adev->gfx.rlc.funcs->stop(adev); /* disable CG */ WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, 0); - gfx_v9_0_rlc_reset(adev); + adev->gfx.rlc.funcs->reset(adev); gfx_v9_0_init_pg(adev); @@ -2514,15 +2400,24 @@ static int gfx_v9_0_rlc_resume(struct amdgpu_device *adev) return r; } - if (adev->asic_type == CHIP_RAVEN || - adev->asic_type == CHIP_VEGA20) { - if (amdgpu_lbpw != 0) + switch (adev->asic_type) { + case CHIP_RAVEN: + if (amdgpu_lbpw == 0) + gfx_v9_0_enable_lbpw(adev, false); + else + gfx_v9_0_enable_lbpw(adev, true); + break; + case CHIP_VEGA20: + if (amdgpu_lbpw > 0) gfx_v9_0_enable_lbpw(adev, true); else gfx_v9_0_enable_lbpw(adev, false); + break; + default: + break; } - gfx_v9_0_rlc_start(adev); + adev->gfx.rlc.funcs->start(adev); return 0; } @@ -2537,7 +2432,7 @@ static void gfx_v9_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable) tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, enable ? 0 : 1); if (!enable) { for (i = 0; i < adev->gfx.num_gfx_rings; i++) - adev->gfx.gfx_ring[i].ready = false; + adev->gfx.gfx_ring[i].sched.ready = false; } WREG32_SOC15(GC, 0, mmCP_ME_CNTL, tmp); udelay(50); @@ -2727,7 +2622,7 @@ static int gfx_v9_0_cp_gfx_resume(struct amdgpu_device *adev) /* start the ring */ gfx_v9_0_cp_gfx_start(adev); - ring->ready = true; + ring->sched.ready = true; return 0; } @@ -2742,8 +2637,8 @@ static void gfx_v9_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) WREG32_SOC15(GC, 0, mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK)); for (i = 0; i < adev->gfx.num_compute_rings; i++) - adev->gfx.compute_ring[i].ready = false; - adev->gfx.kiq.ring.ready = false; + adev->gfx.compute_ring[i].sched.ready = false; + adev->gfx.kiq.ring.sched.ready = false; } udelay(50); } @@ -2866,11 +2761,9 @@ static int gfx_v9_0_kiq_kcq_enable(struct amdgpu_device *adev) amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr)); } - r = amdgpu_ring_test_ring(kiq_ring); - if (r) { + r = amdgpu_ring_test_helper(kiq_ring); + if (r) DRM_ERROR("KCQ enable failed\n"); - kiq_ring->ready = false; - } return r; } @@ -3249,7 +3142,7 @@ static int gfx_v9_0_kiq_resume(struct amdgpu_device *adev) amdgpu_bo_kunmap(ring->mqd_obj); ring->mqd_ptr = NULL; amdgpu_bo_unreserve(ring->mqd_obj); - ring->ready = true; + ring->sched.ready = true; return 0; } @@ -3314,19 +3207,13 @@ static int gfx_v9_0_cp_resume(struct amdgpu_device *adev) return r; ring = &adev->gfx.gfx_ring[0]; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) return r; - } for (i = 0; i < adev->gfx.num_compute_rings; i++) { ring = &adev->gfx.compute_ring[i]; - - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) - ring->ready = false; + amdgpu_ring_test_helper(ring); } gfx_v9_0_enable_gui_idle_interrupt(adev, true); @@ -3353,7 +3240,7 @@ static int gfx_v9_0_hw_init(void *handle) if (r) return r; - r = gfx_v9_0_rlc_resume(adev); + r = adev->gfx.rlc.funcs->resume(adev); if (r) return r; @@ -3391,7 +3278,7 @@ static int gfx_v9_0_kcq_disable(struct amdgpu_device *adev) amdgpu_ring_write(kiq_ring, 0); amdgpu_ring_write(kiq_ring, 0); } - r = amdgpu_ring_test_ring(kiq_ring); + r = amdgpu_ring_test_helper(kiq_ring); if (r) DRM_ERROR("KCQ disable failed\n"); @@ -3433,7 +3320,7 @@ static int gfx_v9_0_hw_fini(void *handle) } gfx_v9_0_cp_enable(adev, false); - gfx_v9_0_rlc_stop(adev); + adev->gfx.rlc.funcs->stop(adev); gfx_v9_0_csb_vram_unpin(adev); @@ -3508,7 +3395,7 @@ static int gfx_v9_0_soft_reset(void *handle) if (grbm_soft_reset) { /* stop the rlc */ - gfx_v9_0_rlc_stop(adev); + adev->gfx.rlc.funcs->stop(adev); /* Disable GFX parsing/prefetching */ gfx_v9_0_cp_gfx_enable(adev, false); @@ -3607,64 +3494,47 @@ static int gfx_v9_0_late_init(void *handle) return 0; } -static void gfx_v9_0_enter_rlc_safe_mode(struct amdgpu_device *adev) +static bool gfx_v9_0_is_rlc_enabled(struct amdgpu_device *adev) { - uint32_t rlc_setting, data; - unsigned i; - - if (adev->gfx.rlc.in_safe_mode) - return; + uint32_t rlc_setting; /* if RLC is not enabled, do nothing */ rlc_setting = RREG32_SOC15(GC, 0, mmRLC_CNTL); if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK)) - return; - - if (adev->cg_flags & - (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG | - AMD_CG_SUPPORT_GFX_3D_CGCG)) { - data = RLC_SAFE_MODE__CMD_MASK; - data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT); - WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data); + return false; - /* wait for RLC_SAFE_MODE */ - for (i = 0; i < adev->usec_timeout; i++) { - if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD)) - break; - udelay(1); - } - adev->gfx.rlc.in_safe_mode = true; - } + return true; } -static void gfx_v9_0_exit_rlc_safe_mode(struct amdgpu_device *adev) +static void gfx_v9_0_set_safe_mode(struct amdgpu_device *adev) { - uint32_t rlc_setting, data; - - if (!adev->gfx.rlc.in_safe_mode) - return; + uint32_t data; + unsigned i; - /* if RLC is not enabled, do nothing */ - rlc_setting = RREG32_SOC15(GC, 0, mmRLC_CNTL); - if (!(rlc_setting & RLC_CNTL__RLC_ENABLE_F32_MASK)) - return; + data = RLC_SAFE_MODE__CMD_MASK; + data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT); + WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data); - if (adev->cg_flags & - (AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_MGCG)) { - /* - * Try to exit safe mode only if it is already in safe - * mode. - */ - data = RLC_SAFE_MODE__CMD_MASK; - WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data); - adev->gfx.rlc.in_safe_mode = false; + /* wait for RLC_SAFE_MODE */ + for (i = 0; i < adev->usec_timeout; i++) { + if (!REG_GET_FIELD(RREG32_SOC15(GC, 0, mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD)) + break; + udelay(1); } } +static void gfx_v9_0_unset_safe_mode(struct amdgpu_device *adev) +{ + uint32_t data; + + data = RLC_SAFE_MODE__CMD_MASK; + WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE, data); +} + static void gfx_v9_0_update_gfx_cg_power_gating(struct amdgpu_device *adev, bool enable) { - gfx_v9_0_enter_rlc_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) && enable) { gfx_v9_0_enable_gfx_cg_power_gating(adev, true); @@ -3675,7 +3545,7 @@ static void gfx_v9_0_update_gfx_cg_power_gating(struct amdgpu_device *adev, gfx_v9_0_enable_gfx_pipeline_powergating(adev, false); } - gfx_v9_0_exit_rlc_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } static void gfx_v9_0_update_gfx_mg_power_gating(struct amdgpu_device *adev, @@ -3773,7 +3643,7 @@ static void gfx_v9_0_update_3d_clock_gating(struct amdgpu_device *adev, { uint32_t data, def; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); /* Enable 3D CGCG/CGLS */ if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_3D_CGCG)) { @@ -3813,7 +3683,7 @@ static void gfx_v9_0_update_3d_clock_gating(struct amdgpu_device *adev, WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D, data); } - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev, @@ -3821,7 +3691,7 @@ static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev { uint32_t def, data; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_CGCG)) { def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE); @@ -3861,7 +3731,7 @@ static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev WREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL, data); } - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } static int gfx_v9_0_update_gfx_clock_gating(struct amdgpu_device *adev, @@ -3890,8 +3760,17 @@ static int gfx_v9_0_update_gfx_clock_gating(struct amdgpu_device *adev, } static const struct amdgpu_rlc_funcs gfx_v9_0_rlc_funcs = { - .enter_safe_mode = gfx_v9_0_enter_rlc_safe_mode, - .exit_safe_mode = gfx_v9_0_exit_rlc_safe_mode + .is_rlc_enabled = gfx_v9_0_is_rlc_enabled, + .set_safe_mode = gfx_v9_0_set_safe_mode, + .unset_safe_mode = gfx_v9_0_unset_safe_mode, + .init = gfx_v9_0_rlc_init, + .get_csb_size = gfx_v9_0_get_csb_size, + .get_csb_buffer = gfx_v9_0_get_csb_buffer, + .get_cp_table_num = gfx_v9_0_cp_jump_table_num, + .resume = gfx_v9_0_rlc_resume, + .stop = gfx_v9_0_rlc_stop, + .reset = gfx_v9_0_rlc_reset, + .start = gfx_v9_0_rlc_start }; static int gfx_v9_0_set_powergating_state(void *handle, @@ -4072,9 +3951,11 @@ static void gfx_v9_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) } static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 header, control = 0; if (ib->flags & AMDGPU_IB_FLAG_CE) @@ -4103,20 +3984,22 @@ static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, } static void gfx_v9_0_ring_emit_ib_compute(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { - u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24); + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24); - amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); + amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); BUG_ON(ib->gpu_addr & 0x3); /* Dword align */ - amdgpu_ring_write(ring, + amdgpu_ring_write(ring, #ifdef __BIG_ENDIAN - (2 << 0) | + (2 << 0) | #endif - lower_32_bits(ib->gpu_addr)); - amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr)); - amdgpu_ring_write(ring, control); + lower_32_bits(ib->gpu_addr)); + amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr)); + amdgpu_ring_write(ring, control); } static void gfx_v9_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, @@ -4695,12 +4578,39 @@ static int gfx_v9_0_eop_irq(struct amdgpu_device *adev, return 0; } +static void gfx_v9_0_fault(struct amdgpu_device *adev, + struct amdgpu_iv_entry *entry) +{ + u8 me_id, pipe_id, queue_id; + struct amdgpu_ring *ring; + int i; + + me_id = (entry->ring_id & 0x0c) >> 2; + pipe_id = (entry->ring_id & 0x03) >> 0; + queue_id = (entry->ring_id & 0x70) >> 4; + + switch (me_id) { + case 0: + drm_sched_fault(&adev->gfx.gfx_ring[0].sched); + break; + case 1: + case 2: + for (i = 0; i < adev->gfx.num_compute_rings; i++) { + ring = &adev->gfx.compute_ring[i]; + if (ring->me == me_id && ring->pipe == pipe_id && + ring->queue == queue_id) + drm_sched_fault(&ring->sched); + } + break; + } +} + static int gfx_v9_0_priv_reg_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { DRM_ERROR("Illegal register access in command stream\n"); - schedule_work(&adev->reset_work); + gfx_v9_0_fault(adev, entry); return 0; } @@ -4709,7 +4619,7 @@ static int gfx_v9_0_priv_inst_irq(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { DRM_ERROR("Illegal instruction in command stream\n"); - schedule_work(&adev->reset_work); + gfx_v9_0_fault(adev, entry); return 0; } @@ -4836,10 +4746,8 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = { 2 + /* gfx_v9_0_ring_emit_vm_flush */ 8 + 8 + 8, /* gfx_v9_0_ring_emit_fence_kiq x3 for user fence, vm fence */ .emit_ib_size = 4, /* gfx_v9_0_ring_emit_ib_compute */ - .emit_ib = gfx_v9_0_ring_emit_ib_compute, .emit_fence = gfx_v9_0_ring_emit_fence_kiq, .test_ring = gfx_v9_0_ring_test_ring, - .test_ib = gfx_v9_0_ring_test_ib, .insert_nop = amdgpu_ring_insert_nop, .pad_ib = amdgpu_ring_generic_pad_ib, .emit_rreg = gfx_v9_0_ring_emit_rreg, diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c index bfa317ad20a9..f5edddf3b29d 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c @@ -35,20 +35,25 @@ u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev) return (u64)RREG32_SOC15(GC, 0, mmMC_VM_FB_OFFSET) << 24; } -static void gfxhub_v1_0_init_gart_pt_regs(struct amdgpu_device *adev) +void gfxhub_v1_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, + uint64_t page_table_base) { - uint64_t value = amdgpu_gmc_pd_addr(adev->gart.bo); + /* two registers distance between mmVM_CONTEXT0_* to mmVM_CONTEXT1_* */ + int offset = mmVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 + - mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32; - WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, - lower_32_bits(value)); + WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, + offset * vmid, lower_32_bits(page_table_base)); - WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, - upper_32_bits(value)); + WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, + offset * vmid, upper_32_bits(page_table_base)); } static void gfxhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev) { - gfxhub_v1_0_init_gart_pt_regs(adev); + uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo); + + gfxhub_v1_0_setup_vm_pt_regs(adev, 0, pt_base); WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, (u32)(adev->gmc.gart_start >> 12)); diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h index 206e29cad753..92d3a70cd9b1 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h @@ -30,5 +30,7 @@ void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev, bool value); void gfxhub_v1_0_init(struct amdgpu_device *adev); u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev); +void gfxhub_v1_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, + uint64_t page_table_base); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c index e1c2b4e9c7b2..2821d1d846e4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c @@ -358,7 +358,8 @@ static int gmc_v6_0_mc_init(struct amdgpu_device *adev) return 0; } -static void gmc_v6_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid) +static void gmc_v6_0_flush_gpu_tlb(struct amdgpu_device *adev, + uint32_t vmid, uint32_t flush_type) { WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid); } @@ -580,7 +581,7 @@ static int gmc_v6_0_gart_enable(struct amdgpu_device *adev) else gmc_v6_0_set_fault_enable_default(adev, true); - gmc_v6_0_flush_gpu_tlb(adev, 0); + gmc_v6_0_flush_gpu_tlb(adev, 0, 0); dev_info(adev->dev, "PCIE GART of %uM enabled (table at 0x%016llX).\n", (unsigned)(adev->gmc.gart_size >> 20), (unsigned long long)table_addr); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c index 910c4ce19cb3..761dcfb2fec0 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c @@ -430,7 +430,8 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev) * * Flush the TLB for the requested page table (CIK). */ -static void gmc_v7_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid) +static void gmc_v7_0_flush_gpu_tlb(struct amdgpu_device *adev, + uint32_t vmid, uint32_t flush_type) { /* bits 0-15 are the VM contexts0-15 */ WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid); @@ -698,7 +699,7 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev) WREG32(mmCHUB_CONTROL, tmp); } - gmc_v7_0_flush_gpu_tlb(adev, 0); + gmc_v7_0_flush_gpu_tlb(adev, 0, 0); DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n", (unsigned)(adev->gmc.gart_size >> 20), (unsigned long long)table_addr); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index 1d3265c97b70..531aaf377592 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -611,7 +611,7 @@ static int gmc_v8_0_mc_init(struct amdgpu_device *adev) * Flush the TLB for the requested page table (CIK). */ static void gmc_v8_0_flush_gpu_tlb(struct amdgpu_device *adev, - uint32_t vmid) + uint32_t vmid, uint32_t flush_type) { /* bits 0-15 are the VM contexts0-15 */ WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid); @@ -920,7 +920,7 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev) else gmc_v8_0_set_fault_enable_default(adev, true); - gmc_v8_0_flush_gpu_tlb(adev, 0); + gmc_v8_0_flush_gpu_tlb(adev, 0, 0); DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n", (unsigned)(adev->gmc.gart_size >> 20), (unsigned long long)table_addr); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index f35d7a554ad5..811231e4ec53 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -293,14 +293,14 @@ static void gmc_v9_0_set_irq_funcs(struct amdgpu_device *adev) adev->gmc.vm_fault.funcs = &gmc_v9_0_irq_funcs; } -static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vmid) +static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vmid, + uint32_t flush_type) { u32 req = 0; - /* invalidate using legacy mode on vmid*/ req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, PER_VMID_INVALIDATE_REQ, 1 << vmid); - req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, 0); + req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type); req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1); req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1); req = REG_SET_FIELD(req, VM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1); @@ -312,48 +312,6 @@ static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vmid) return req; } -static signed long amdgpu_kiq_reg_write_reg_wait(struct amdgpu_device *adev, - uint32_t reg0, uint32_t reg1, - uint32_t ref, uint32_t mask) -{ - signed long r, cnt = 0; - unsigned long flags; - uint32_t seq; - struct amdgpu_kiq *kiq = &adev->gfx.kiq; - struct amdgpu_ring *ring = &kiq->ring; - - spin_lock_irqsave(&kiq->ring_lock, flags); - - amdgpu_ring_alloc(ring, 32); - amdgpu_ring_emit_reg_write_reg_wait(ring, reg0, reg1, - ref, mask); - amdgpu_fence_emit_polling(ring, &seq); - amdgpu_ring_commit(ring); - spin_unlock_irqrestore(&kiq->ring_lock, flags); - - r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); - - /* don't wait anymore for IRQ context */ - if (r < 1 && in_interrupt()) - goto failed_kiq; - - might_sleep(); - - while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) { - msleep(MAX_KIQ_REG_BAILOUT_INTERVAL); - r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); - } - - if (cnt > MAX_KIQ_REG_TRY) - goto failed_kiq; - - return 0; - -failed_kiq: - pr_err("failed to invalidate tlb with kiq\n"); - return r; -} - /* * GART * VMID 0 is the physical GPU addresses as used by the kernel. @@ -362,64 +320,47 @@ failed_kiq: */ /** - * gmc_v9_0_flush_gpu_tlb - gart tlb flush callback + * gmc_v9_0_flush_gpu_tlb - tlb flush with certain type * * @adev: amdgpu_device pointer * @vmid: vm instance to flush + * @flush_type: the flush type * - * Flush the TLB for the requested page table. + * Flush the TLB for the requested page table using certain type. */ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, - uint32_t vmid) + uint32_t vmid, uint32_t flush_type) { - /* Use register 17 for GART */ const unsigned eng = 17; unsigned i, j; - int r; for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) { struct amdgpu_vmhub *hub = &adev->vmhub[i]; - u32 tmp = gmc_v9_0_get_invalidate_req(vmid); - - if (adev->gfx.kiq.ring.ready && - (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev)) && - !adev->in_gpu_reset) { - r = amdgpu_kiq_reg_write_reg_wait(adev, hub->vm_inv_eng0_req + eng, - hub->vm_inv_eng0_ack + eng, tmp, 1 << vmid); - if (!r) - continue; - } + u32 tmp = gmc_v9_0_get_invalidate_req(vmid, flush_type); - spin_lock(&adev->gmc.invalidate_lock); + if (i == AMDGPU_GFXHUB && !adev->in_gpu_reset && + adev->gfx.kiq.ring.sched.ready && + (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) { + uint32_t req = hub->vm_inv_eng0_req + eng; + uint32_t ack = hub->vm_inv_eng0_ack + eng; - WREG32_NO_KIQ(hub->vm_inv_eng0_req + eng, tmp); - - /* Busy wait for ACK.*/ - for (j = 0; j < 100; j++) { - tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng); - tmp &= 1 << vmid; - if (tmp) - break; - cpu_relax(); - } - if (j < 100) { - spin_unlock(&adev->gmc.invalidate_lock); + amdgpu_virt_kiq_reg_write_reg_wait(adev, req, ack, tmp, + 1 << vmid); continue; } - /* Wait for ACK with a delay.*/ + spin_lock(&adev->gmc.invalidate_lock); + WREG32_NO_KIQ(hub->vm_inv_eng0_req + eng, tmp); for (j = 0; j < adev->usec_timeout; j++) { tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng); - tmp &= 1 << vmid; - if (tmp) + if (tmp & (1 << vmid)) break; udelay(1); } - if (j < adev->usec_timeout) { - spin_unlock(&adev->gmc.invalidate_lock); - continue; - } spin_unlock(&adev->gmc.invalidate_lock); + if (j < adev->usec_timeout) + continue; + DRM_ERROR("Timeout waiting for VM flush ACK!\n"); } } @@ -429,7 +370,7 @@ static uint64_t gmc_v9_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring, { struct amdgpu_device *adev = ring->adev; struct amdgpu_vmhub *hub = &adev->vmhub[ring->funcs->vmhub]; - uint32_t req = gmc_v9_0_get_invalidate_req(vmid); + uint32_t req = gmc_v9_0_get_invalidate_req(vmid, 0); unsigned eng = ring->vm_inv_eng; amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 + (2 * vmid), @@ -739,9 +680,8 @@ static int gmc_v9_0_late_init(void *handle) unsigned vmhub = ring->funcs->vmhub; ring->vm_inv_eng = vm_inv_eng[vmhub]++; - dev_info(adev->dev, "ring %u(%s) uses VM inv eng %u on hub %u\n", - ring->idx, ring->name, ring->vm_inv_eng, - ring->funcs->vmhub); + dev_info(adev->dev, "ring %s uses VM inv eng %u on hub %u\n", + ring->name, ring->vm_inv_eng, ring->funcs->vmhub); } /* Engine 16 is used for KFD and 17 for GART flushes */ @@ -1122,7 +1062,7 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev) gfxhub_v1_0_set_fault_enable_default(adev, value); mmhub_v1_0_set_fault_enable_default(adev, value); - gmc_v9_0_flush_gpu_tlb(adev, 0); + gmc_v9_0_flush_gpu_tlb(adev, 0, 0); DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n", (unsigned)(adev->gmc.gart_size >> 20), diff --git a/drivers/gpu/drm/amd/amdgpu/kv_dpm.c b/drivers/gpu/drm/amd/amdgpu/kv_dpm.c index d0e478f43443..0c9a2c03504e 100644 --- a/drivers/gpu/drm/amd/amdgpu/kv_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/kv_dpm.c @@ -508,19 +508,19 @@ static int kv_enable_didt(struct amdgpu_device *adev, bool enable) pi->caps_db_ramping || pi->caps_td_ramping || pi->caps_tcp_ramping) { - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); if (enable) { ret = kv_program_pt_config_registers(adev, didt_config_kv); if (ret) { - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return ret; } } kv_do_enable_didt(adev, enable); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c index a0db67adc34c..d0d966d6080a 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c @@ -52,20 +52,25 @@ u64 mmhub_v1_0_get_fb_location(struct amdgpu_device *adev) return base; } -static void mmhub_v1_0_init_gart_pt_regs(struct amdgpu_device *adev) +void mmhub_v1_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, + uint64_t page_table_base) { - uint64_t value = amdgpu_gmc_pd_addr(adev->gart.bo); + /* two registers distance between mmVM_CONTEXT0_* to mmVM_CONTEXT1_* */ + int offset = mmVM_CONTEXT1_PAGE_TABLE_BASE_ADDR_LO32 + - mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32; - WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, - lower_32_bits(value)); + WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32, + offset * vmid, lower_32_bits(page_table_base)); - WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, - upper_32_bits(value)); + WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32, + offset * vmid, upper_32_bits(page_table_base)); } static void mmhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev) { - mmhub_v1_0_init_gart_pt_regs(adev); + uint64_t pt_base = amdgpu_gmc_pd_addr(adev->gart.bo); + + mmhub_v1_0_setup_vm_pt_regs(adev, 0, pt_base); WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, (u32)(adev->gmc.gart_start >> 12)); diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h index bef3d0c0c117..0de0fdf98c00 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h @@ -34,5 +34,7 @@ int mmhub_v1_0_set_clockgating(struct amdgpu_device *adev, void mmhub_v1_0_get_clockgating(struct amdgpu_device *adev, u32 *flags); void mmhub_v1_0_update_power_gating(struct amdgpu_device *adev, bool enable); +void mmhub_v1_0_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmid, + uint64_t page_table_base); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index 3f3fac2d50cd..e5dd052d9e06 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -34,6 +34,7 @@ #include "nbio/nbio_7_4_offset.h" MODULE_FIRMWARE("amdgpu/vega20_sos.bin"); +MODULE_FIRMWARE("amdgpu/vega20_ta.bin"); /* address block */ #define smnMP1_FIRMWARE_FLAGS 0x3010024 @@ -98,7 +99,8 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) const char *chip_name; char fw_name[30]; int err = 0; - const struct psp_firmware_header_v1_0 *hdr; + const struct psp_firmware_header_v1_0 *sos_hdr; + const struct ta_firmware_header_v1_0 *ta_hdr; DRM_DEBUG("\n"); @@ -119,16 +121,32 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) if (err) goto out; - hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.sos_fw->data; - adev->psp.sos_fw_version = le32_to_cpu(hdr->header.ucode_version); - adev->psp.sos_feature_version = le32_to_cpu(hdr->ucode_feature_version); - adev->psp.sos_bin_size = le32_to_cpu(hdr->sos_size_bytes); - adev->psp.sys_bin_size = le32_to_cpu(hdr->header.ucode_size_bytes) - - le32_to_cpu(hdr->sos_size_bytes); - adev->psp.sys_start_addr = (uint8_t *)hdr + - le32_to_cpu(hdr->header.ucode_array_offset_bytes); + sos_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.sos_fw->data; + adev->psp.sos_fw_version = le32_to_cpu(sos_hdr->header.ucode_version); + adev->psp.sos_feature_version = le32_to_cpu(sos_hdr->ucode_feature_version); + adev->psp.sos_bin_size = le32_to_cpu(sos_hdr->sos_size_bytes); + adev->psp.sys_bin_size = le32_to_cpu(sos_hdr->header.ucode_size_bytes) - + le32_to_cpu(sos_hdr->sos_size_bytes); + adev->psp.sys_start_addr = (uint8_t *)sos_hdr + + le32_to_cpu(sos_hdr->header.ucode_array_offset_bytes); adev->psp.sos_start_addr = (uint8_t *)adev->psp.sys_start_addr + - le32_to_cpu(hdr->sos_offset_bytes); + le32_to_cpu(sos_hdr->sos_offset_bytes); + + snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ta.bin", chip_name); + err = request_firmware(&adev->psp.ta_fw, fw_name, adev->dev); + if (err) + goto out; + + err = amdgpu_ucode_validate(adev->psp.ta_fw); + if (err) + goto out; + + ta_hdr = (const struct ta_firmware_header_v1_0 *)adev->psp.ta_fw->data; + adev->psp.ta_xgmi_ucode_version = le32_to_cpu(ta_hdr->ta_xgmi_ucode_version); + adev->psp.ta_xgmi_ucode_size = le32_to_cpu(ta_hdr->ta_xgmi_size_bytes); + adev->psp.ta_xgmi_start_addr = (uint8_t *)ta_hdr + + le32_to_cpu(ta_hdr->header.ucode_array_offset_bytes); + return 0; out: if (err) { @@ -167,7 +185,7 @@ static int psp_v11_0_bootloader_load_sysdrv(struct psp_context *psp) /* Copy PSP System Driver binary to memory */ memcpy(psp->fw_pri_buf, psp->sys_start_addr, psp->sys_bin_size); - /* Provide the sys driver to bootrom */ + /* Provide the sys driver to bootloader */ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, (uint32_t)(psp->fw_pri_mc_addr >> 20)); psp_gfxdrv_command_reg = 1 << 16; @@ -208,7 +226,7 @@ static int psp_v11_0_bootloader_load_sos(struct psp_context *psp) /* Copy Secure OS binary to PSP memory */ memcpy(psp->fw_pri_buf, psp->sos_start_addr, psp->sos_bin_size); - /* Provide the PSP secure OS to bootrom */ + /* Provide the PSP secure OS to bootloader */ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, (uint32_t)(psp->fw_pri_mc_addr >> 20)); psp_gfxdrv_command_reg = 2 << 16; @@ -552,24 +570,110 @@ static int psp_v11_0_mode1_reset(struct psp_context *psp) static int psp_v11_0_xgmi_get_topology_info(struct psp_context *psp, int number_devices, struct psp_xgmi_topology_info *topology) { + struct ta_xgmi_shared_memory *xgmi_cmd; + struct ta_xgmi_cmd_get_topology_info_input *topology_info_input; + struct ta_xgmi_cmd_get_topology_info_output *topology_info_output; + int i; + int ret; + + if (!topology || topology->num_nodes > TA_XGMI__MAX_CONNECTED_NODES) + return -EINVAL; + + xgmi_cmd = (struct ta_xgmi_shared_memory*)psp->xgmi_context.xgmi_shared_buf; + memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); + + /* Fill in the shared memory with topology information as input */ + topology_info_input = &xgmi_cmd->xgmi_in_message.get_topology_info; + xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_GET_TOPOLOGY_INFO; + topology_info_input->num_nodes = number_devices; + + for (i = 0; i < topology_info_input->num_nodes; i++) { + topology_info_input->nodes[i].node_id = topology->nodes[i].node_id; + topology_info_input->nodes[i].num_hops = topology->nodes[i].num_hops; + topology_info_input->nodes[i].is_sharing_enabled = topology->nodes[i].is_sharing_enabled; + topology_info_input->nodes[i].sdma_engine = topology->nodes[i].sdma_engine; + } + + /* Invoke xgmi ta to get the topology information */ + ret = psp_xgmi_invoke(psp, TA_COMMAND_XGMI__GET_GET_TOPOLOGY_INFO); + if (ret) + return ret; + + /* Read the output topology information from the shared memory */ + topology_info_output = &xgmi_cmd->xgmi_out_message.get_topology_info; + topology->num_nodes = xgmi_cmd->xgmi_out_message.get_topology_info.num_nodes; + for (i = 0; i < topology->num_nodes; i++) { + topology->nodes[i].node_id = topology_info_output->nodes[i].node_id; + topology->nodes[i].num_hops = topology_info_output->nodes[i].num_hops; + topology->nodes[i].is_sharing_enabled = topology_info_output->nodes[i].is_sharing_enabled; + topology->nodes[i].sdma_engine = topology_info_output->nodes[i].sdma_engine; + } + return 0; } static int psp_v11_0_xgmi_set_topology_info(struct psp_context *psp, int number_devices, struct psp_xgmi_topology_info *topology) { - return 0; + struct ta_xgmi_shared_memory *xgmi_cmd; + struct ta_xgmi_cmd_get_topology_info_input *topology_info_input; + int i; + + if (!topology || topology->num_nodes > TA_XGMI__MAX_CONNECTED_NODES) + return -EINVAL; + + xgmi_cmd = (struct ta_xgmi_shared_memory*)psp->xgmi_context.xgmi_shared_buf; + memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); + + topology_info_input = &xgmi_cmd->xgmi_in_message.get_topology_info; + xgmi_cmd->cmd_id = TA_COMMAND_XGMI__SET_TOPOLOGY_INFO; + topology_info_input->num_nodes = number_devices; + + for (i = 0; i < topology_info_input->num_nodes; i++) { + topology_info_input->nodes[i].node_id = topology->nodes[i].node_id; + topology_info_input->nodes[i].num_hops = topology->nodes[i].num_hops; + topology_info_input->nodes[i].is_sharing_enabled = topology->nodes[i].is_sharing_enabled; + topology_info_input->nodes[i].sdma_engine = topology->nodes[i].sdma_engine; + } + + /* Invoke xgmi ta to set topology information */ + return psp_xgmi_invoke(psp, TA_COMMAND_XGMI__SET_TOPOLOGY_INFO); } static u64 psp_v11_0_xgmi_get_hive_id(struct psp_context *psp) { - u64 hive_id = 0; + struct ta_xgmi_shared_memory *xgmi_cmd; + int ret; + + xgmi_cmd = (struct ta_xgmi_shared_memory*)psp->xgmi_context.xgmi_shared_buf; + memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); + + xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_HIVE_ID; + + /* Invoke xgmi ta to get hive id */ + ret = psp_xgmi_invoke(psp, xgmi_cmd->cmd_id); + if (ret) + return 0; + else + return xgmi_cmd->xgmi_out_message.get_hive_id.hive_id; +} + +static u64 psp_v11_0_xgmi_get_node_id(struct psp_context *psp) +{ + struct ta_xgmi_shared_memory *xgmi_cmd; + int ret; + + xgmi_cmd = (struct ta_xgmi_shared_memory*)psp->xgmi_context.xgmi_shared_buf; + memset(xgmi_cmd, 0, sizeof(struct ta_xgmi_shared_memory)); - /* Remove me when we can get correct hive_id through PSP */ - if (psp->adev->gmc.xgmi.num_physical_nodes) - hive_id = 0x123456789abcdef; + xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_NODE_ID; - return hive_id; + /* Invoke xgmi ta to get the node id */ + ret = psp_xgmi_invoke(psp, xgmi_cmd->cmd_id); + if (ret) + return 0; + else + return xgmi_cmd->xgmi_out_message.get_node_id.node_id; } static const struct psp_funcs psp_v11_0_funcs = { @@ -587,6 +691,7 @@ static const struct psp_funcs psp_v11_0_funcs = { .xgmi_get_topology_info = psp_v11_0_xgmi_get_topology_info, .xgmi_set_topology_info = psp_v11_0_xgmi_set_topology_info, .xgmi_get_hive_id = psp_v11_0_xgmi_get_hive_id, + .xgmi_get_node_id = psp_v11_0_xgmi_get_node_id, }; void psp_v11_0_set_psp_funcs(struct psp_context *psp) diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c index e1ebf770c303..9cea0bbe4525 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c @@ -194,7 +194,7 @@ static int psp_v3_1_bootloader_load_sysdrv(struct psp_context *psp) /* Copy PSP System Driver binary to memory */ memcpy(psp->fw_pri_buf, psp->sys_start_addr, psp->sys_bin_size); - /* Provide the sys driver to bootrom */ + /* Provide the sys driver to bootloader */ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, (uint32_t)(psp->fw_pri_mc_addr >> 20)); psp_gfxdrv_command_reg = 1 << 16; @@ -254,7 +254,7 @@ static int psp_v3_1_bootloader_load_sos(struct psp_context *psp) /* Copy Secure OS binary to PSP memory */ memcpy(psp->fw_pri_buf, psp->sos_start_addr, psp->sos_bin_size); - /* Provide the PSP secure OS to bootrom */ + /* Provide the PSP secure OS to bootloader */ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, (uint32_t)(psp->fw_pri_mc_addr >> 20)); psp_gfxdrv_command_reg = 2 << 16; diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c index 2d4770e173dd..9f3cb2aec7c2 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c @@ -225,7 +225,7 @@ static void sdma_v2_4_ring_set_wptr(struct amdgpu_ring *ring) static void sdma_v2_4_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) { - struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring); + struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring); int i; for (i = 0; i < count; i++) @@ -245,9 +245,12 @@ static void sdma_v2_4_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) * Schedule an IB in the DMA ring (VI). */ static void sdma_v2_4_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + /* IB packet must end on a 8 DW boundary */ sdma_v2_4_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8); @@ -349,8 +352,8 @@ static void sdma_v2_4_gfx_stop(struct amdgpu_device *adev) ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0); WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl); } - sdma0->ready = false; - sdma1->ready = false; + sdma0->sched.ready = false; + sdma1->sched.ready = false; } /** @@ -471,17 +474,15 @@ static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev) /* enable DMA IBs */ WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl); - ring->ready = true; + ring->sched.ready = true; } sdma_v2_4_enable(adev, true); for (i = 0; i < adev->sdma.num_instances; i++) { ring = &adev->sdma.instance[i].ring; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) return r; - } if (adev->mman.buffer_funcs_ring == ring) amdgpu_ttm_set_buffer_funcs_status(adev, true); @@ -550,21 +551,16 @@ static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring) u64 gpu_addr; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); r = amdgpu_ring_alloc(ring, 5); - if (r) { - DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_device_wb_free(adev, index); - return r; - } + if (r) + goto error_free_wb; amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR)); @@ -581,15 +577,11 @@ static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } - amdgpu_device_wb_free(adev, index); + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; +error_free_wb: + amdgpu_device_wb_free(adev, index); return r; } @@ -612,20 +604,16 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 256, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err0; - } ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR); @@ -644,21 +632,16 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out\n"); r = -ETIMEDOUT; goto err1; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto err1; } tmp = le32_to_cpu(adev->wb.wb[index]); - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + if (tmp == 0xDEADBEEF) r = 0; - } else { - DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp); + else r = -EINVAL; - } err1: amdgpu_ib_free(adev, &ib, NULL); @@ -760,7 +743,7 @@ static void sdma_v2_4_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe, */ static void sdma_v2_4_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib) { - struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring); + struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring); u32 pad_count; int i; @@ -1105,8 +1088,14 @@ static int sdma_v2_4_process_illegal_inst_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { + u8 instance_id, queue_id; + DRM_ERROR("Illegal instruction in SDMA command stream\n"); - schedule_work(&adev->reset_work); + instance_id = (entry->ring_id & 0x3) >> 0; + queue_id = (entry->ring_id & 0xc) >> 2; + + if (instance_id <= 1 && queue_id == 0) + drm_sched_fault(&adev->sdma.instance[instance_id].ring.sched); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index 6fb3edaba0ec..b6a25f92d566 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -399,7 +399,7 @@ static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring) static void sdma_v3_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) { - struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring); + struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring); int i; for (i = 0; i < count; i++) @@ -419,9 +419,12 @@ static void sdma_v3_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) * Schedule an IB in the DMA ring (VI). */ static void sdma_v3_0_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + /* IB packet must end on a 8 DW boundary */ sdma_v3_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8); @@ -523,8 +526,8 @@ static void sdma_v3_0_gfx_stop(struct amdgpu_device *adev) ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0); WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl); } - sdma0->ready = false; - sdma1->ready = false; + sdma0->sched.ready = false; + sdma1->sched.ready = false; } /** @@ -739,7 +742,7 @@ static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev) /* enable DMA IBs */ WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl); - ring->ready = true; + ring->sched.ready = true; } /* unhalt the MEs */ @@ -749,11 +752,9 @@ static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev) for (i = 0; i < adev->sdma.num_instances; i++) { ring = &adev->sdma.instance[i].ring; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) return r; - } if (adev->mman.buffer_funcs_ring == ring) amdgpu_ttm_set_buffer_funcs_status(adev, true); @@ -822,21 +823,16 @@ static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring) u64 gpu_addr; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); r = amdgpu_ring_alloc(ring, 5); - if (r) { - DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_device_wb_free(adev, index); - return r; - } + if (r) + goto error_free_wb; amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR)); @@ -853,15 +849,11 @@ static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } - amdgpu_device_wb_free(adev, index); + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; +error_free_wb: + amdgpu_device_wb_free(adev, index); return r; } @@ -884,20 +876,16 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 256, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err0; - } ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR); @@ -916,21 +904,16 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out\n"); r = -ETIMEDOUT; goto err1; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto err1; } tmp = le32_to_cpu(adev->wb.wb[index]); - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + if (tmp == 0xDEADBEEF) r = 0; - } else { - DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp); + else r = -EINVAL; - } err1: amdgpu_ib_free(adev, &ib, NULL); dma_fence_put(f); @@ -1031,7 +1014,7 @@ static void sdma_v3_0_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe, */ static void sdma_v3_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib) { - struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring); + struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring); u32 pad_count; int i; @@ -1440,8 +1423,14 @@ static int sdma_v3_0_process_illegal_inst_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { + u8 instance_id, queue_id; + DRM_ERROR("Illegal instruction in SDMA command stream\n"); - schedule_work(&adev->reset_work); + instance_id = (entry->ring_id & 0x3) >> 0; + queue_id = (entry->ring_id & 0xc) >> 2; + + if (instance_id <= 1 && queue_id == 0) + drm_sched_fault(&adev->sdma.instance[instance_id].ring.sched); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index 7a8c9172d30a..f4490cdd9804 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -54,6 +54,11 @@ MODULE_FIRMWARE("amdgpu/raven2_sdma.bin"); #define SDMA0_POWER_CNTL__ON_OFF_CONDITION_HOLD_TIME_MASK 0x000000F8L #define SDMA0_POWER_CNTL__ON_OFF_STATUS_DURATION_TIME_MASK 0xFC000000L +#define WREG32_SDMA(instance, offset, value) \ + WREG32(sdma_v4_0_get_reg_offset(adev, (instance), (offset)), value) +#define RREG32_SDMA(instance, offset) \ + RREG32(sdma_v4_0_get_reg_offset(adev, (instance), (offset))) + static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev); static void sdma_v4_0_set_buffer_funcs(struct amdgpu_device *adev); static void sdma_v4_0_set_vm_pte_funcs(struct amdgpu_device *adev); @@ -367,16 +372,11 @@ static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring) wptr = READ_ONCE(*((u64 *)&adev->wb.wb[ring->wptr_offs])); DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr); } else { - u32 lowbit, highbit; - - lowbit = RREG32(sdma_v4_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR)) >> 2; - highbit = RREG32(sdma_v4_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI)) >> 2; - - DRM_DEBUG("wptr [%i]high== 0x%08x low==0x%08x\n", - ring->me, highbit, lowbit); - wptr = highbit; + wptr = RREG32_SDMA(ring->me, mmSDMA0_GFX_RB_WPTR_HI); wptr = wptr << 32; - wptr |= lowbit; + wptr |= RREG32_SDMA(ring->me, mmSDMA0_GFX_RB_WPTR); + DRM_DEBUG("wptr before shift [%i] wptr == 0x%016llx\n", + ring->me, wptr); } return wptr >> 2; @@ -417,14 +417,67 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) lower_32_bits(ring->wptr << 2), ring->me, upper_32_bits(ring->wptr << 2)); - WREG32(sdma_v4_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr << 2)); - WREG32(sdma_v4_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr << 2)); + WREG32_SDMA(ring->me, mmSDMA0_GFX_RB_WPTR, + lower_32_bits(ring->wptr << 2)); + WREG32_SDMA(ring->me, mmSDMA0_GFX_RB_WPTR_HI, + upper_32_bits(ring->wptr << 2)); + } +} + +/** + * sdma_v4_0_page_ring_get_wptr - get the current write pointer + * + * @ring: amdgpu ring pointer + * + * Get the current wptr from the hardware (VEGA10+). + */ +static uint64_t sdma_v4_0_page_ring_get_wptr(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + u64 wptr; + + if (ring->use_doorbell) { + /* XXX check if swapping is necessary on BE */ + wptr = READ_ONCE(*((u64 *)&adev->wb.wb[ring->wptr_offs])); + } else { + wptr = RREG32_SDMA(ring->me, mmSDMA0_PAGE_RB_WPTR_HI); + wptr = wptr << 32; + wptr |= RREG32_SDMA(ring->me, mmSDMA0_PAGE_RB_WPTR); + } + + return wptr >> 2; +} + +/** + * sdma_v4_0_ring_set_wptr - commit the write pointer + * + * @ring: amdgpu ring pointer + * + * Write the wptr back to the hardware (VEGA10+). + */ +static void sdma_v4_0_page_ring_set_wptr(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + + if (ring->use_doorbell) { + u64 *wb = (u64 *)&adev->wb.wb[ring->wptr_offs]; + + /* XXX check if swapping is necessary on BE */ + WRITE_ONCE(*wb, (ring->wptr << 2)); + WDOORBELL64(ring->doorbell_index, ring->wptr << 2); + } else { + uint64_t wptr = ring->wptr << 2; + + WREG32_SDMA(ring->me, mmSDMA0_PAGE_RB_WPTR, + lower_32_bits(wptr)); + WREG32_SDMA(ring->me, mmSDMA0_PAGE_RB_WPTR_HI, + upper_32_bits(wptr)); } } static void sdma_v4_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) { - struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring); + struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring); int i; for (i = 0; i < count; i++) @@ -444,9 +497,12 @@ static void sdma_v4_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) * Schedule an IB in the DMA ring (VEGA10). */ static void sdma_v4_0_ring_emit_ib(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + /* IB packet must end on a 8 DW boundary */ sdma_v4_0_ring_insert_nop(ring, (10 - (lower_32_bits(ring->wptr) & 7)) % 8); @@ -568,16 +624,16 @@ static void sdma_v4_0_gfx_stop(struct amdgpu_device *adev) amdgpu_ttm_set_buffer_funcs_status(adev, false); for (i = 0; i < adev->sdma.num_instances; i++) { - rb_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL)); + rb_cntl = RREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL); rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl); - ib_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL)); + WREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL, rb_cntl); + ib_cntl = RREG32_SDMA(i, mmSDMA0_GFX_IB_CNTL); ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl); + WREG32_SDMA(i, mmSDMA0_GFX_IB_CNTL, ib_cntl); } - sdma0->ready = false; - sdma1->ready = false; + sdma0->sched.ready = false; + sdma1->sched.ready = false; } /** @@ -593,6 +649,39 @@ static void sdma_v4_0_rlc_stop(struct amdgpu_device *adev) } /** + * sdma_v4_0_page_stop - stop the page async dma engines + * + * @adev: amdgpu_device pointer + * + * Stop the page async dma ring buffers (VEGA10). + */ +static void sdma_v4_0_page_stop(struct amdgpu_device *adev) +{ + struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].page; + struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].page; + u32 rb_cntl, ib_cntl; + int i; + + if ((adev->mman.buffer_funcs_ring == sdma0) || + (adev->mman.buffer_funcs_ring == sdma1)) + amdgpu_ttm_set_buffer_funcs_status(adev, false); + + for (i = 0; i < adev->sdma.num_instances; i++) { + rb_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_PAGE_RB_CNTL, + RB_ENABLE, 0); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL, rb_cntl); + ib_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_IB_CNTL); + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_PAGE_IB_CNTL, + IB_ENABLE, 0); + WREG32_SDMA(i, mmSDMA0_PAGE_IB_CNTL, ib_cntl); + } + + sdma0->sched.ready = false; + sdma1->sched.ready = false; +} + +/** * sdma_v_0_ctx_switch_enable - stop the async dma engines context switch * * @adev: amdgpu_device pointer @@ -630,18 +719,15 @@ static void sdma_v4_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable) } for (i = 0; i < adev->sdma.num_instances; i++) { - f32_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_CNTL)); + f32_cntl = RREG32_SDMA(i, mmSDMA0_CNTL); f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL, AUTO_CTXSW_ENABLE, enable ? 1 : 0); if (enable && amdgpu_sdma_phase_quantum) { - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_PHASE0_QUANTUM), - phase_quantum); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_PHASE1_QUANTUM), - phase_quantum); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_PHASE2_QUANTUM), - phase_quantum); + WREG32_SDMA(i, mmSDMA0_PHASE0_QUANTUM, phase_quantum); + WREG32_SDMA(i, mmSDMA0_PHASE1_QUANTUM, phase_quantum); + WREG32_SDMA(i, mmSDMA0_PHASE2_QUANTUM, phase_quantum); } - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_CNTL), f32_cntl); + WREG32_SDMA(i, mmSDMA0_CNTL, f32_cntl); } } @@ -662,156 +748,217 @@ static void sdma_v4_0_enable(struct amdgpu_device *adev, bool enable) if (enable == false) { sdma_v4_0_gfx_stop(adev); sdma_v4_0_rlc_stop(adev); + if (adev->sdma.has_page_queue) + sdma_v4_0_page_stop(adev); } for (i = 0; i < adev->sdma.num_instances; i++) { - f32_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL)); + f32_cntl = RREG32_SDMA(i, mmSDMA0_F32_CNTL); f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, enable ? 0 : 1); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), f32_cntl); + WREG32_SDMA(i, mmSDMA0_F32_CNTL, f32_cntl); } } /** + * sdma_v4_0_rb_cntl - get parameters for rb_cntl + */ +static uint32_t sdma_v4_0_rb_cntl(struct amdgpu_ring *ring, uint32_t rb_cntl) +{ + /* Set ring buffer size in dwords */ + uint32_t rb_bufsz = order_base_2(ring->ring_size / 4); + + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz); +#ifdef __BIG_ENDIAN + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, + RPTR_WRITEBACK_SWAP_ENABLE, 1); +#endif + return rb_cntl; +} + +/** * sdma_v4_0_gfx_resume - setup and start the async dma engines * * @adev: amdgpu_device pointer + * @i: instance to resume * * Set up the gfx DMA ring buffers and enable them (VEGA10). * Returns 0 for success, error for failure. */ -static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev) +static void sdma_v4_0_gfx_resume(struct amdgpu_device *adev, unsigned int i) { - struct amdgpu_ring *ring; + struct amdgpu_ring *ring = &adev->sdma.instance[i].ring; u32 rb_cntl, ib_cntl, wptr_poll_cntl; - u32 rb_bufsz; u32 wb_offset; u32 doorbell; u32 doorbell_offset; - u32 temp; u64 wptr_gpu_addr; - int i, r; - - for (i = 0; i < adev->sdma.num_instances; i++) { - ring = &adev->sdma.instance[i].ring; - wb_offset = (ring->rptr_offs * 4); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0); + wb_offset = (ring->rptr_offs * 4); - /* Set ring buffer size in dwords */ - rb_bufsz = order_base_2(ring->ring_size / 4); - rb_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL)); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz); -#ifdef __BIG_ENDIAN - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, - RPTR_WRITEBACK_SWAP_ENABLE, 1); -#endif - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl); + rb_cntl = RREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL); + rb_cntl = sdma_v4_0_rb_cntl(ring, rb_cntl); + WREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL, rb_cntl); - /* Initialize the ring buffer's read and write pointers */ - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), 0); + /* Initialize the ring buffer's read and write pointers */ + WREG32_SDMA(i, mmSDMA0_GFX_RB_RPTR, 0); + WREG32_SDMA(i, mmSDMA0_GFX_RB_RPTR_HI, 0); + WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR, 0); + WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_HI, 0); - /* set the wb address whether it's enabled or not */ - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI), - upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO), - lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC); + /* set the wb address whether it's enabled or not */ + WREG32_SDMA(i, mmSDMA0_GFX_RB_RPTR_ADDR_HI, + upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF); + WREG32_SDMA(i, mmSDMA0_GFX_RB_RPTR_ADDR_LO, + lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, + RPTR_WRITEBACK_ENABLE, 1); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE), ring->gpu_addr >> 8); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI), ring->gpu_addr >> 40); + WREG32_SDMA(i, mmSDMA0_GFX_RB_BASE, ring->gpu_addr >> 8); + WREG32_SDMA(i, mmSDMA0_GFX_RB_BASE_HI, ring->gpu_addr >> 40); - ring->wptr = 0; + ring->wptr = 0; - /* before programing wptr to a less value, need set minor_ptr_update first */ - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1); + /* before programing wptr to a less value, need set minor_ptr_update first */ + WREG32_SDMA(i, mmSDMA0_GFX_MINOR_PTR_UPDATE, 1); - if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */ - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr) << 2); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2); - } + doorbell = RREG32_SDMA(i, mmSDMA0_GFX_DOORBELL); + doorbell_offset = RREG32_SDMA(i, mmSDMA0_GFX_DOORBELL_OFFSET); - doorbell = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL)); - doorbell_offset = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET)); - - if (ring->use_doorbell) { - doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1); - doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_GFX_DOORBELL_OFFSET, + doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, + ring->use_doorbell); + doorbell_offset = REG_SET_FIELD(doorbell_offset, + SDMA0_GFX_DOORBELL_OFFSET, OFFSET, ring->doorbell_index); - } else { - doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0); - } - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET), doorbell_offset); - adev->nbio_funcs->sdma_doorbell_range(adev, i, ring->use_doorbell, - ring->doorbell_index); + WREG32_SDMA(i, mmSDMA0_GFX_DOORBELL, doorbell); + WREG32_SDMA(i, mmSDMA0_GFX_DOORBELL_OFFSET, doorbell_offset); + adev->nbio_funcs->sdma_doorbell_range(adev, i, ring->use_doorbell, + ring->doorbell_index); + + sdma_v4_0_ring_set_wptr(ring); + + /* set minor_ptr_update to 0 after wptr programed */ + WREG32_SDMA(i, mmSDMA0_GFX_MINOR_PTR_UPDATE, 0); + + /* setup the wptr shadow polling */ + wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4); + WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO, + lower_32_bits(wptr_gpu_addr)); + WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI, + upper_32_bits(wptr_gpu_addr)); + wptr_poll_cntl = RREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL); + wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, + SDMA0_GFX_RB_WPTR_POLL_CNTL, + F32_POLL_ENABLE, amdgpu_sriov_vf(adev)); + WREG32_SDMA(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL, wptr_poll_cntl); + + /* enable DMA RB */ + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1); + WREG32_SDMA(i, mmSDMA0_GFX_RB_CNTL, rb_cntl); + + ib_cntl = RREG32_SDMA(i, mmSDMA0_GFX_IB_CNTL); + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1); +#ifdef __BIG_ENDIAN + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1); +#endif + /* enable DMA IBs */ + WREG32_SDMA(i, mmSDMA0_GFX_IB_CNTL, ib_cntl); - if (amdgpu_sriov_vf(adev)) - sdma_v4_0_ring_set_wptr(ring); + ring->sched.ready = true; +} - /* set minor_ptr_update to 0 after wptr programed */ - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0); +/** + * sdma_v4_0_page_resume - setup and start the async dma engines + * + * @adev: amdgpu_device pointer + * @i: instance to resume + * + * Set up the page DMA ring buffers and enable them (VEGA10). + * Returns 0 for success, error for failure. + */ +static void sdma_v4_0_page_resume(struct amdgpu_device *adev, unsigned int i) +{ + struct amdgpu_ring *ring = &adev->sdma.instance[i].page; + u32 rb_cntl, ib_cntl, wptr_poll_cntl; + u32 wb_offset; + u32 doorbell; + u32 doorbell_offset; + u64 wptr_gpu_addr; - /* set utc l1 enable flag always to 1 */ - temp = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_CNTL)); - temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_CNTL), temp); + wb_offset = (ring->rptr_offs * 4); - if (!amdgpu_sriov_vf(adev)) { - /* unhalt engine */ - temp = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL)); - temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), temp); - } + rb_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL); + rb_cntl = sdma_v4_0_rb_cntl(ring, rb_cntl); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL, rb_cntl); - /* setup the wptr shadow polling */ - wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO), - lower_32_bits(wptr_gpu_addr)); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI), - upper_32_bits(wptr_gpu_addr)); - wptr_poll_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL)); - if (amdgpu_sriov_vf(adev)) - wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 1); - else - wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 0); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), wptr_poll_cntl); + /* Initialize the ring buffer's read and write pointers */ + WREG32_SDMA(i, mmSDMA0_PAGE_RB_RPTR, 0); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_RPTR_HI, 0); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR, 0); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_HI, 0); - /* enable DMA RB */ - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl); + /* set the wb address whether it's enabled or not */ + WREG32_SDMA(i, mmSDMA0_PAGE_RB_RPTR_ADDR_HI, + upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_RPTR_ADDR_LO, + lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC); - ib_cntl = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL)); - ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1); -#ifdef __BIG_ENDIAN - ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1); -#endif - /* enable DMA IBs */ - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_PAGE_RB_CNTL, + RPTR_WRITEBACK_ENABLE, 1); - ring->ready = true; + WREG32_SDMA(i, mmSDMA0_PAGE_RB_BASE, ring->gpu_addr >> 8); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_BASE_HI, ring->gpu_addr >> 40); - if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */ - sdma_v4_0_ctx_switch_enable(adev, true); - sdma_v4_0_enable(adev, true); - } + ring->wptr = 0; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; - return r; - } + /* before programing wptr to a less value, need set minor_ptr_update first */ + WREG32_SDMA(i, mmSDMA0_PAGE_MINOR_PTR_UPDATE, 1); - if (adev->mman.buffer_funcs_ring == ring) - amdgpu_ttm_set_buffer_funcs_status(adev, true); + doorbell = RREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL); + doorbell_offset = RREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL_OFFSET); - } + doorbell = REG_SET_FIELD(doorbell, SDMA0_PAGE_DOORBELL, ENABLE, + ring->use_doorbell); + doorbell_offset = REG_SET_FIELD(doorbell_offset, + SDMA0_PAGE_DOORBELL_OFFSET, + OFFSET, ring->doorbell_index); + WREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL, doorbell); + WREG32_SDMA(i, mmSDMA0_PAGE_DOORBELL_OFFSET, doorbell_offset); + /* TODO: enable doorbell support */ + /*adev->nbio_funcs->sdma_doorbell_range(adev, i, ring->use_doorbell, + ring->doorbell_index);*/ + + sdma_v4_0_ring_set_wptr(ring); + + /* set minor_ptr_update to 0 after wptr programed */ + WREG32_SDMA(i, mmSDMA0_PAGE_MINOR_PTR_UPDATE, 0); + + /* setup the wptr shadow polling */ + wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_POLL_ADDR_LO, + lower_32_bits(wptr_gpu_addr)); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_POLL_ADDR_HI, + upper_32_bits(wptr_gpu_addr)); + wptr_poll_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL); + wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, + SDMA0_PAGE_RB_WPTR_POLL_CNTL, + F32_POLL_ENABLE, amdgpu_sriov_vf(adev)); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_WPTR_POLL_CNTL, wptr_poll_cntl); + + /* enable DMA RB */ + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_PAGE_RB_CNTL, RB_ENABLE, 1); + WREG32_SDMA(i, mmSDMA0_PAGE_RB_CNTL, rb_cntl); + + ib_cntl = RREG32_SDMA(i, mmSDMA0_PAGE_IB_CNTL); + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_PAGE_IB_CNTL, IB_ENABLE, 1); +#ifdef __BIG_ENDIAN + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_PAGE_IB_CNTL, IB_SWAP_ENABLE, 1); +#endif + /* enable DMA IBs */ + WREG32_SDMA(i, mmSDMA0_PAGE_IB_CNTL, ib_cntl); - return 0; + ring->sched.ready = true; } static void @@ -922,12 +1069,14 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev) (adev->sdma.instance[i].fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes)); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_UCODE_ADDR), 0); + WREG32_SDMA(i, mmSDMA0_UCODE_ADDR, 0); for (j = 0; j < fw_size; j++) - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_UCODE_DATA), le32_to_cpup(fw_data++)); + WREG32_SDMA(i, mmSDMA0_UCODE_DATA, + le32_to_cpup(fw_data++)); - WREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_UCODE_ADDR), adev->sdma.instance[i].fw_version); + WREG32_SDMA(i, mmSDMA0_UCODE_ADDR, + adev->sdma.instance[i].fw_version); } return 0; @@ -943,33 +1092,78 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev) */ static int sdma_v4_0_start(struct amdgpu_device *adev) { - int r = 0; + struct amdgpu_ring *ring; + int i, r; if (amdgpu_sriov_vf(adev)) { sdma_v4_0_ctx_switch_enable(adev, false); sdma_v4_0_enable(adev, false); + } else { - /* set RB registers */ - r = sdma_v4_0_gfx_resume(adev); - return r; + if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) { + r = sdma_v4_0_load_microcode(adev); + if (r) + return r; + } + + /* unhalt the MEs */ + sdma_v4_0_enable(adev, true); + /* enable sdma ring preemption */ + sdma_v4_0_ctx_switch_enable(adev, true); + } + + /* start the gfx rings and rlc compute queues */ + for (i = 0; i < adev->sdma.num_instances; i++) { + uint32_t temp; + + WREG32_SDMA(i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL, 0); + sdma_v4_0_gfx_resume(adev, i); + if (adev->sdma.has_page_queue) + sdma_v4_0_page_resume(adev, i); + + /* set utc l1 enable flag always to 1 */ + temp = RREG32_SDMA(i, mmSDMA0_CNTL); + temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1); + WREG32_SDMA(i, mmSDMA0_CNTL, temp); + + if (!amdgpu_sriov_vf(adev)) { + /* unhalt engine */ + temp = RREG32_SDMA(i, mmSDMA0_F32_CNTL); + temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0); + WREG32_SDMA(i, mmSDMA0_F32_CNTL, temp); + } } - if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) { - r = sdma_v4_0_load_microcode(adev); + if (amdgpu_sriov_vf(adev)) { + sdma_v4_0_ctx_switch_enable(adev, true); + sdma_v4_0_enable(adev, true); + } else { + r = sdma_v4_0_rlc_resume(adev); if (r) return r; } - /* unhalt the MEs */ - sdma_v4_0_enable(adev, true); - /* enable sdma ring preemption */ - sdma_v4_0_ctx_switch_enable(adev, true); + for (i = 0; i < adev->sdma.num_instances; i++) { + ring = &adev->sdma.instance[i].ring; - /* start the gfx rings and rlc compute queues */ - r = sdma_v4_0_gfx_resume(adev); - if (r) - return r; - r = sdma_v4_0_rlc_resume(adev); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + + if (adev->sdma.has_page_queue) { + struct amdgpu_ring *page = &adev->sdma.instance[i].page; + + r = amdgpu_ring_test_helper(page); + if (r) + return r; + + if (adev->mman.buffer_funcs_ring == page) + amdgpu_ttm_set_buffer_funcs_status(adev, true); + } + + if (adev->mman.buffer_funcs_ring == ring) + amdgpu_ttm_set_buffer_funcs_status(adev, true); + } return r; } @@ -993,21 +1187,16 @@ static int sdma_v4_0_ring_test_ring(struct amdgpu_ring *ring) u64 gpu_addr; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); r = amdgpu_ring_alloc(ring, 5); - if (r) { - DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_device_wb_free(adev, index); - return r; - } + if (r) + goto error_free_wb; amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR)); @@ -1024,15 +1213,11 @@ static int sdma_v4_0_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } - amdgpu_device_wb_free(adev, index); + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; +error_free_wb: + amdgpu_device_wb_free(adev, index); return r; } @@ -1055,20 +1240,16 @@ static int sdma_v4_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) u64 gpu_addr; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 256, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err0; - } ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR); @@ -1087,21 +1268,17 @@ static int sdma_v4_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out\n"); r = -ETIMEDOUT; goto err1; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto err1; } tmp = le32_to_cpu(adev->wb.wb[index]); - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + if (tmp == 0xDEADBEEF) r = 0; - } else { - DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp); + else r = -EINVAL; - } + err1: amdgpu_ib_free(adev, &ib, NULL); dma_fence_put(f); @@ -1206,7 +1383,7 @@ static void sdma_v4_0_vm_set_pte_pde(struct amdgpu_ib *ib, */ static void sdma_v4_0_ring_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib) { - struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring); + struct amdgpu_sdma_instance *sdma = amdgpu_sdma_get_instance_from_ring(ring); u32 pad_count; int i; @@ -1276,10 +1453,18 @@ static int sdma_v4_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (adev->asic_type == CHIP_RAVEN) + if (adev->asic_type == CHIP_RAVEN) { adev->sdma.num_instances = 1; - else + adev->sdma.has_page_queue = false; + } else { adev->sdma.num_instances = 2; + /* TODO: Page queue breaks driver reload under SRIOV */ + if ((adev->asic_type == CHIP_VEGA10) && amdgpu_sriov_vf((adev))) + adev->sdma.has_page_queue = false; + else if (adev->asic_type != CHIP_VEGA20 && + adev->asic_type != CHIP_VEGA12) + adev->sdma.has_page_queue = true; + } sdma_v4_0_set_ring_funcs(adev); sdma_v4_0_set_buffer_funcs(adev); @@ -1340,6 +1525,21 @@ static int sdma_v4_0_sw_init(void *handle) AMDGPU_SDMA_IRQ_TRAP1); if (r) return r; + + if (adev->sdma.has_page_queue) { + ring = &adev->sdma.instance[i].page; + ring->ring_obj = NULL; + ring->use_doorbell = false; + + sprintf(ring->name, "page%d", i); + r = amdgpu_ring_init(adev, ring, 1024, + &adev->sdma.trap_irq, + (i == 0) ? + AMDGPU_SDMA_IRQ_TRAP0 : + AMDGPU_SDMA_IRQ_TRAP1); + if (r) + return r; + } } return r; @@ -1350,8 +1550,11 @@ static int sdma_v4_0_sw_fini(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; int i; - for (i = 0; i < adev->sdma.num_instances; i++) + for (i = 0; i < adev->sdma.num_instances; i++) { amdgpu_ring_fini(&adev->sdma.instance[i].ring); + if (adev->sdma.has_page_queue) + amdgpu_ring_fini(&adev->sdma.instance[i].page); + } for (i = 0; i < adev->sdma.num_instances; i++) { release_firmware(adev->sdma.instance[i].fw); @@ -1414,7 +1617,7 @@ static bool sdma_v4_0_is_idle(void *handle) u32 i; for (i = 0; i < adev->sdma.num_instances; i++) { - u32 tmp = RREG32(sdma_v4_0_get_reg_offset(adev, i, mmSDMA0_STATUS_REG)); + u32 tmp = RREG32_SDMA(i, mmSDMA0_STATUS_REG); if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK)) return false; @@ -1430,8 +1633,8 @@ static int sdma_v4_0_wait_for_idle(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; for (i = 0; i < adev->usec_timeout; i++) { - sdma0 = RREG32(sdma_v4_0_get_reg_offset(adev, 0, mmSDMA0_STATUS_REG)); - sdma1 = RREG32(sdma_v4_0_get_reg_offset(adev, 1, mmSDMA0_STATUS_REG)); + sdma0 = RREG32_SDMA(0, mmSDMA0_STATUS_REG); + sdma1 = RREG32_SDMA(1, mmSDMA0_STATUS_REG); if (sdma0 & sdma1 & SDMA0_STATUS_REG__IDLE_MASK) return 0; @@ -1452,16 +1655,13 @@ static int sdma_v4_0_set_trap_irq_state(struct amdgpu_device *adev, unsigned type, enum amdgpu_interrupt_state state) { + unsigned int instance = (type == AMDGPU_SDMA_IRQ_TRAP0) ? 0 : 1; u32 sdma_cntl; - u32 reg_offset = (type == AMDGPU_SDMA_IRQ_TRAP0) ? - sdma_v4_0_get_reg_offset(adev, 0, mmSDMA0_CNTL) : - sdma_v4_0_get_reg_offset(adev, 1, mmSDMA0_CNTL); - - sdma_cntl = RREG32(reg_offset); + sdma_cntl = RREG32_SDMA(instance, mmSDMA0_CNTL); sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, state == AMDGPU_IRQ_STATE_ENABLE ? 1 : 0); - WREG32(reg_offset, sdma_cntl); + WREG32_SDMA(instance, mmSDMA0_CNTL, sdma_cntl); return 0; } @@ -1470,39 +1670,32 @@ static int sdma_v4_0_process_trap_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { + uint32_t instance; + DRM_DEBUG("IH: SDMA trap\n"); switch (entry->client_id) { case SOC15_IH_CLIENTID_SDMA0: - switch (entry->ring_id) { - case 0: - amdgpu_fence_process(&adev->sdma.instance[0].ring); - break; - case 1: - /* XXX compute */ - break; - case 2: - /* XXX compute */ - break; - case 3: - /* XXX page queue*/ - break; - } + instance = 0; break; case SOC15_IH_CLIENTID_SDMA1: - switch (entry->ring_id) { - case 0: - amdgpu_fence_process(&adev->sdma.instance[1].ring); - break; - case 1: - /* XXX compute */ - break; - case 2: - /* XXX compute */ - break; - case 3: - /* XXX page queue*/ - break; - } + instance = 1; + break; + default: + return 0; + } + + switch (entry->ring_id) { + case 0: + amdgpu_fence_process(&adev->sdma.instance[instance].ring); + break; + case 1: + /* XXX compute */ + break; + case 2: + /* XXX compute */ + break; + case 3: + amdgpu_fence_process(&adev->sdma.instance[instance].page); break; } return 0; @@ -1512,12 +1705,29 @@ static int sdma_v4_0_process_illegal_inst_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { + int instance; + DRM_ERROR("Illegal instruction in SDMA command stream\n"); - schedule_work(&adev->reset_work); + + switch (entry->client_id) { + case SOC15_IH_CLIENTID_SDMA0: + instance = 0; + break; + case SOC15_IH_CLIENTID_SDMA1: + instance = 1; + break; + default: + return 0; + } + + switch (entry->ring_id) { + case 0: + drm_sched_fault(&adev->sdma.instance[instance].ring.sched); + break; + } return 0; } - static void sdma_v4_0_update_medium_grain_clock_gating( struct amdgpu_device *adev, bool enable) @@ -1730,6 +1940,38 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = { .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, }; +static const struct amdgpu_ring_funcs sdma_v4_0_page_ring_funcs = { + .type = AMDGPU_RING_TYPE_SDMA, + .align_mask = 0xf, + .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), + .support_64bit_ptrs = true, + .vmhub = AMDGPU_MMHUB, + .get_rptr = sdma_v4_0_ring_get_rptr, + .get_wptr = sdma_v4_0_page_ring_get_wptr, + .set_wptr = sdma_v4_0_page_ring_set_wptr, + .emit_frame_size = + 6 + /* sdma_v4_0_ring_emit_hdp_flush */ + 3 + /* hdp invalidate */ + 6 + /* sdma_v4_0_ring_emit_pipeline_sync */ + /* sdma_v4_0_ring_emit_vm_flush */ + SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 + + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 + + 10 + 10 + 10, /* sdma_v4_0_ring_emit_fence x3 for user fence, vm fence */ + .emit_ib_size = 7 + 6, /* sdma_v4_0_ring_emit_ib */ + .emit_ib = sdma_v4_0_ring_emit_ib, + .emit_fence = sdma_v4_0_ring_emit_fence, + .emit_pipeline_sync = sdma_v4_0_ring_emit_pipeline_sync, + .emit_vm_flush = sdma_v4_0_ring_emit_vm_flush, + .emit_hdp_flush = sdma_v4_0_ring_emit_hdp_flush, + .test_ring = sdma_v4_0_ring_test_ring, + .test_ib = sdma_v4_0_ring_test_ib, + .insert_nop = sdma_v4_0_ring_insert_nop, + .pad_ib = sdma_v4_0_ring_pad_ib, + .emit_wreg = sdma_v4_0_ring_emit_wreg, + .emit_reg_wait = sdma_v4_0_ring_emit_reg_wait, + .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, +}; + static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) { int i; @@ -1737,6 +1979,10 @@ static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) for (i = 0; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs; adev->sdma.instance[i].ring.me = i; + if (adev->sdma.has_page_queue) { + adev->sdma.instance[i].page.funcs = &sdma_v4_0_page_ring_funcs; + adev->sdma.instance[i].page.me = i; + } } } @@ -1818,7 +2064,10 @@ static const struct amdgpu_buffer_funcs sdma_v4_0_buffer_funcs = { static void sdma_v4_0_set_buffer_funcs(struct amdgpu_device *adev) { adev->mman.buffer_funcs = &sdma_v4_0_buffer_funcs; - adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring; + if (adev->sdma.has_page_queue) + adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].page; + else + adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring; } static const struct amdgpu_vm_pte_funcs sdma_v4_0_vm_pte_funcs = { @@ -1836,7 +2085,10 @@ static void sdma_v4_0_set_vm_pte_funcs(struct amdgpu_device *adev) adev->vm_manager.vm_pte_funcs = &sdma_v4_0_vm_pte_funcs; for (i = 0; i < adev->sdma.num_instances; i++) { - sched = &adev->sdma.instance[i].ring.sched; + if (adev->sdma.has_page_queue) + sched = &adev->sdma.instance[i].page.sched; + else + sched = &adev->sdma.instance[i].ring.sched; adev->vm_manager.vm_pte_rqs[i] = &sched->sched_rq[DRM_SCHED_PRIORITY_KERNEL]; } diff --git a/drivers/gpu/drm/amd/amdgpu/si_dma.c b/drivers/gpu/drm/amd/amdgpu/si_dma.c index adbaea6da0d7..b6e473134e19 100644 --- a/drivers/gpu/drm/amd/amdgpu/si_dma.c +++ b/drivers/gpu/drm/amd/amdgpu/si_dma.c @@ -61,9 +61,11 @@ static void si_dma_ring_set_wptr(struct amdgpu_ring *ring) } static void si_dma_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); /* The indirect buffer packet must end on an 8 DW boundary in the DMA ring. * Pad as necessary with NOPs. */ @@ -122,7 +124,7 @@ static void si_dma_stop(struct amdgpu_device *adev) if (adev->mman.buffer_funcs_ring == ring) amdgpu_ttm_set_buffer_funcs_status(adev, false); - ring->ready = false; + ring->sched.ready = false; } } @@ -175,13 +177,11 @@ static int si_dma_start(struct amdgpu_device *adev) WREG32(DMA_RB_WPTR + sdma_offsets[i], lower_32_bits(ring->wptr) << 2); WREG32(DMA_RB_CNTL + sdma_offsets[i], rb_cntl | DMA_RB_ENABLE); - ring->ready = true; + ring->sched.ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) return r; - } if (adev->mman.buffer_funcs_ring == ring) amdgpu_ttm_set_buffer_funcs_status(adev, true); @@ -209,21 +209,16 @@ static int si_dma_ring_test_ring(struct amdgpu_ring *ring) u64 gpu_addr; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); r = amdgpu_ring_alloc(ring, 4); - if (r) { - DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); - amdgpu_device_wb_free(adev, index); - return r; - } + if (r) + goto error_free_wb; amdgpu_ring_write(ring, DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, 1)); amdgpu_ring_write(ring, lower_32_bits(gpu_addr)); @@ -238,15 +233,11 @@ static int si_dma_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } - amdgpu_device_wb_free(adev, index); + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; +error_free_wb: + amdgpu_device_wb_free(adev, index); return r; } @@ -269,20 +260,16 @@ static int si_dma_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); + if (r) return r; - } gpu_addr = adev->wb.gpu_addr + (index * 4); tmp = 0xCAFEDEAD; adev->wb.wb[index] = cpu_to_le32(tmp); memset(&ib, 0, sizeof(ib)); r = amdgpu_ib_get(adev, NULL, 256, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + if (r) goto err0; - } ib.ptr[0] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, 1); ib.ptr[1] = lower_32_bits(gpu_addr); @@ -295,21 +282,16 @@ static int si_dma_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = dma_fence_wait_timeout(f, false, timeout); if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out\n"); r = -ETIMEDOUT; goto err1; } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); goto err1; } tmp = le32_to_cpu(adev->wb.wb[index]); - if (tmp == 0xDEADBEEF) { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + if (tmp == 0xDEADBEEF) r = 0; - } else { - DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp); + else r = -EINVAL; - } err1: amdgpu_ib_free(adev, &ib, NULL); @@ -658,15 +640,6 @@ static int si_dma_process_trap_irq(struct amdgpu_device *adev, return 0; } -static int si_dma_process_illegal_inst_irq(struct amdgpu_device *adev, - struct amdgpu_irq_src *source, - struct amdgpu_iv_entry *entry) -{ - DRM_ERROR("Illegal instruction in SDMA command stream\n"); - schedule_work(&adev->reset_work); - return 0; -} - static int si_dma_set_clockgating_state(void *handle, enum amd_clockgating_state state) { @@ -781,15 +754,10 @@ static const struct amdgpu_irq_src_funcs si_dma_trap_irq_funcs = { .process = si_dma_process_trap_irq, }; -static const struct amdgpu_irq_src_funcs si_dma_illegal_inst_irq_funcs = { - .process = si_dma_process_illegal_inst_irq, -}; - static void si_dma_set_irq_funcs(struct amdgpu_device *adev) { adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST; adev->sdma.trap_irq.funcs = &si_dma_trap_irq_funcs; - adev->sdma.illegal_inst_irq.funcs = &si_dma_illegal_inst_irq_funcs; } /** diff --git a/drivers/gpu/drm/amd/amdgpu/ta_xgmi_if.h b/drivers/gpu/drm/amd/amdgpu/ta_xgmi_if.h new file mode 100644 index 000000000000..ac2c27b7630c --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/ta_xgmi_if.h @@ -0,0 +1,130 @@ +/* + * Copyright 2018 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 _TA_XGMI_IF_H +#define _TA_XGMI_IF_H + +/* Responses have bit 31 set */ +#define RSP_ID_MASK (1U << 31) +#define RSP_ID(cmdId) (((uint32_t)(cmdId)) | RSP_ID_MASK) + +enum ta_command_xgmi { + TA_COMMAND_XGMI__INITIALIZE = 0x00, + TA_COMMAND_XGMI__GET_NODE_ID = 0x01, + TA_COMMAND_XGMI__GET_HIVE_ID = 0x02, + TA_COMMAND_XGMI__GET_GET_TOPOLOGY_INFO = 0x03, + TA_COMMAND_XGMI__SET_TOPOLOGY_INFO = 0x04 +}; + +/* XGMI related enumerations */ +/**********************************************************/; +enum ta_xgmi_connected_nodes { + TA_XGMI__MAX_CONNECTED_NODES = 64 +}; + +enum ta_xgmi_status { + TA_XGMI_STATUS__SUCCESS = 0x00, + TA_XGMI_STATUS__GENERIC_FAILURE = 0x01, + TA_XGMI_STATUS__NULL_POINTER = 0x02, + TA_XGMI_STATUS__INVALID_PARAMETER = 0x03, + TA_XGMI_STATUS__NOT_INITIALIZED = 0x04, + TA_XGMI_STATUS__INVALID_NODE_NUM = 0x05, + TA_XGMI_STATUS__INVALID_NODE_ID = 0x06, + TA_XGMI_STATUS__INVALID_TOPOLOGY = 0x07, + TA_XGMI_STATUS__FAILED_ID_GEN = 0x08, + TA_XGMI_STATUS__FAILED_TOPOLOGY_INIT = 0x09, + TA_XGMI_STATUS__SET_SHARING_ERROR = 0x0A +}; + +enum ta_xgmi_assigned_sdma_engine { + TA_XGMI_ASSIGNED_SDMA_ENGINE__NOT_ASSIGNED = -1, + TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA0 = 0, + TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA1 = 1, + TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA2 = 2, + TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA3 = 3, + TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA4 = 4, + TA_XGMI_ASSIGNED_SDMA_ENGINE__SDMA5 = 5 +}; + +/* input/output structures for XGMI commands */ +/**********************************************************/ +struct ta_xgmi_node_info { + uint64_t node_id; + uint8_t num_hops; + uint8_t is_sharing_enabled; + enum ta_xgmi_assigned_sdma_engine sdma_engine; +}; + +struct ta_xgmi_cmd_initialize_output { + uint32_t status; +}; + +struct ta_xgmi_cmd_get_node_id_output { + uint64_t node_id; +}; + +struct ta_xgmi_cmd_get_hive_id_output { + uint64_t hive_id; +}; + +struct ta_xgmi_cmd_get_topology_info_input { + uint32_t num_nodes; + struct ta_xgmi_node_info nodes[TA_XGMI__MAX_CONNECTED_NODES]; +}; + +struct ta_xgmi_cmd_get_topology_info_output { + uint32_t num_nodes; + struct ta_xgmi_node_info nodes[TA_XGMI__MAX_CONNECTED_NODES]; +}; + +struct ta_xgmi_cmd_set_topology_info_input { + uint32_t num_nodes; + struct ta_xgmi_node_info nodes[TA_XGMI__MAX_CONNECTED_NODES]; +}; + +/**********************************************************/ +/* Common input structure for XGMI callbacks */ +union ta_xgmi_cmd_input { + struct ta_xgmi_cmd_get_topology_info_input get_topology_info; + struct ta_xgmi_cmd_set_topology_info_input set_topology_info; +}; + +/* Common output structure for XGMI callbacks */ +union ta_xgmi_cmd_output { + struct ta_xgmi_cmd_initialize_output initialize; + struct ta_xgmi_cmd_get_node_id_output get_node_id; + struct ta_xgmi_cmd_get_hive_id_output get_hive_id; + struct ta_xgmi_cmd_get_topology_info_output get_topology_info; +}; +/**********************************************************/ + +struct ta_xgmi_shared_memory { + uint32_t cmd_id; + uint32_t resp_id; + enum ta_xgmi_status xgmi_status; + uint32_t reserved; + union ta_xgmi_cmd_input xgmi_in_message; + union ta_xgmi_cmd_output xgmi_out_message; +}; + +#endif //_TA_XGMI_IF_H diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c index 1fc17bf39fed..90bbcee00f28 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c @@ -162,12 +162,9 @@ static int uvd_v4_2_hw_init(void *handle) uvd_v4_2_enable_mgcg(adev, true); amdgpu_asic_set_uvd_clocks(adev, 10000, 10000); - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } r = amdgpu_ring_alloc(ring, 10); if (r) { @@ -218,7 +215,7 @@ static int uvd_v4_2_hw_fini(void *handle) if (RREG32(mmUVD_STATUS) != 0) uvd_v4_2_stop(adev); - ring->ready = false; + ring->sched.ready = false; return 0; } @@ -484,11 +481,9 @@ static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring) WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", - ring->idx, r); + if (r) return r; - } + amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0)); amdgpu_ring_write(ring, 0xDEADBEEF); amdgpu_ring_commit(ring); @@ -499,14 +494,9 @@ static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + return r; } @@ -519,8 +509,9 @@ static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring) * Write ring commands to execute the indirect buffer */ static void uvd_v4_2_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_BASE, 0)); amdgpu_ring_write(ring, ib->gpu_addr); diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c index fde6ad5ac9ab..1c5e12703103 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c @@ -158,12 +158,9 @@ static int uvd_v5_0_hw_init(void *handle) uvd_v5_0_set_clockgating_state(adev, AMD_CG_STATE_UNGATE); uvd_v5_0_enable_mgcg(adev, true); - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } r = amdgpu_ring_alloc(ring, 10); if (r) { @@ -215,7 +212,7 @@ static int uvd_v5_0_hw_fini(void *handle) if (RREG32(mmUVD_STATUS) != 0) uvd_v5_0_stop(adev); - ring->ready = false; + ring->sched.ready = false; return 0; } @@ -500,11 +497,8 @@ static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring) WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", - ring->idx, r); + if (r) return r; - } amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0)); amdgpu_ring_write(ring, 0xDEADBEEF); amdgpu_ring_commit(ring); @@ -515,14 +509,9 @@ static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + return r; } @@ -535,8 +524,9 @@ static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring) * Write ring commands to execute the indirect buffer */ static void uvd_v5_0_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0)); amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c index 7a5b40275e8e..f184842ef2a2 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c @@ -175,11 +175,8 @@ static int uvd_v6_0_enc_ring_test_ring(struct amdgpu_ring *ring) int r; r = amdgpu_ring_alloc(ring, 16); - if (r) { - DRM_ERROR("amdgpu: uvd enc failed to lock ring %d (%d).\n", - ring->idx, r); + if (r) return r; - } amdgpu_ring_write(ring, HEVC_ENC_CMD_END); amdgpu_ring_commit(ring); @@ -189,14 +186,8 @@ static int uvd_v6_0_enc_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed\n", - ring->idx); + if (i >= adev->usec_timeout) r = -ETIMEDOUT; - } return r; } @@ -336,31 +327,24 @@ static int uvd_v6_0_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = uvd_v6_0_enc_get_create_msg(ring, 1, NULL); - if (r) { - DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r); + if (r) goto error; - } r = uvd_v6_0_enc_get_destroy_msg(ring, 1, &fence); - if (r) { - DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r); + if (r) goto error; - } r = dma_fence_wait_timeout(fence, false, timeout); - if (r == 0) { - DRM_ERROR("amdgpu: IB test timed out.\n"); + if (r == 0) r = -ETIMEDOUT; - } else if (r < 0) { - DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r); - } else { - DRM_DEBUG("ib test on ring %d succeeded\n", ring->idx); + else if (r > 0) r = 0; - } + error: dma_fence_put(fence); return r; } + static int uvd_v6_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -476,12 +460,9 @@ static int uvd_v6_0_hw_init(void *handle) uvd_v6_0_set_clockgating_state(adev, AMD_CG_STATE_UNGATE); uvd_v6_0_enable_mgcg(adev, true); - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } r = amdgpu_ring_alloc(ring, 10); if (r) { @@ -513,12 +494,9 @@ static int uvd_v6_0_hw_init(void *handle) if (uvd_v6_0_enc_support(adev)) { for (i = 0; i < adev->uvd.num_enc_rings; ++i) { ring = &adev->uvd.inst->ring_enc[i]; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } } } @@ -548,7 +526,7 @@ static int uvd_v6_0_hw_fini(void *handle) if (RREG32(mmUVD_STATUS) != 0) uvd_v6_0_stop(adev); - ring->ready = false; + ring->sched.ready = false; return 0; } @@ -969,11 +947,9 @@ static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring) WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", - ring->idx, r); + if (r) return r; - } + amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0)); amdgpu_ring_write(ring, 0xDEADBEEF); amdgpu_ring_commit(ring); @@ -984,14 +960,9 @@ static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("ring test on %d succeeded in %d usecs\n", - ring->idx, i); - } else { - DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + return r; } @@ -1004,9 +975,12 @@ static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring) * Write ring commands to execute the indirect buffer */ static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_VMID, 0)); amdgpu_ring_write(ring, vmid); @@ -1027,8 +1001,12 @@ static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring, * Write enc ring commands to execute the indirect buffer */ static void uvd_v6_0_enc_ring_emit_ib(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + amdgpu_ring_write(ring, HEVC_ENC_CMD_IB_VM); amdgpu_ring_write(ring, vmid); amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c index 58b39afcfb86..8a4595968d98 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c @@ -183,11 +183,8 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring) return 0; r = amdgpu_ring_alloc(ring, 16); - if (r) { - DRM_ERROR("amdgpu: uvd enc failed to lock (%d)ring %d (%d).\n", - ring->me, ring->idx, r); + if (r) return r; - } amdgpu_ring_write(ring, HEVC_ENC_CMD_END); amdgpu_ring_commit(ring); @@ -197,14 +194,8 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("(%d)ring test on %d succeeded in %d usecs\n", - ring->me, ring->idx, i); - } else { - DRM_ERROR("amdgpu: (%d)ring %d test failed\n", - ring->me, ring->idx); + if (i >= adev->usec_timeout) r = -ETIMEDOUT; - } return r; } @@ -343,27 +334,19 @@ static int uvd_v7_0_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; r = uvd_v7_0_enc_get_create_msg(ring, 1, NULL); - if (r) { - DRM_ERROR("amdgpu: (%d)failed to get create msg (%ld).\n", ring->me, r); + if (r) goto error; - } r = uvd_v7_0_enc_get_destroy_msg(ring, 1, &fence); - if (r) { - DRM_ERROR("amdgpu: (%d)failed to get destroy ib (%ld).\n", ring->me, r); + if (r) goto error; - } r = dma_fence_wait_timeout(fence, false, timeout); - if (r == 0) { - DRM_ERROR("amdgpu: (%d)IB test timed out.\n", ring->me); + if (r == 0) r = -ETIMEDOUT; - } else if (r < 0) { - DRM_ERROR("amdgpu: (%d)fence wait failed (%ld).\n", ring->me, r); - } else { - DRM_DEBUG("ib test on (%d)ring %d succeeded\n", ring->me, ring->idx); + else if (r > 0) r = 0; - } + error: dma_fence_put(fence); return r; @@ -540,12 +523,9 @@ static int uvd_v7_0_hw_init(void *handle) ring = &adev->uvd.inst[j].ring; if (!amdgpu_sriov_vf(adev)) { - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } r = amdgpu_ring_alloc(ring, 10); if (r) { @@ -582,12 +562,9 @@ static int uvd_v7_0_hw_init(void *handle) for (i = 0; i < adev->uvd.num_enc_rings; ++i) { ring = &adev->uvd.inst[j].ring_enc[i]; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } } } done: @@ -619,7 +596,7 @@ static int uvd_v7_0_hw_fini(void *handle) for (i = 0; i < adev->uvd.num_uvd_inst; ++i) { if (adev->uvd.harvest_config & (1 << i)) continue; - adev->uvd.inst[i].ring.ready = false; + adev->uvd.inst[i].ring.sched.ready = false; } return 0; @@ -1235,11 +1212,9 @@ static int uvd_v7_0_ring_test_ring(struct amdgpu_ring *ring) WREG32_SOC15(UVD, ring->me, mmUVD_CONTEXT_ID, 0xCAFEDEAD); r = amdgpu_ring_alloc(ring, 3); - if (r) { - DRM_ERROR("amdgpu: (%d)cp failed to lock ring %d (%d).\n", - ring->me, ring->idx, r); + if (r) return r; - } + amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_CONTEXT_ID), 0)); amdgpu_ring_write(ring, 0xDEADBEEF); @@ -1251,14 +1226,9 @@ static int uvd_v7_0_ring_test_ring(struct amdgpu_ring *ring) DRM_UDELAY(1); } - if (i < adev->usec_timeout) { - DRM_DEBUG("(%d)ring test on %d succeeded in %d usecs\n", - ring->me, ring->idx, i); - } else { - DRM_ERROR("(%d)amdgpu: ring %d test failed (0x%08X)\n", - ring->me, ring->idx, tmp); - r = -EINVAL; - } + if (i >= adev->usec_timeout) + r = -ETIMEDOUT; + return r; } @@ -1300,10 +1270,12 @@ static int uvd_v7_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, * Write ring commands to execute the indirect buffer */ static void uvd_v7_0_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_job *job, struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + bool ctx_switch) { struct amdgpu_device *adev = ring->adev; + unsigned vmid = AMDGPU_JOB_GET_VMID(job); amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, ring->me, mmUVD_LMI_RBC_IB_VMID), 0)); @@ -1329,8 +1301,12 @@ static void uvd_v7_0_ring_emit_ib(struct amdgpu_ring *ring, * Write enc ring commands to execute the indirect buffer */ static void uvd_v7_0_enc_ring_emit_ib(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + amdgpu_ring_write(ring, HEVC_ENC_CMD_IB_VM); amdgpu_ring_write(ring, vmid); amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c index ea28828360d3..bed78a778e3f 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c @@ -463,15 +463,11 @@ static int vce_v2_0_hw_init(void *handle) amdgpu_asic_set_vce_clocks(adev, 10000, 10000); vce_v2_0_enable_mgcg(adev, true, false); - for (i = 0; i < adev->vce.num_rings; i++) - adev->vce.ring[i].ready = false; for (i = 0; i < adev->vce.num_rings; i++) { - r = amdgpu_ring_test_ring(&adev->vce.ring[i]); + r = amdgpu_ring_test_helper(&adev->vce.ring[i]); if (r) return r; - else - adev->vce.ring[i].ready = true; } DRM_INFO("VCE initialized successfully.\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index 6dbd39730070..3e84840859a7 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -474,15 +474,10 @@ static int vce_v3_0_hw_init(void *handle) amdgpu_asic_set_vce_clocks(adev, 10000, 10000); - for (i = 0; i < adev->vce.num_rings; i++) - adev->vce.ring[i].ready = false; - for (i = 0; i < adev->vce.num_rings; i++) { - r = amdgpu_ring_test_ring(&adev->vce.ring[i]); + r = amdgpu_ring_test_helper(&adev->vce.ring[i]); if (r) return r; - else - adev->vce.ring[i].ready = true; } DRM_INFO("VCE initialized successfully.\n"); @@ -838,8 +833,12 @@ out: } static void vce_v3_0_ring_emit_ib(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + amdgpu_ring_write(ring, VCE_CMD_IB_VM); amdgpu_ring_write(ring, vmid); amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c index 1c9471890bf7..0054ba1b9a68 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c @@ -519,15 +519,10 @@ static int vce_v4_0_hw_init(void *handle) if (r) return r; - for (i = 0; i < adev->vce.num_rings; i++) - adev->vce.ring[i].ready = false; - for (i = 0; i < adev->vce.num_rings; i++) { - r = amdgpu_ring_test_ring(&adev->vce.ring[i]); + r = amdgpu_ring_test_helper(&adev->vce.ring[i]); if (r) return r; - else - adev->vce.ring[i].ready = true; } DRM_INFO("VCE initialized successfully.\n"); @@ -549,7 +544,7 @@ static int vce_v4_0_hw_fini(void *handle) } for (i = 0; i < adev->vce.num_rings; i++) - adev->vce.ring[i].ready = false; + adev->vce.ring[i].sched.ready = false; return 0; } @@ -951,9 +946,11 @@ static int vce_v4_0_set_powergating_state(void *handle, } #endif -static void vce_v4_0_ring_emit_ib(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch) +static void vce_v4_0_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job, + struct amdgpu_ib *ib, bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + amdgpu_ring_write(ring, VCE_CMD_IB_VM); amdgpu_ring_write(ring, vmid); amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index eae90922fdbe..c1a03505f956 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -176,30 +176,22 @@ static int vcn_v1_0_hw_init(void *handle) struct amdgpu_ring *ring = &adev->vcn.ring_dec; int i, r; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } for (i = 0; i < adev->vcn.num_enc_rings; ++i) { ring = &adev->vcn.ring_enc[i]; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + ring->sched.ready = true; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } } ring = &adev->vcn.ring_jpeg; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; + r = amdgpu_ring_test_helper(ring); + if (r) goto done; - } done: if (!r) @@ -224,7 +216,7 @@ static int vcn_v1_0_hw_fini(void *handle) if (RREG32_SOC15(VCN, 0, mmUVD_STATUS)) vcn_v1_0_stop(adev); - ring->ready = false; + ring->sched.ready = false; return 0; } @@ -1366,10 +1358,12 @@ static void vcn_v1_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 * Write ring commands to execute the indirect buffer */ static void vcn_v1_0_dec_ring_emit_ib(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { struct amdgpu_device *adev = ring->adev; + unsigned vmid = AMDGPU_JOB_GET_VMID(job); amdgpu_ring_write(ring, PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_IB_VMID), 0)); @@ -1524,8 +1518,12 @@ static void vcn_v1_0_enc_ring_insert_end(struct amdgpu_ring *ring) * Write enc ring commands to execute the indirect buffer */ static void vcn_v1_0_enc_ring_emit_ib(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, unsigned int vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { + unsigned vmid = AMDGPU_JOB_GET_VMID(job); + amdgpu_ring_write(ring, VCN_ENC_CMD_IB); amdgpu_ring_write(ring, vmid); amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); @@ -1725,10 +1723,12 @@ static void vcn_v1_0_jpeg_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u6 * Write ring commands to execute the indirect buffer. */ static void vcn_v1_0_jpeg_ring_emit_ib(struct amdgpu_ring *ring, - struct amdgpu_ib *ib, - unsigned vmid, bool ctx_switch) + struct amdgpu_job *job, + struct amdgpu_ib *ib, + bool ctx_switch) { struct amdgpu_device *adev = ring->adev; + unsigned vmid = AMDGPU_JOB_GET_VMID(job); amdgpu_ring_write(ring, PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_JRBC_IB_VMID), 0, 0, PACKETJ_TYPE0)); diff --git a/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c b/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c index 5d2475d5392c..177d1e5329a5 100644 --- a/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c +++ b/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c @@ -23,6 +23,7 @@ #include "kfd_priv.h" #include "kfd_events.h" #include "cik_int.h" +#include "amdgpu_amdkfd.h" static bool cik_event_interrupt_isr(struct kfd_dev *dev, const uint32_t *ih_ring_entry, @@ -107,7 +108,7 @@ static void cik_event_interrupt_wq(struct kfd_dev *dev, kfd_process_vm_fault(dev->dqm, pasid); memset(&info, 0, sizeof(info)); - dev->kfd2kgd->get_vm_fault_info(dev->kgd, &info); + amdgpu_amdkfd_gpuvm_get_vm_fault_info(dev->kgd, &info); if (!info.page_addr && !info.status) return; diff --git a/drivers/gpu/drm/amd/amdkfd/cik_regs.h b/drivers/gpu/drm/amd/amdkfd/cik_regs.h index 37ce6dd65391..8e2a1663c4db 100644 --- a/drivers/gpu/drm/amd/amdkfd/cik_regs.h +++ b/drivers/gpu/drm/amd/amdkfd/cik_regs.h @@ -68,6 +68,4 @@ #define GRBM_GFX_INDEX 0x30800 -#define ATC_VMID_PASID_MAPPING_VALID (1U << 31) - #endif diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index 14d5b5fa822d..5f4062b41add 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -37,6 +37,7 @@ #include "kfd_priv.h" #include "kfd_device_queue_manager.h" #include "kfd_dbgmgr.h" +#include "amdgpu_amdkfd.h" static long kfd_ioctl(struct file *, unsigned int, unsigned long); static int kfd_open(struct inode *, struct file *); @@ -834,8 +835,7 @@ static int kfd_ioctl_get_clock_counters(struct file *filep, dev = kfd_device_by_id(args->gpu_id); if (dev) /* Reading GPU clock counter from KGD */ - args->gpu_clock_counter = - dev->kfd2kgd->get_gpu_clock_counter(dev->kgd); + args->gpu_clock_counter = amdgpu_amdkfd_get_gpu_clock_counter(dev->kgd); else /* Node without GPU resource */ args->gpu_clock_counter = 0; @@ -1042,7 +1042,7 @@ static int kfd_ioctl_create_event(struct file *filp, struct kfd_process *p, } mutex_unlock(&p->mutex); - err = kfd->kfd2kgd->map_gtt_bo_to_kernel(kfd->kgd, + err = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(kfd->kgd, mem, &kern_addr, &size); if (err) { pr_err("Failed to map event page to kernel\n"); @@ -1240,7 +1240,7 @@ bool kfd_dev_is_large_bar(struct kfd_dev *dev) if (dev->device_info->needs_iommu_device) return false; - dev->kfd2kgd->get_local_mem_info(dev->kgd, &mem_info); + amdgpu_amdkfd_get_local_mem_info(dev->kgd, &mem_info); if (mem_info.local_mem_size_private == 0 && mem_info.local_mem_size_public > 0) return true; @@ -1281,7 +1281,7 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep, goto err_unlock; } - err = dev->kfd2kgd->alloc_memory_of_gpu( + err = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu( dev->kgd, args->va_addr, args->size, pdd->vm, (struct kgd_mem **) &mem, &offset, flags); @@ -1303,7 +1303,7 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep, return 0; err_free: - dev->kfd2kgd->free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem); + amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem); err_unlock: mutex_unlock(&p->mutex); return err; @@ -1338,7 +1338,8 @@ static int kfd_ioctl_free_memory_of_gpu(struct file *filep, goto err_unlock; } - ret = dev->kfd2kgd->free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem); + ret = amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, + (struct kgd_mem *)mem); /* If freeing the buffer failed, leave the handle in place for * clean-up during process tear-down. @@ -1418,7 +1419,7 @@ static int kfd_ioctl_map_memory_to_gpu(struct file *filep, err = PTR_ERR(peer_pdd); goto get_mem_obj_from_handle_failed; } - err = peer->kfd2kgd->map_memory_to_gpu( + err = amdgpu_amdkfd_gpuvm_map_memory_to_gpu( peer->kgd, (struct kgd_mem *)mem, peer_pdd->vm); if (err) { pr_err("Failed to map to gpu %d/%d\n", @@ -1430,7 +1431,7 @@ static int kfd_ioctl_map_memory_to_gpu(struct file *filep, mutex_unlock(&p->mutex); - err = dev->kfd2kgd->sync_memory(dev->kgd, (struct kgd_mem *) mem, true); + err = amdgpu_amdkfd_gpuvm_sync_memory(dev->kgd, (struct kgd_mem *) mem, true); if (err) { pr_debug("Sync memory failed, wait interrupted by user signal\n"); goto sync_memory_failed; @@ -1525,7 +1526,7 @@ static int kfd_ioctl_unmap_memory_from_gpu(struct file *filep, err = -ENODEV; goto get_mem_obj_from_handle_failed; } - err = dev->kfd2kgd->unmap_memory_to_gpu( + err = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu( peer->kgd, (struct kgd_mem *)mem, peer_pdd->vm); if (err) { pr_err("Failed to unmap from gpu %d/%d\n", diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c index 56412b0e7e1c..3783d122f283 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c @@ -26,6 +26,7 @@ #include "kfd_priv.h" #include "kfd_topology.h" #include "kfd_iommu.h" +#include "amdgpu_amdkfd.h" /* GPU Processor ID base for dGPUs for which VCRAT needs to be created. * GPU processor ID are expressed with Bit[31]=1. @@ -753,12 +754,10 @@ int kfd_create_crat_image_acpi(void **crat_image, size_t *size) return -ENODATA; } - pcrat_image = kmalloc(crat_table->length, GFP_KERNEL); + pcrat_image = kmemdup(crat_table, crat_table->length, GFP_KERNEL); if (!pcrat_image) return -ENOMEM; - memcpy(pcrat_image, crat_table, crat_table->length); - *crat_image = pcrat_image; *size = crat_table->length; @@ -1161,7 +1160,7 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image, cu->flags |= CRAT_CU_FLAGS_GPU_PRESENT; cu->proximity_domain = proximity_domain; - kdev->kfd2kgd->get_cu_info(kdev->kgd, &cu_info); + amdgpu_amdkfd_get_cu_info(kdev->kgd, &cu_info); cu->num_simd_per_cu = cu_info.simd_per_cu; cu->num_simd_cores = cu_info.simd_per_cu * cu_info.cu_active_number; cu->max_waves_simd = cu_info.max_waves_per_simd; @@ -1192,7 +1191,7 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image, * report the total FB size (public+private) as a single * private heap. */ - kdev->kfd2kgd->get_local_mem_info(kdev->kgd, &local_mem_info); + amdgpu_amdkfd_get_local_mem_info(kdev->kgd, &local_mem_info); sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr + sub_type_hdr->length); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c index a9f18ea7e354..c004647c8cb4 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c @@ -28,6 +28,7 @@ #include "kfd_pm4_headers_vi.h" #include "cwsr_trap_handler.h" #include "kfd_iommu.h" +#include "amdgpu_amdkfd.h" #define MQD_SIZE_ALIGNED 768 @@ -478,7 +479,7 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd, /* add another 512KB for all other allocations on gart (HPD, fences) */ size += 512 * 1024; - if (kfd->kfd2kgd->init_gtt_mem_allocation( + if (amdgpu_amdkfd_alloc_gtt_mem( kfd->kgd, size, &kfd->gtt_mem, &kfd->gtt_start_gpu_addr, &kfd->gtt_start_cpu_ptr, false)) { @@ -552,7 +553,7 @@ kfd_topology_add_device_error: kfd_doorbell_error: kfd_gtt_sa_fini(kfd); kfd_gtt_sa_init_error: - kfd->kfd2kgd->free_gtt_mem(kfd->kgd, kfd->gtt_mem); + amdgpu_amdkfd_free_gtt_mem(kfd->kgd, kfd->gtt_mem); dev_err(kfd_device, "device %x:%x NOT added due to errors\n", kfd->pdev->vendor, kfd->pdev->device); @@ -569,7 +570,7 @@ void kgd2kfd_device_exit(struct kfd_dev *kfd) kfd_topology_remove_device(kfd); kfd_doorbell_fini(kfd); kfd_gtt_sa_fini(kfd); - kfd->kfd2kgd->free_gtt_mem(kfd->kgd, kfd->gtt_mem); + amdgpu_amdkfd_free_gtt_mem(kfd->kgd, kfd->gtt_mem); } kfree(kfd); @@ -681,6 +682,7 @@ void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry) { uint32_t patched_ihre[KFD_MAX_RING_ENTRY_SIZE]; bool is_patched = false; + unsigned long flags; if (!kfd->init_complete) return; @@ -690,7 +692,7 @@ void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry) return; } - spin_lock(&kfd->interrupt_lock); + spin_lock_irqsave(&kfd->interrupt_lock, flags); if (kfd->interrupts_active && interrupt_is_wanted(kfd, ih_ring_entry, @@ -699,7 +701,7 @@ void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry) is_patched ? patched_ihre : ih_ring_entry)) queue_work(kfd->ih_wq, &kfd->interrupt_work); - spin_unlock(&kfd->interrupt_lock); + spin_unlock_irqrestore(&kfd->interrupt_lock, flags); } int kgd2kfd_quiesce_mm(struct mm_struct *mm) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c index a3b933967171..fb9d66ea13b7 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -33,6 +33,7 @@ #include "kfd_mqd_manager.h" #include "cik_regs.h" #include "kfd_kernel_queue.h" +#include "amdgpu_amdkfd.h" /* Size of the per-pipe EOP queue */ #define CIK_HPD_EOP_BYTES_LOG2 11 @@ -219,7 +220,7 @@ static int flush_texture_cache_nocpsch(struct kfd_dev *kdev, if (ret) return ret; - return kdev->kfd2kgd->submit_ib(kdev->kgd, KGD_ENGINE_MEC1, qpd->vmid, + return amdgpu_amdkfd_submit_ib(kdev->kgd, KGD_ENGINE_MEC1, qpd->vmid, qpd->ib_base, (uint32_t *)qpd->ib_kaddr, pmf->release_mem_size / sizeof(uint32_t)); } @@ -672,7 +673,7 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm, pdd = qpd_to_pdd(qpd); /* Retrieve PD base */ - pd_base = dqm->dev->kfd2kgd->get_process_page_dir(pdd->vm); + pd_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->vm); dqm_lock(dqm); if (WARN_ON_ONCE(!qpd->evicted)) /* already restored, do nothing */ @@ -743,7 +744,7 @@ static int restore_process_queues_cpsch(struct device_queue_manager *dqm, pdd = qpd_to_pdd(qpd); /* Retrieve PD base */ - pd_base = dqm->dev->kfd2kgd->get_process_page_dir(pdd->vm); + pd_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->vm); dqm_lock(dqm); if (WARN_ON_ONCE(!qpd->evicted)) /* already restored, do nothing */ @@ -793,7 +794,7 @@ static int register_process(struct device_queue_manager *dqm, pdd = qpd_to_pdd(qpd); /* Retrieve PD base */ - pd_base = dqm->dev->kfd2kgd->get_process_page_dir(pdd->vm); + pd_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->vm); dqm_lock(dqm); list_add(&n->list, &dqm->queues); @@ -805,7 +806,7 @@ static int register_process(struct device_queue_manager *dqm, retval = dqm->asic_ops.update_qpd(dqm, qpd); if (dqm->processes_count++ == 0) - dqm->dev->kfd2kgd->set_compute_idle(dqm->dev->kgd, false); + amdgpu_amdkfd_set_compute_idle(dqm->dev->kgd, false); dqm_unlock(dqm); @@ -829,7 +830,7 @@ static int unregister_process(struct device_queue_manager *dqm, list_del(&cur->list); kfree(cur); if (--dqm->processes_count == 0) - dqm->dev->kfd2kgd->set_compute_idle( + amdgpu_amdkfd_set_compute_idle( dqm->dev->kgd, true); goto out; } @@ -845,15 +846,8 @@ static int set_pasid_vmid_mapping(struct device_queue_manager *dqm, unsigned int pasid, unsigned int vmid) { - uint32_t pasid_mapping; - - pasid_mapping = (pasid == 0) ? 0 : - (uint32_t)pasid | - ATC_VMID_PASID_MAPPING_VALID; - return dqm->dev->kfd2kgd->set_pasid_vmid_mapping( - dqm->dev->kgd, pasid_mapping, - vmid); + dqm->dev->kgd, pasid, vmid); } static void init_interrupts(struct device_queue_manager *dqm) @@ -1796,7 +1790,7 @@ static void kfd_process_hw_exception(struct work_struct *work) { struct device_queue_manager *dqm = container_of(work, struct device_queue_manager, hw_exception_work); - dqm->dev->kfd2kgd->gpu_recover(dqm->dev->kgd); + amdgpu_amdkfd_gpu_reset(dqm->dev->kgd); } #if defined(CONFIG_DEBUG_FS) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c index e33019a7a883..6910028010d6 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c @@ -22,6 +22,7 @@ */ #include "kfd_mqd_manager.h" +#include "amdgpu_amdkfd.h" struct mqd_manager *mqd_manager_init(enum KFD_MQD_TYPE type, struct kfd_dev *dev) @@ -58,7 +59,7 @@ void mqd_symmetrically_map_cu_mask(struct mqd_manager *mm, uint32_t cu_per_sh[4] = {0}; int i, se, cu = 0; - mm->dev->kfd2kgd->get_cu_info(mm->dev->kgd, &cu_info); + amdgpu_amdkfd_get_cu_info(mm->dev->kgd, &cu_info); if (cu_mask_count > cu_info.cu_active_number) cu_mask_count = cu_info.cu_active_number; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c index f381c1cb27bd..9dbba609450e 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c @@ -30,6 +30,7 @@ #include "gc/gc_9_0_offset.h" #include "gc/gc_9_0_sh_mask.h" #include "sdma0/sdma0_4_0_sh_mask.h" +#include "amdgpu_amdkfd.h" static inline struct v9_mqd *get_mqd(void *mqd) { @@ -83,7 +84,7 @@ static int init_mqd(struct mqd_manager *mm, void **mqd, *mqd_mem_obj = kzalloc(sizeof(struct kfd_mem_obj), GFP_KERNEL); if (!*mqd_mem_obj) return -ENOMEM; - retval = kfd->kfd2kgd->init_gtt_mem_allocation(kfd->kgd, + retval = amdgpu_amdkfd_alloc_gtt_mem(kfd->kgd, ALIGN(q->ctl_stack_size, PAGE_SIZE) + ALIGN(sizeof(struct v9_mqd), PAGE_SIZE), &((*mqd_mem_obj)->gtt_mem), @@ -250,7 +251,7 @@ static void uninit_mqd(struct mqd_manager *mm, void *mqd, struct kfd_dev *kfd = mm->dev; if (mqd_mem_obj->gtt_mem) { - kfd->kfd2kgd->free_gtt_mem(kfd->kgd, mqd_mem_obj->gtt_mem); + amdgpu_amdkfd_free_gtt_mem(kfd->kgd, mqd_mem_obj->gtt_mem); kfree(mqd_mem_obj); } else { kfd_gtt_sa_free(mm->dev, mqd_mem_obj); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pasid.c b/drivers/gpu/drm/amd/amdkfd/kfd_pasid.c index 15fff4420e53..33b08ff00b50 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_pasid.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_pasid.c @@ -22,6 +22,7 @@ #include <linux/types.h> #include "kfd_priv.h" +#include "amdgpu_ids.h" static unsigned int pasid_bits = 16; static const struct kfd2kgd_calls *kfd2kgd; @@ -71,7 +72,7 @@ unsigned int kfd_pasid_alloc(void) return false; } - r = kfd2kgd->alloc_pasid(pasid_bits); + r = amdgpu_pasid_alloc(pasid_bits); return r > 0 ? r : 0; } @@ -79,5 +80,5 @@ unsigned int kfd_pasid_alloc(void) void kfd_pasid_free(unsigned int pasid) { if (kfd2kgd) - kfd2kgd->free_pasid(pasid); + amdgpu_pasid_free(pasid); } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index 53ff86d45d91..dec8e64f36bd 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -507,6 +507,7 @@ struct qcm_process_device { * All the memory management data should be here too */ uint64_t gds_context_area; + /* Contains page table flags such as AMDGPU_PTE_VALID since gfx9 */ uint64_t page_table_base; uint32_t sh_mem_config; uint32_t sh_mem_bases; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index 0039e451d9af..80b36e860a0a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -31,6 +31,7 @@ #include <linux/compat.h> #include <linux/mman.h> #include <linux/file.h> +#include "amdgpu_amdkfd.h" struct mm_struct; @@ -100,8 +101,8 @@ static void kfd_process_free_gpuvm(struct kgd_mem *mem, { struct kfd_dev *dev = pdd->dev; - dev->kfd2kgd->unmap_memory_to_gpu(dev->kgd, mem, pdd->vm); - dev->kfd2kgd->free_memory_of_gpu(dev->kgd, mem); + amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(dev->kgd, mem, pdd->vm); + amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, mem); } /* kfd_process_alloc_gpuvm - Allocate GPU VM for the KFD process @@ -119,16 +120,16 @@ static int kfd_process_alloc_gpuvm(struct kfd_process_device *pdd, int handle; int err; - err = kdev->kfd2kgd->alloc_memory_of_gpu(kdev->kgd, gpu_va, size, + err = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(kdev->kgd, gpu_va, size, pdd->vm, &mem, NULL, flags); if (err) goto err_alloc_mem; - err = kdev->kfd2kgd->map_memory_to_gpu(kdev->kgd, mem, pdd->vm); + err = amdgpu_amdkfd_gpuvm_map_memory_to_gpu(kdev->kgd, mem, pdd->vm); if (err) goto err_map_mem; - err = kdev->kfd2kgd->sync_memory(kdev->kgd, mem, true); + err = amdgpu_amdkfd_gpuvm_sync_memory(kdev->kgd, mem, true); if (err) { pr_debug("Sync memory failed, wait interrupted by user signal\n"); goto sync_memory_failed; @@ -147,7 +148,7 @@ static int kfd_process_alloc_gpuvm(struct kfd_process_device *pdd, } if (kptr) { - err = kdev->kfd2kgd->map_gtt_bo_to_kernel(kdev->kgd, + err = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(kdev->kgd, (struct kgd_mem *)mem, kptr, NULL); if (err) { pr_debug("Map GTT BO to kernel failed\n"); @@ -165,7 +166,7 @@ sync_memory_failed: return err; err_map_mem: - kdev->kfd2kgd->free_memory_of_gpu(kdev->kgd, mem); + amdgpu_amdkfd_gpuvm_free_memory_of_gpu(kdev->kgd, mem); err_alloc_mem: *kptr = NULL; return err; @@ -296,11 +297,11 @@ static void kfd_process_device_free_bos(struct kfd_process_device *pdd) per_device_list) { if (!peer_pdd->vm) continue; - peer_pdd->dev->kfd2kgd->unmap_memory_to_gpu( + amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu( peer_pdd->dev->kgd, mem, peer_pdd->vm); } - pdd->dev->kfd2kgd->free_memory_of_gpu(pdd->dev->kgd, mem); + amdgpu_amdkfd_gpuvm_free_memory_of_gpu(pdd->dev->kgd, mem); kfd_process_device_remove_obj_handle(pdd, id); } } @@ -323,11 +324,12 @@ static void kfd_process_destroy_pdds(struct kfd_process *p) pdd->dev->id, p->pasid); if (pdd->drm_file) { - pdd->dev->kfd2kgd->release_process_vm(pdd->dev->kgd, pdd->vm); + amdgpu_amdkfd_gpuvm_release_process_vm( + pdd->dev->kgd, pdd->vm); fput(pdd->drm_file); } else if (pdd->vm) - pdd->dev->kfd2kgd->destroy_process_vm( + amdgpu_amdkfd_gpuvm_destroy_process_vm( pdd->dev->kgd, pdd->vm); list_del(&pdd->per_device_list); @@ -688,12 +690,12 @@ int kfd_process_device_init_vm(struct kfd_process_device *pdd, dev = pdd->dev; if (drm_file) - ret = dev->kfd2kgd->acquire_process_vm( + ret = amdgpu_amdkfd_gpuvm_acquire_process_vm( dev->kgd, drm_file, p->pasid, &pdd->vm, &p->kgd_process_info, &p->ef); else - ret = dev->kfd2kgd->create_process_vm( - dev->kgd, p->pasid, &pdd->vm, &p->kgd_process_info, &p->ef); + ret = amdgpu_amdkfd_gpuvm_create_process_vm(dev->kgd, p->pasid, + &pdd->vm, &p->kgd_process_info, &p->ef); if (ret) { pr_err("Failed to create process VM object\n"); return ret; @@ -714,7 +716,7 @@ err_init_cwsr: err_reserve_ib_mem: kfd_process_device_free_bos(pdd); if (!drm_file) - dev->kfd2kgd->destroy_process_vm(dev->kgd, pdd->vm); + amdgpu_amdkfd_gpuvm_destroy_process_vm(dev->kgd, pdd->vm); pdd->vm = NULL; return ret; @@ -972,7 +974,7 @@ static void restore_process_worker(struct work_struct *work) */ p->last_restore_timestamp = get_jiffies_64(); - ret = pdd->dev->kfd2kgd->restore_process_bos(p->kgd_process_info, + ret = amdgpu_amdkfd_gpuvm_restore_process_bos(p->kgd_process_info, &p->ef); if (ret) { pr_debug("Failed to restore BOs of pasid %d, retry after %d ms\n", diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index e3843c5929ed..c73b4ff61f99 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -36,6 +36,7 @@ #include "kfd_topology.h" #include "kfd_device_queue_manager.h" #include "kfd_iommu.h" +#include "amdgpu_amdkfd.h" /* topology_device_list - Master list of all topology devices */ static struct list_head topology_device_list; @@ -1052,7 +1053,7 @@ static uint32_t kfd_generate_gpu_id(struct kfd_dev *gpu) if (!gpu) return 0; - gpu->kfd2kgd->get_local_mem_info(gpu->kgd, &local_mem_info); + amdgpu_amdkfd_get_local_mem_info(gpu->kgd, &local_mem_info); local_mem_size = local_mem_info.local_mem_size_private + local_mem_info.local_mem_size_public; @@ -1118,8 +1119,7 @@ static void kfd_fill_mem_clk_max_info(struct kfd_topology_device *dev) * for APUs - If CRAT from ACPI reports more than one bank, then * all the banks will report the same mem_clk_max information */ - dev->gpu->kfd2kgd->get_local_mem_info(dev->gpu->kgd, - &local_mem_info); + amdgpu_amdkfd_get_local_mem_info(dev->gpu->kgd, &local_mem_info); list_for_each_entry(mem, &dev->mem_props, list) mem->mem_clk_max = local_mem_info.mem_clk_max; @@ -1240,7 +1240,7 @@ int kfd_topology_add_device(struct kfd_dev *gpu) * needed for the topology */ - dev->gpu->kfd2kgd->get_cu_info(dev->gpu->kgd, &cu_info); + amdgpu_amdkfd_get_cu_info(dev->gpu->kgd, &cu_info); dev->node_props.simd_arrays_per_engine = cu_info.num_shader_arrays_per_engine; @@ -1249,7 +1249,7 @@ int kfd_topology_add_device(struct kfd_dev *gpu) dev->node_props.location_id = PCI_DEVID(gpu->pdev->bus->number, gpu->pdev->devfn); dev->node_props.max_engine_clk_fcompute = - dev->gpu->kfd2kgd->get_max_engine_clock_in_mhz(dev->gpu->kgd); + amdgpu_amdkfd_get_max_engine_clock_in_mhz(dev->gpu->kgd); dev->node_props.max_engine_clk_ccompute = cpufreq_quick_get_max(0) / 1000; dev->node_props.drm_render_minor = diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 5064768642f3..aa43bb253ea2 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -76,6 +76,16 @@ #define FIRMWARE_RAVEN_DMCU "amdgpu/raven_dmcu.bin" MODULE_FIRMWARE(FIRMWARE_RAVEN_DMCU); +/** + * DOC: overview + * + * The AMDgpu display manager, **amdgpu_dm** (or even simpler, + * **dm**) sits between DRM and DC. It acts as a liason, converting DRM + * requests into DC requests, and DC responses into DRM responses. + * + * The root control structure is &struct amdgpu_display_manager. + */ + /* basic init/fini API */ static int amdgpu_dm_init(struct amdgpu_device *adev); static void amdgpu_dm_fini(struct amdgpu_device *adev); @@ -95,7 +105,7 @@ static void amdgpu_dm_update_connector_after_detect(struct amdgpu_dm_connector *aconnector); static int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, - struct amdgpu_plane *aplane, + struct drm_plane *plane, unsigned long possible_crtcs); static int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, struct drm_plane *plane, @@ -379,11 +389,6 @@ static void amdgpu_dm_fbc_init(struct drm_connector *connector) } -/* - * Init display KMS - * - * Returns 0 on success - */ static int amdgpu_dm_init(struct amdgpu_device *adev) { struct dc_init_data init_data; @@ -663,6 +668,26 @@ static void s3_handle_mst(struct drm_device *dev, bool suspend) drm_modeset_unlock(&dev->mode_config.connection_mutex); } +/** + * dm_hw_init() - Initialize DC device + * @handle: The base driver device containing the amdpgu_dm device. + * + * Initialize the &struct amdgpu_display_manager device. This involves calling + * the initializers of each DM component, then populating the struct with them. + * + * Although the function implies hardware initialization, both hardware and + * software are initialized here. Splitting them out to their relevant init + * hooks is a future TODO item. + * + * Some notable things that are initialized here: + * + * - Display Core, both software and hardware + * - DC modules that we need (freesync and color management) + * - DRM software states + * - Interrupt sources and handlers + * - Vblank support + * - Debug FS entries, if enabled + */ static int dm_hw_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -673,6 +698,14 @@ static int dm_hw_init(void *handle) return 0; } +/** + * dm_hw_fini() - Teardown DC device + * @handle: The base driver device containing the amdpgu_dm device. + * + * Teardown components within &struct amdgpu_display_manager that require + * cleanup. This involves cleaning up the DRM device, DC, and any modules that + * were loaded. Also flush IRQ workqueues and disable them. + */ static int dm_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -898,6 +931,16 @@ static int dm_resume(void *handle) return ret; } +/** + * DOC: DM Lifecycle + * + * DM (and consequently DC) is registered in the amdgpu base driver as a IP + * block. When CONFIG_DRM_AMD_DC is enabled, the DM device IP block is added to + * the base driver's device list to be initialized and torn down accordingly. + * + * The functions to do so are provided as hooks in &struct amd_ip_funcs. + */ + static const struct amd_ip_funcs amdgpu_dm_funcs = { .name = "dm", .early_init = dm_early_init, @@ -965,6 +1008,12 @@ dm_atomic_state_alloc_free(struct drm_atomic_state *state) kfree(dm_state); } +/** + * DOC: atomic + * + * *WIP* + */ + static const struct drm_mode_config_funcs amdgpu_dm_mode_funcs = { .fb_create = amdgpu_display_user_framebuffer_create, .output_poll_changed = drm_fb_helper_output_poll_changed, @@ -1527,8 +1576,23 @@ static int amdgpu_dm_backlight_update_status(struct backlight_device *bd) { struct amdgpu_display_manager *dm = bl_get_data(bd); + /* backlight_pwm_u16_16 parameter is in unsigned 32 bit, 16 bit integer + * and 16 bit fractional, where 1.0 is max backlight value. + * bd->props.brightness is 8 bit format and needs to be converted by + * scaling via copy lower byte to upper byte of 16 bit value. + */ + uint32_t brightness = bd->props.brightness * 0x101; + + /* + * PWM interperts 0 as 100% rather than 0% because of HW + * limitation for level 0. So limiting minimum brightness level + * to 1. + */ + if (bd->props.brightness < 1) + brightness = 0x101; + if (dc_link_set_backlight_level(dm->backlight_link, - bd->props.brightness, 0, 0)) + brightness, 0, 0)) return 0; else return 1; @@ -1580,18 +1644,18 @@ static int initialize_plane(struct amdgpu_display_manager *dm, struct amdgpu_mode_info *mode_info, int plane_id) { - struct amdgpu_plane *plane; + struct drm_plane *plane; unsigned long possible_crtcs; int ret = 0; - plane = kzalloc(sizeof(struct amdgpu_plane), GFP_KERNEL); + plane = kzalloc(sizeof(struct drm_plane), GFP_KERNEL); mode_info->planes[plane_id] = plane; if (!plane) { DRM_ERROR("KMS: Failed to allocate plane\n"); return -ENOMEM; } - plane->base.type = mode_info->plane_type[plane_id]; + plane->type = mode_info->plane_type[plane_id]; /* * HACK: IGT tests expect that each plane can only have @@ -1682,7 +1746,7 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev) } for (i = 0; i < dm->dc->caps.max_streams; i++) - if (amdgpu_dm_crtc_init(dm, &mode_info->planes[i]->base, i)) { + if (amdgpu_dm_crtc_init(dm, mode_info->planes[i], i)) { DRM_ERROR("KMS: Failed to initialize crtc\n"); goto fail; } @@ -3457,49 +3521,49 @@ static const u32 cursor_formats[] = { }; static int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, - struct amdgpu_plane *aplane, + struct drm_plane *plane, unsigned long possible_crtcs) { int res = -EPERM; - switch (aplane->base.type) { + switch (plane->type) { case DRM_PLANE_TYPE_PRIMARY: res = drm_universal_plane_init( dm->adev->ddev, - &aplane->base, + plane, possible_crtcs, &dm_plane_funcs, rgb_formats, ARRAY_SIZE(rgb_formats), - NULL, aplane->base.type, NULL); + NULL, plane->type, NULL); break; case DRM_PLANE_TYPE_OVERLAY: res = drm_universal_plane_init( dm->adev->ddev, - &aplane->base, + plane, possible_crtcs, &dm_plane_funcs, yuv_formats, ARRAY_SIZE(yuv_formats), - NULL, aplane->base.type, NULL); + NULL, plane->type, NULL); break; case DRM_PLANE_TYPE_CURSOR: res = drm_universal_plane_init( dm->adev->ddev, - &aplane->base, + plane, possible_crtcs, &dm_plane_funcs, cursor_formats, ARRAY_SIZE(cursor_formats), - NULL, aplane->base.type, NULL); + NULL, plane->type, NULL); break; } - drm_plane_helper_add(&aplane->base, &dm_plane_helper_funcs); + drm_plane_helper_add(plane, &dm_plane_helper_funcs); /* Create (reset) the plane state */ - if (aplane->base.funcs->reset) - aplane->base.funcs->reset(&aplane->base); + if (plane->funcs->reset) + plane->funcs->reset(plane); return res; @@ -3510,7 +3574,7 @@ static int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, uint32_t crtc_index) { struct amdgpu_crtc *acrtc = NULL; - struct amdgpu_plane *cursor_plane; + struct drm_plane *cursor_plane; int res = -ENOMEM; @@ -3518,7 +3582,7 @@ static int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, if (!cursor_plane) goto fail; - cursor_plane->base.type = DRM_PLANE_TYPE_CURSOR; + cursor_plane->type = DRM_PLANE_TYPE_CURSOR; res = amdgpu_dm_plane_init(dm, cursor_plane, 0); acrtc = kzalloc(sizeof(struct amdgpu_crtc), GFP_KERNEL); @@ -3529,7 +3593,7 @@ static int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, dm->ddev, &acrtc->base, plane, - &cursor_plane->base, + cursor_plane, &amdgpu_dm_crtc_funcs, NULL); if (res) @@ -3768,12 +3832,12 @@ void amdgpu_dm_connector_init_helper(struct amdgpu_display_manager *dm, case DRM_MODE_CONNECTOR_HDMIA: aconnector->base.polled = DRM_CONNECTOR_POLL_HPD; aconnector->base.ycbcr_420_allowed = - link->link_enc->features.ycbcr420_supported ? true : false; + link->link_enc->features.hdmi_ycbcr420_supported ? true : false; break; case DRM_MODE_CONNECTOR_DisplayPort: aconnector->base.polled = DRM_CONNECTOR_POLL_HPD; aconnector->base.ycbcr_420_allowed = - link->link_enc->features.ycbcr420_supported ? true : false; + link->link_enc->features.dp_ycbcr420_supported ? true : false; break; case DRM_MODE_CONNECTOR_DVID: aconnector->base.polled = DRM_CONNECTOR_POLL_HPD; @@ -4531,6 +4595,14 @@ static int amdgpu_dm_atomic_commit(struct drm_device *dev, /*TODO Handle EINTR, reenable IRQ*/ } +/** + * amdgpu_dm_atomic_commit_tail() - AMDgpu DM's commit tail implementation. + * @state: The atomic state to commit + * + * This will tell DC to commit the constructed DC state from atomic_check, + * programming the hardware. Any failures here implies a hardware failure, since + * atomic check should have filtered anything non-kosher. + */ static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state) { struct drm_device *dev = state->dev; @@ -5302,6 +5374,12 @@ enum surface_update_type dm_determine_update_type_for_commit(struct dc *dc, stru struct dc_stream_update stream_update; enum surface_update_type update_type = UPDATE_TYPE_FAST; + if (!updates || !surface) { + DRM_ERROR("Plane or surface update failed to allocate"); + /* Set type to FULL to avoid crashing in DC*/ + update_type = UPDATE_TYPE_FULL; + goto ret; + } for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state, new_crtc_state, i) { new_dm_crtc_state = to_dm_crtc_state(new_crtc_state); @@ -5377,6 +5455,31 @@ ret: return update_type; } +/** + * amdgpu_dm_atomic_check() - Atomic check implementation for AMDgpu DM. + * @dev: The DRM device + * @state: The atomic state to commit + * + * Validate that the given atomic state is programmable by DC into hardware. + * This involves constructing a &struct dc_state reflecting the new hardware + * state we wish to commit, then querying DC to see if it is programmable. It's + * important not to modify the existing DC state. Otherwise, atomic_check + * may unexpectedly commit hardware changes. + * + * When validating the DC state, it's important that the right locks are + * acquired. For full updates case which removes/adds/updates streams on one + * CRTC while flipping on another CRTC, acquiring global lock will guarantee + * that any such full update commit will wait for completion of any outstanding + * flip using DRMs synchronization events. See + * dm_determine_update_type_for_commit() + * + * Note that DM adds the affected connectors for all CRTCs in state, when that + * might not seem necessary. This is because DC stream creation requires the + * DC sink, which is tied to the DRM connector state. Cleaning this up should + * be possible but non-trivial - a possible TODO item. + * + * Return: -Error code if validation failed. + */ static int amdgpu_dm_atomic_check(struct drm_device *dev, struct drm_atomic_state *state) { @@ -5479,15 +5582,6 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, lock_and_validation_needed = true; } - /* - * For full updates case when - * removing/adding/updating streams on one CRTC while flipping - * on another CRTC, - * acquiring global lock will guarantee that any such full - * update commit - * will wait for completion of any outstanding flip using DRMs - * synchronization events. - */ update_type = dm_determine_update_type_for_commit(dc, state); if (overall_update_type < update_type) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h index 924a38a1fc44..d6960644d714 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h @@ -59,49 +59,100 @@ struct common_irq_params { enum dc_irq_source irq_src; }; +/** + * struct irq_list_head - Linked-list for low context IRQ handlers. + * + * @head: The list_head within &struct handler_data + * @work: A work_struct containing the deferred handler work + */ struct irq_list_head { struct list_head head; /* In case this interrupt needs post-processing, 'work' will be queued*/ struct work_struct work; }; +/** + * struct dm_compressor_info - Buffer info used by frame buffer compression + * @cpu_addr: MMIO cpu addr + * @bo_ptr: Pointer to the buffer object + * @gpu_addr: MMIO gpu addr + */ struct dm_comressor_info { void *cpu_addr; struct amdgpu_bo *bo_ptr; uint64_t gpu_addr; }; +/** + * struct amdgpu_display_manager - Central amdgpu display manager device + * + * @dc: Display Core control structure + * @adev: AMDGPU base driver structure + * @ddev: DRM base driver structure + * @display_indexes_num: Max number of display streams supported + * @irq_handler_list_table_lock: Synchronizes access to IRQ tables + * @backlight_dev: Backlight control device + * @cached_state: Caches device atomic state for suspend/resume + * @compressor: Frame buffer compression buffer. See &struct dm_comressor_info + */ struct amdgpu_display_manager { + struct dc *dc; + + /** + * @cgs_device: + * + * The Common Graphics Services device. It provides an interface for + * accessing registers. + */ struct cgs_device *cgs_device; - struct amdgpu_device *adev; /*AMD base driver*/ - struct drm_device *ddev; /*DRM base driver*/ + struct amdgpu_device *adev; + struct drm_device *ddev; u16 display_indexes_num; - /* - * 'irq_source_handler_table' holds a list of handlers - * per (DAL) IRQ source. + /** + * @irq_handler_list_low_tab: + * + * Low priority IRQ handler table. * - * Each IRQ source may need to be handled at different contexts. - * By 'context' we mean, for example: - * - The ISR context, which is the direct interrupt handler. - * - The 'deferred' context - this is the post-processing of the - * interrupt, but at a lower priority. + * It is a n*m table consisting of n IRQ sources, and m handlers per IRQ + * source. Low priority IRQ handlers are deferred to a workqueue to be + * processed. Hence, they can sleep. * * Note that handlers are called in the same order as they were * registered (FIFO). */ struct irq_list_head irq_handler_list_low_tab[DAL_IRQ_SOURCES_NUMBER]; + + /** + * @irq_handler_list_high_tab: + * + * High priority IRQ handler table. + * + * It is a n*m table, same as &irq_handler_list_low_tab. However, + * handlers in this table are not deferred and are called immediately. + */ struct list_head irq_handler_list_high_tab[DAL_IRQ_SOURCES_NUMBER]; + /** + * @pflip_params: + * + * Page flip IRQ parameters, passed to registered handlers when + * triggered. + */ struct common_irq_params pflip_params[DC_IRQ_SOURCE_PFLIP_LAST - DC_IRQ_SOURCE_PFLIP_FIRST + 1]; + /** + * @vblank_params: + * + * Vertical blanking IRQ parameters, passed to registered handlers when + * triggered. + */ struct common_irq_params vblank_params[DC_IRQ_SOURCE_VBLANK6 - DC_IRQ_SOURCE_VBLANK1 + 1]; - /* this spin lock synchronizes access to 'irq_handler_list_table' */ spinlock_t irq_handler_list_table_lock; struct backlight_device *backlight_dev; @@ -110,9 +161,6 @@ struct amdgpu_display_manager { struct mod_freesync *freesync_module; - /** - * Caches device atomic state for suspend/resume - */ struct drm_atomic_state *cached_state; struct dm_comressor_info compressor; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c index be19e6861189..216e48cec716 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c @@ -164,7 +164,7 @@ int amdgpu_dm_set_regamma_lut(struct dm_crtc_state *crtc) */ stream->out_transfer_func->type = TF_TYPE_DISTRIBUTED_POINTS; ret = mod_color_calculate_regamma_params(stream->out_transfer_func, - gamma, true, adev->asic_type <= CHIP_RAVEN); + gamma, true, adev->asic_type <= CHIP_RAVEN, NULL); dc_gamma_release(&gamma); if (!ret) { stream->out_transfer_func->type = old_type; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c index 01fc5717b657..f088ac585978 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c @@ -75,6 +75,11 @@ int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name) return -EINVAL; } + if (!stream_state) { + DRM_ERROR("No stream state for CRTC%d\n", crtc->index); + return -EINVAL; + } + /* When enabling CRC, we should also disable dithering. */ if (source == AMDGPU_DM_PIPE_CRC_SOURCE_AUTO) { if (dc_stream_configure_crc(stream_state->ctx->dc, diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c index a212178f2edc..cd10f77cdeb0 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c @@ -32,16 +32,55 @@ #include "amdgpu_dm.h" #include "amdgpu_dm_irq.h" +/** + * DOC: overview + * + * DM provides another layer of IRQ management on top of what the base driver + * already provides. This is something that could be cleaned up, and is a + * future TODO item. + * + * The base driver provides IRQ source registration with DRM, handler + * registration into the base driver's IRQ table, and a handler callback + * amdgpu_irq_handler(), with which DRM calls on interrupts. This generic + * handler looks up the IRQ table, and calls the respective + * &amdgpu_irq_src_funcs.process hookups. + * + * What DM provides on top are two IRQ tables specifically for top-half and + * bottom-half IRQ handling, with the bottom-half implementing workqueues: + * + * - &amdgpu_display_manager.irq_handler_list_high_tab + * - &amdgpu_display_manager.irq_handler_list_low_tab + * + * They override the base driver's IRQ table, and the effect can be seen + * in the hooks that DM provides for &amdgpu_irq_src_funcs.process. They + * are all set to the DM generic handler amdgpu_dm_irq_handler(), which looks up + * DM's IRQ tables. However, in order for base driver to recognize this hook, DM + * still needs to register the IRQ with the base driver. See + * dce110_register_irq_handlers() and dcn10_register_irq_handlers(). + * + * To expose DC's hardware interrupt toggle to the base driver, DM implements + * &amdgpu_irq_src_funcs.set hooks. Base driver calls it through + * amdgpu_irq_update() to enable or disable the interrupt. + */ + /****************************************************************************** * Private declarations. *****************************************************************************/ +/** + * struct amdgpu_dm_irq_handler_data - Data for DM interrupt handlers. + * + * @list: Linked list entry referencing the next/previous handler + * @handler: Handler function + * @handler_arg: Argument passed to the handler when triggered + * @dm: DM which this handler belongs to + * @irq_source: DC interrupt source that this handler is registered for + */ struct amdgpu_dm_irq_handler_data { struct list_head list; interrupt_handler handler; void *handler_arg; - /* DM which this handler belongs to */ struct amdgpu_display_manager *dm; /* DAL irq source which registered for this interrupt. */ enum dc_irq_source irq_source; @@ -68,7 +107,7 @@ static void init_handler_common_data(struct amdgpu_dm_irq_handler_data *hcd, } /** - * dm_irq_work_func - Handle an IRQ outside of the interrupt handler proper. + * dm_irq_work_func() - Handle an IRQ outside of the interrupt handler proper. * * @work: work struct */ @@ -99,8 +138,8 @@ static void dm_irq_work_func(struct work_struct *work) * (The most common use is HPD interrupt) */ } -/** - * Remove a handler and return a pointer to hander list from which the +/* + * Remove a handler and return a pointer to handler list from which the * handler was removed. */ static struct list_head *remove_irq_handler(struct amdgpu_device *adev, @@ -203,6 +242,24 @@ static bool validate_irq_unregistration_params(enum dc_irq_source irq_source, * Note: caller is responsible for input validation. *****************************************************************************/ +/** + * amdgpu_dm_irq_register_interrupt() - Register a handler within DM. + * @adev: The base driver device containing the DM device. + * @int_params: Interrupt parameters containing the source, and handler context + * @ih: Function pointer to the interrupt handler to register + * @handler_args: Arguments passed to the handler when the interrupt occurs + * + * Register an interrupt handler for the given IRQ source, under the given + * context. The context can either be high or low. High context handlers are + * executed directly within ISR context, while low context is executed within a + * workqueue, thereby allowing operations that sleep. + * + * Registered handlers are called in a FIFO manner, i.e. the most recently + * registered handler will be called first. + * + * Return: Handler data &struct amdgpu_dm_irq_handler_data containing the IRQ + * source, handler function, and args + */ void *amdgpu_dm_irq_register_interrupt(struct amdgpu_device *adev, struct dc_interrupt_params *int_params, void (*ih)(void *), @@ -261,6 +318,15 @@ void *amdgpu_dm_irq_register_interrupt(struct amdgpu_device *adev, return handler_data; } +/** + * amdgpu_dm_irq_unregister_interrupt() - Remove a handler from the DM IRQ table + * @adev: The base driver device containing the DM device + * @irq_source: IRQ source to remove the given handler from + * @ih: Function pointer to the interrupt handler to unregister + * + * Go through both low and high context IRQ tables, and find the given handler + * for the given irq source. If found, remove it. Otherwise, do nothing. + */ void amdgpu_dm_irq_unregister_interrupt(struct amdgpu_device *adev, enum dc_irq_source irq_source, void *ih) @@ -295,6 +361,20 @@ void amdgpu_dm_irq_unregister_interrupt(struct amdgpu_device *adev, } } +/** + * amdgpu_dm_irq_init() - Initialize DM IRQ management + * @adev: The base driver device containing the DM device + * + * Initialize DM's high and low context IRQ tables. + * + * The N by M table contains N IRQ sources, with M + * &struct amdgpu_dm_irq_handler_data hooked together in a linked list. The + * list_heads are initialized here. When an interrupt n is triggered, all m + * handlers are called in sequence, FIFO according to registration order. + * + * The low context table requires special steps to initialize, since handlers + * will be deferred to a workqueue. See &struct irq_list_head. + */ int amdgpu_dm_irq_init(struct amdgpu_device *adev) { int src; @@ -317,7 +397,12 @@ int amdgpu_dm_irq_init(struct amdgpu_device *adev) return 0; } -/* DM IRQ and timer resource release */ +/** + * amdgpu_dm_irq_fini() - Tear down DM IRQ management + * @adev: The base driver device containing the DM device + * + * Flush all work within the low context IRQ table. + */ void amdgpu_dm_irq_fini(struct amdgpu_device *adev) { int src; @@ -414,7 +499,7 @@ int amdgpu_dm_irq_resume_late(struct amdgpu_device *adev) return 0; } -/** +/* * amdgpu_dm_irq_schedule_work - schedule all work items registered for the * "irq_source". */ @@ -439,8 +524,9 @@ static void amdgpu_dm_irq_schedule_work(struct amdgpu_device *adev, } -/** amdgpu_dm_irq_immediate_work - * Callback high irq work immediately, don't send to work queue +/* + * amdgpu_dm_irq_immediate_work + * Callback high irq work immediately, don't send to work queue */ static void amdgpu_dm_irq_immediate_work(struct amdgpu_device *adev, enum dc_irq_source irq_source) @@ -467,11 +553,14 @@ static void amdgpu_dm_irq_immediate_work(struct amdgpu_device *adev, DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags); } -/* - * amdgpu_dm_irq_handler +/** + * amdgpu_dm_irq_handler - Generic DM IRQ handler + * @adev: amdgpu base driver device containing the DM device + * @source: Unused + * @entry: Data about the triggered interrupt * - * Generic IRQ handler, calls all registered high irq work immediately, and - * schedules work for low irq + * Calls all registered high irq work immediately, and schedules work for low + * irq. The DM IRQ table is used to find the corresponding handlers. */ static int amdgpu_dm_irq_handler(struct amdgpu_device *adev, struct amdgpu_irq_src *source, @@ -613,7 +702,7 @@ void amdgpu_dm_set_irq_funcs(struct amdgpu_device *adev) adev->hpd_irq.funcs = &dm_hpd_irq_funcs; } -/* +/** * amdgpu_dm_hpd_init - hpd setup callback. * * @adev: amdgpu_device pointer diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c index 12001a006b2d..9d2d6986b983 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c @@ -485,11 +485,11 @@ void pp_rv_set_display_requirement(struct pp_smu *pp, return; clock.clock_type = amd_pp_dcf_clock; - clock.clock_freq_in_khz = req->hard_min_dcefclk_khz; + clock.clock_freq_in_khz = req->hard_min_dcefclk_mhz * 1000; pp_funcs->display_clock_voltage_request(pp_handle, &clock); clock.clock_type = amd_pp_f_clock; - clock.clock_freq_in_khz = req->hard_min_fclk_khz; + clock.clock_freq_in_khz = req->hard_min_fclk_mhz * 1000; pp_funcs->display_clock_voltage_request(pp_handle, &clock); } @@ -518,13 +518,13 @@ void pp_rv_set_wm_ranges(struct pp_smu *pp, wm_dce_clocks[i].wm_set_id = ranges->reader_wm_sets[i].wm_inst; wm_dce_clocks[i].wm_max_dcfclk_clk_in_khz = - ranges->reader_wm_sets[i].max_drain_clk_khz; + ranges->reader_wm_sets[i].max_drain_clk_mhz * 1000; wm_dce_clocks[i].wm_min_dcfclk_clk_in_khz = - ranges->reader_wm_sets[i].min_drain_clk_khz; + ranges->reader_wm_sets[i].min_drain_clk_mhz * 1000; wm_dce_clocks[i].wm_max_mem_clk_in_khz = - ranges->reader_wm_sets[i].max_fill_clk_khz; + ranges->reader_wm_sets[i].max_fill_clk_mhz * 1000; wm_dce_clocks[i].wm_min_mem_clk_in_khz = - ranges->reader_wm_sets[i].min_fill_clk_khz; + ranges->reader_wm_sets[i].min_fill_clk_mhz * 1000; } for (i = 0; i < wm_with_clock_ranges.num_wm_mcif_sets; i++) { @@ -534,13 +534,13 @@ void pp_rv_set_wm_ranges(struct pp_smu *pp, wm_soc_clocks[i].wm_set_id = ranges->writer_wm_sets[i].wm_inst; wm_soc_clocks[i].wm_max_socclk_clk_in_khz = - ranges->writer_wm_sets[i].max_fill_clk_khz; + ranges->writer_wm_sets[i].max_fill_clk_mhz * 1000; wm_soc_clocks[i].wm_min_socclk_clk_in_khz = - ranges->writer_wm_sets[i].min_fill_clk_khz; + ranges->writer_wm_sets[i].min_fill_clk_mhz * 1000; wm_soc_clocks[i].wm_max_mem_clk_in_khz = - ranges->writer_wm_sets[i].max_drain_clk_khz; + ranges->writer_wm_sets[i].max_drain_clk_mhz * 1000; wm_soc_clocks[i].wm_min_mem_clk_in_khz = - ranges->writer_wm_sets[i].min_drain_clk_khz; + ranges->writer_wm_sets[i].min_drain_clk_mhz * 1000; } pp_funcs->set_watermarks_for_clocks_ranges(pp_handle, &wm_with_clock_ranges); diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c index 0e1dc1b1a48d..c2ab026aee91 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c @@ -2030,7 +2030,7 @@ static uint32_t get_src_obj_list(struct bios_parser *bp, ATOM_OBJECT *object, static struct device_id device_type_from_device_id(uint16_t device_id) { - struct device_id result_device_id; + struct device_id result_device_id = {0}; switch (device_id) { case ATOM_DEVICE_LCD1_SUPPORT: diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c index ff764da21b6f..751bb614fc0e 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c @@ -1884,6 +1884,8 @@ static const struct dc_vbios_funcs vbios_funcs = { .is_accelerated_mode = bios_parser_is_accelerated_mode, + .is_active_display = bios_is_active_display, + .set_scratch_critical_state = bios_parser_set_scratch_critical_state, diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.c index d4589470985c..fdda8aa8e303 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.c @@ -88,3 +88,96 @@ uint32_t bios_get_vga_enabled_displays( return active_disp; } +bool bios_is_active_display( + struct dc_bios *bios, + enum signal_type signal, + const struct connector_device_tag_info *device_tag) +{ + uint32_t active = 0; + uint32_t connected = 0; + uint32_t bios_scratch_0 = 0; + uint32_t bios_scratch_3 = 0; + + switch (signal) { + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + case SIGNAL_TYPE_HDMI_TYPE_A: + case SIGNAL_TYPE_DISPLAY_PORT: + case SIGNAL_TYPE_DISPLAY_PORT_MST: + { + if (device_tag->dev_id.device_type == DEVICE_TYPE_DFP) { + switch (device_tag->dev_id.enum_id) { + case 1: + { + active = ATOM_S3_DFP1_ACTIVE; + connected = 0x0008; //ATOM_DISPLAY_DFP1_CONNECT + } + break; + + case 2: + { + active = ATOM_S3_DFP2_ACTIVE; + connected = 0x0080; //ATOM_DISPLAY_DFP2_CONNECT + } + break; + + case 3: + { + active = ATOM_S3_DFP3_ACTIVE; + connected = 0x0200; //ATOM_DISPLAY_DFP3_CONNECT + } + break; + + case 4: + { + active = ATOM_S3_DFP4_ACTIVE; + connected = 0x0400; //ATOM_DISPLAY_DFP4_CONNECT + } + break; + + case 5: + { + active = ATOM_S3_DFP5_ACTIVE; + connected = 0x0800; //ATOM_DISPLAY_DFP5_CONNECT + } + break; + + case 6: + { + active = ATOM_S3_DFP6_ACTIVE; + connected = 0x0040; //ATOM_DISPLAY_DFP6_CONNECT + } + break; + + default: + break; + } + } + } + break; + + case SIGNAL_TYPE_LVDS: + case SIGNAL_TYPE_EDP: + { + active = ATOM_S3_LCD1_ACTIVE; + connected = 0x0002; //ATOM_DISPLAY_LCD1_CONNECT + } + break; + + default: + break; + } + + + if (bios->regs->BIOS_SCRATCH_0) /*follow up with other asic, todo*/ + bios_scratch_0 = REG_READ(BIOS_SCRATCH_0); + if (bios->regs->BIOS_SCRATCH_3) /*follow up with other asic, todo*/ + bios_scratch_3 = REG_READ(BIOS_SCRATCH_3); + + bios_scratch_3 &= ATOM_S3_DEVICE_ACTIVE_MASK; + if ((active & bios_scratch_3) && (connected & bios_scratch_0)) + return true; + + return false; +} + diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.h b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.h index 75a29e68fb27..f33cac2147e3 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.h +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.h @@ -35,6 +35,10 @@ bool bios_is_accelerated_mode(struct dc_bios *bios); void bios_set_scratch_acc_mode_change(struct dc_bios *bios); void bios_set_scratch_critical_state(struct dc_bios *bios, bool state); uint32_t bios_get_vga_enabled_displays(struct dc_bios *bios); +bool bios_is_active_display( + struct dc_bios *bios, + enum signal_type signal, + const struct connector_device_tag_info *device_tag); #define GET_IMAGE(type, offset) ((type *) bios_get_image(&bp->base, offset, sizeof(type))) diff --git a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c index 3208188b7ed4..43e4a2be0fa6 100644 --- a/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/calcs/dcn_calcs.c @@ -1423,27 +1423,27 @@ void dcn_bw_notify_pplib_of_wm_ranges(struct dc *dc) ranges.num_reader_wm_sets = WM_SET_COUNT; ranges.num_writer_wm_sets = WM_SET_COUNT; ranges.reader_wm_sets[0].wm_inst = WM_A; - ranges.reader_wm_sets[0].min_drain_clk_khz = min_dcfclk_khz; - ranges.reader_wm_sets[0].max_drain_clk_khz = overdrive; - ranges.reader_wm_sets[0].min_fill_clk_khz = min_fclk_khz; - ranges.reader_wm_sets[0].max_fill_clk_khz = overdrive; + ranges.reader_wm_sets[0].min_drain_clk_mhz = min_dcfclk_khz / 1000; + ranges.reader_wm_sets[0].max_drain_clk_mhz = overdrive / 1000; + ranges.reader_wm_sets[0].min_fill_clk_mhz = min_fclk_khz / 1000; + ranges.reader_wm_sets[0].max_fill_clk_mhz = overdrive / 1000; ranges.writer_wm_sets[0].wm_inst = WM_A; - ranges.writer_wm_sets[0].min_fill_clk_khz = socclk_khz; - ranges.writer_wm_sets[0].max_fill_clk_khz = overdrive; - ranges.writer_wm_sets[0].min_drain_clk_khz = min_fclk_khz; - ranges.writer_wm_sets[0].max_drain_clk_khz = overdrive; + ranges.writer_wm_sets[0].min_fill_clk_mhz = socclk_khz / 1000; + ranges.writer_wm_sets[0].max_fill_clk_mhz = overdrive / 1000; + ranges.writer_wm_sets[0].min_drain_clk_mhz = min_fclk_khz / 1000; + ranges.writer_wm_sets[0].max_drain_clk_mhz = overdrive / 1000; if (dc->debug.pplib_wm_report_mode == WM_REPORT_OVERRIDE) { ranges.reader_wm_sets[0].wm_inst = WM_A; - ranges.reader_wm_sets[0].min_drain_clk_khz = 300000; - ranges.reader_wm_sets[0].max_drain_clk_khz = 5000000; - ranges.reader_wm_sets[0].min_fill_clk_khz = 800000; - ranges.reader_wm_sets[0].max_fill_clk_khz = 5000000; + ranges.reader_wm_sets[0].min_drain_clk_mhz = 300; + ranges.reader_wm_sets[0].max_drain_clk_mhz = 5000; + ranges.reader_wm_sets[0].min_fill_clk_mhz = 800; + ranges.reader_wm_sets[0].max_fill_clk_mhz = 5000; ranges.writer_wm_sets[0].wm_inst = WM_A; - ranges.writer_wm_sets[0].min_fill_clk_khz = 200000; - ranges.writer_wm_sets[0].max_fill_clk_khz = 5000000; - ranges.writer_wm_sets[0].min_drain_clk_khz = 800000; - ranges.writer_wm_sets[0].max_drain_clk_khz = 5000000; + ranges.writer_wm_sets[0].min_fill_clk_mhz = 200; + ranges.writer_wm_sets[0].max_fill_clk_mhz = 5000; + ranges.writer_wm_sets[0].min_drain_clk_mhz = 800; + ranges.writer_wm_sets[0].max_drain_clk_mhz = 5000; } ranges.reader_wm_sets[1] = ranges.writer_wm_sets[0]; diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 7c491c91465f..3279e26c3440 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -391,9 +391,11 @@ bool dc_stream_program_csc_matrix(struct dc *dc, struct dc_stream_state *stream) == stream) { pipes = &dc->current_state->res_ctx.pipe_ctx[i]; - dc->hwss.program_csc_matrix(pipes, - stream->output_color_space, - stream->csc_color_matrix.matrix); + dc->hwss.program_output_csc(dc, + pipes, + stream->output_color_space, + stream->csc_color_matrix.matrix, + pipes->plane_res.hubp->opp_id); ret = true; } } @@ -941,7 +943,7 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c if (!dcb->funcs->is_accelerated_mode(dcb)) dc->hwss.enable_accelerated_mode(dc, context); - dc->hwss.set_bandwidth(dc, context, false); + dc->hwss.prepare_bandwidth(dc, context); /* re-program planes for existing stream, in case we need to * free up plane resource for later use @@ -957,8 +959,6 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c } /* Program hardware */ - dc->hwss.ready_shared_resources(dc, context); - for (i = 0; i < dc->res_pool->pipe_count; i++) { pipe = &context->res_ctx.pipe_ctx[i]; dc->hwss.wait_for_mpcc_disconnect(dc, dc->res_pool, pipe); @@ -1012,7 +1012,7 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c dc_enable_stereo(dc, context, dc_streams, context->stream_count); /* pplib is notified if disp_num changed */ - dc->hwss.set_bandwidth(dc, context, true); + dc->hwss.optimize_bandwidth(dc, context); dc_release_state(dc->current_state); @@ -1020,8 +1020,6 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c dc_retain_state(dc->current_state); - dc->hwss.optimize_shared_resources(dc); - return result; } @@ -1063,7 +1061,7 @@ bool dc_post_update_surfaces_to_stream(struct dc *dc) dc->optimized_required = false; - dc->hwss.set_bandwidth(dc, context, true); + dc->hwss.optimize_bandwidth(dc, context); return true; } @@ -1369,35 +1367,6 @@ static struct dc_stream_status *stream_get_status( static const enum surface_update_type update_surface_trace_level = UPDATE_TYPE_FULL; -static void notify_display_count_to_smu( - struct dc *dc, - struct dc_state *context) -{ - int i, display_count; - struct pp_smu_funcs_rv *pp_smu = dc->res_pool->pp_smu; - - /* - * if function pointer not set up, this message is - * sent as part of pplib_apply_display_requirements. - * So just return. - */ - if (!pp_smu || !pp_smu->set_display_count) - return; - - display_count = 0; - for (i = 0; i < context->stream_count; i++) { - const struct dc_stream_state *stream = context->streams[i]; - - /* only notify active stream */ - if (stream->dpms_off) - continue; - - display_count++; - } - - pp_smu->set_display_count(&pp_smu->pp_smu, display_count); -} - static void commit_planes_do_stream_update(struct dc *dc, struct dc_stream_state *stream, struct dc_stream_update *stream_update, @@ -1422,7 +1391,6 @@ static void commit_planes_do_stream_update(struct dc *dc, stream_update->adjust->v_total_max); if (stream_update->periodic_fn_vsync_delta && - pipe_ctx->stream_res.tg && pipe_ctx->stream_res.tg->funcs->program_vline_interrupt) pipe_ctx->stream_res.tg->funcs->program_vline_interrupt( pipe_ctx->stream_res.tg, &pipe_ctx->stream->timing, @@ -1448,19 +1416,13 @@ static void commit_planes_do_stream_update(struct dc *dc, if (stream_update->dpms_off) { if (*stream_update->dpms_off) { core_link_disable_stream(pipe_ctx, KEEP_ACQUIRED_RESOURCE); - dc->hwss.pplib_apply_display_requirements( - dc, dc->current_state); - notify_display_count_to_smu(dc, dc->current_state); + dc->hwss.optimize_bandwidth(dc, dc->current_state); } else { - dc->hwss.pplib_apply_display_requirements( - dc, dc->current_state); - notify_display_count_to_smu(dc, dc->current_state); + dc->hwss.prepare_bandwidth(dc, dc->current_state); core_link_enable_stream(dc->current_state, pipe_ctx); } } - - if (stream_update->abm_level && pipe_ctx->stream_res.abm) { if (pipe_ctx->stream_res.tg->funcs->is_blanked) { // if otg funcs defined check if blanked before programming @@ -1487,7 +1449,7 @@ static void commit_planes_for_stream(struct dc *dc, struct pipe_ctx *top_pipe_to_program = NULL; if (update_type == UPDATE_TYPE_FULL) { - dc->hwss.set_bandwidth(dc, context, false); + dc->hwss.prepare_bandwidth(dc, context); context_clock_trace(dc, context); } diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_debug.c b/drivers/gpu/drm/amd/display/dc/core/dc_debug.c index e1ebdf7b5eaf..73d049506618 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_debug.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_debug.c @@ -311,7 +311,7 @@ void context_timing_trace( { int i; struct dc *core_dc = dc; - int h_pos[MAX_PIPES], v_pos[MAX_PIPES]; + int h_pos[MAX_PIPES] = {0}, v_pos[MAX_PIPES] = {0}; struct crtc_position position; unsigned int underlay_idx = core_dc->res_pool->underlay_pipe_index; DC_LOGGER_INIT(dc->ctx->logger); @@ -322,8 +322,7 @@ void context_timing_trace( /* get_position() returns CRTC vertical/horizontal counter * hence not applicable for underlay pipe */ - if (pipe_ctx->stream == NULL - || pipe_ctx->pipe_idx == underlay_idx) + if (pipe_ctx->stream == NULL || pipe_ctx->pipe_idx == underlay_idx) continue; pipe_ctx->stream_res.tg->funcs->get_position(pipe_ctx->stream_res.tg, &position); @@ -333,7 +332,7 @@ void context_timing_trace( for (i = 0; i < core_dc->res_pool->pipe_count; i++) { struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; - if (pipe_ctx->stream == NULL) + if (pipe_ctx->stream == NULL || pipe_ctx->pipe_idx == underlay_idx) continue; TIMING_TRACE("OTG_%d H_tot:%d V_tot:%d H_pos:%d V_pos:%d\n", diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c index 5da2186b3615..7ee9c033acbd 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c @@ -1357,28 +1357,13 @@ static enum dc_status enable_link_dp( struct dc_link *link = stream->sink->link; struct dc_link_settings link_settings = {0}; enum dp_panel_mode panel_mode; - enum dc_link_rate max_link_rate = LINK_RATE_HIGH2; /* get link settings for video mode timing */ decide_link_settings(stream, &link_settings); - /* raise clock state for HBR3 if required. Confirmed with HW DCE/DPCS - * logic for HBR3 still needs Nominal (0.8V) on VDDC rail - */ - if (link->link_enc->features.flags.bits.IS_HBR3_CAPABLE) - max_link_rate = LINK_RATE_HIGH3; - - if (link_settings.link_rate == max_link_rate) { - struct dc_clocks clocks = state->bw.dcn.clk; - - /* dce/dcn compat, do not update dispclk */ - clocks.dispclk_khz = 0; - /* 27mhz = 27000000hz= 27000khz */ - clocks.phyclk_khz = link_settings.link_rate * 27000; - - state->dis_clk->funcs->update_clocks( - state->dis_clk, &clocks, false); - } + pipe_ctx->stream_res.pix_clk_params.requested_sym_clk = + link_settings.link_rate * LINK_RATE_REF_FREQ_IN_KHZ; + state->dccg->funcs->update_clocks(state->dccg, state, false); dp_enable_link_phy( link, @@ -2156,14 +2141,16 @@ int dc_link_get_backlight_level(const struct dc_link *link) { struct abm *abm = link->ctx->dc->res_pool->abm; - if (abm == NULL || abm->funcs->get_current_backlight_8_bit == NULL) + if (abm == NULL || abm->funcs->get_current_backlight == NULL) return DC_ERROR_UNEXPECTED; - return (int) abm->funcs->get_current_backlight_8_bit(abm); + return (int) abm->funcs->get_current_backlight(abm); } -bool dc_link_set_backlight_level(const struct dc_link *link, uint32_t level, - uint32_t frame_ramp, const struct dc_stream_state *stream) +bool dc_link_set_backlight_level(const struct dc_link *link, + uint32_t backlight_pwm_u16_16, + uint32_t frame_ramp, + const struct dc_stream_state *stream) { struct dc *core_dc = link->ctx->dc; struct abm *abm = core_dc->res_pool->abm; @@ -2175,19 +2162,17 @@ bool dc_link_set_backlight_level(const struct dc_link *link, uint32_t level, if ((dmcu == NULL) || (abm == NULL) || - (abm->funcs->set_backlight_level == NULL)) + (abm->funcs->set_backlight_level_pwm == NULL)) return false; - if (stream) { - if (stream->bl_pwm_level == EDP_BACKLIGHT_RAMP_DISABLE_LEVEL) - frame_ramp = 0; - - ((struct dc_stream_state *)stream)->bl_pwm_level = level; - } + if (stream) + ((struct dc_stream_state *)stream)->bl_pwm_level = + backlight_pwm_u16_16; use_smooth_brightness = dmcu->funcs->is_dmcu_initialized(dmcu); - DC_LOG_BACKLIGHT("New Backlight level: %d (0x%X)\n", level, level); + DC_LOG_BACKLIGHT("New Backlight level: %d (0x%X)\n", + backlight_pwm_u16_16, backlight_pwm_u16_16); if (dc_is_embedded_signal(link->connector_signal)) { if (stream != NULL) { @@ -2204,9 +2189,9 @@ bool dc_link_set_backlight_level(const struct dc_link *link, uint32_t level, 1; } } - abm->funcs->set_backlight_level( + abm->funcs->set_backlight_level_pwm( abm, - level, + backlight_pwm_u16_16, frame_ramp, controller_id, use_smooth_brightness); @@ -2220,7 +2205,7 @@ bool dc_link_set_abm_disable(const struct dc_link *link) struct dc *core_dc = link->ctx->dc; struct abm *abm = core_dc->res_pool->abm; - if ((abm == NULL) || (abm->funcs->set_backlight_level == NULL)) + if ((abm == NULL) || (abm->funcs->set_backlight_level_pwm == NULL)) return false; abm->funcs->set_abm_immediate_disable(abm); @@ -2609,6 +2594,10 @@ void core_link_enable_stream( core_dc->hwss.unblank_stream(pipe_ctx, &pipe_ctx->stream->sink->link->cur_link_settings); + dc_link_set_backlight_level(pipe_ctx->stream->sink->link, + pipe_ctx->stream->bl_pwm_level, + 0, + pipe_ctx->stream); } } diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index b6fe29b9fb65..fc65b0055167 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -499,8 +499,13 @@ static void calculate_viewport(struct pipe_ctx *pipe_ctx) pipe_ctx->top_pipe->plane_state == pipe_ctx->plane_state; bool flip_vert_scan_dir = false, flip_horz_scan_dir = false; + /* - * Need to calculate the scan direction for viewport to properly determine offset + * We need take horizontal mirror into account. On an unrotated surface this means + * that the viewport offset is actually the offset from the other side of source + * image so we have to subtract the right edge of the viewport from the right edge of + * the source window. Similar to mirror we need to take into account how offset is + * affected for 270/180 rotations */ if (pipe_ctx->plane_state->rotation == ROTATION_ANGLE_180) { flip_vert_scan_dir = true; @@ -510,6 +515,9 @@ static void calculate_viewport(struct pipe_ctx *pipe_ctx) else if (pipe_ctx->plane_state->rotation == ROTATION_ANGLE_270) flip_horz_scan_dir = true; + if (pipe_ctx->plane_state->horizontal_mirror) + flip_horz_scan_dir = !flip_horz_scan_dir; + if (stream->view_format == VIEW_3D_FORMAT_SIDE_BY_SIDE || stream->view_format == VIEW_3D_FORMAT_TOP_AND_BOTTOM) { pri_split = false; @@ -540,45 +548,27 @@ static void calculate_viewport(struct pipe_ctx *pipe_ctx) plane_state->clip_rect.y + plane_state->clip_rect.height - clip.y ; /* offset = surf_src.ofs + (clip.ofs - surface->dst_rect.ofs) * scl_ratio + * note: surf_src.ofs should be added after rotation/mirror offset direction + * adjustment since it is already in viewport space * num_pixels = clip.num_pix * scl_ratio */ - data->viewport.x = surf_src.x + (clip.x - plane_state->dst_rect.x) * + data->viewport.x = (clip.x - plane_state->dst_rect.x) * surf_src.width / plane_state->dst_rect.width; data->viewport.width = clip.width * surf_src.width / plane_state->dst_rect.width; - data->viewport.y = surf_src.y + (clip.y - plane_state->dst_rect.y) * + data->viewport.y = (clip.y - plane_state->dst_rect.y) * surf_src.height / plane_state->dst_rect.height; data->viewport.height = clip.height * surf_src.height / plane_state->dst_rect.height; - /* To transfer the x, y to correct coordinate on mirror image (camera). - * deg 0 : transfer x, - * deg 90 : don't need to transfer, - * deg180 : transfer y, - * deg270 : transfer x and y. - * To transfer the x, y to correct coordinate on non-mirror image (video). - * deg 0 : don't need to transfer, - * deg 90 : transfer y, - * deg180 : transfer x and y, - * deg270 : transfer x. - */ - if (pipe_ctx->plane_state->horizontal_mirror) { - if (flip_horz_scan_dir && !flip_vert_scan_dir) { - data->viewport.y = surf_src.height - data->viewport.y - data->viewport.height; - data->viewport.x = surf_src.width - data->viewport.x - data->viewport.width; - } else if (flip_horz_scan_dir && flip_vert_scan_dir) - data->viewport.y = surf_src.height - data->viewport.y - data->viewport.height; - else { - if (!flip_horz_scan_dir && !flip_vert_scan_dir) - data->viewport.x = surf_src.width - data->viewport.x - data->viewport.width; - } - } else { - if (flip_horz_scan_dir) - data->viewport.x = surf_src.width - data->viewport.x - data->viewport.width; - if (flip_vert_scan_dir) - data->viewport.y = surf_src.height - data->viewport.y - data->viewport.height; - } + if (flip_vert_scan_dir) + data->viewport.y = surf_src.height - data->viewport.y - data->viewport.height; + if (flip_horz_scan_dir) + data->viewport.x = surf_src.width - data->viewport.x - data->viewport.width; + + data->viewport.x += surf_src.x; + data->viewport.y += surf_src.y; /* Round down, compensate in init */ data->viewport_c.x = data->viewport.x / vpc_div; @@ -773,22 +763,15 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx, struct rect *r else if (pipe_ctx->plane_state->rotation == ROTATION_ANGLE_270) flip_horz_scan_dir = true; + if (pipe_ctx->plane_state->horizontal_mirror) + flip_horz_scan_dir = !flip_horz_scan_dir; + if (pipe_ctx->plane_state->rotation == ROTATION_ANGLE_90 || pipe_ctx->plane_state->rotation == ROTATION_ANGLE_270) { rect_swap_helper(&src); rect_swap_helper(&data->viewport_c); rect_swap_helper(&data->viewport); - - if (pipe_ctx->plane_state->rotation == ROTATION_ANGLE_270 && - pipe_ctx->plane_state->horizontal_mirror) { - flip_vert_scan_dir = true; - } - if (pipe_ctx->plane_state->rotation == ROTATION_ANGLE_90 && - pipe_ctx->plane_state->horizontal_mirror) { - flip_vert_scan_dir = false; - } - } else if (pipe_ctx->plane_state->horizontal_mirror) - flip_horz_scan_dir = !flip_horz_scan_dir; + } /* * Init calculated according to formula: @@ -1115,9 +1098,6 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx) pipe_ctx->plane_res.scl_data.format = convert_pixel_format_to_dalsurface( pipe_ctx->plane_state->format); - if (pipe_ctx->stream->timing.flags.INTERLACE) - pipe_ctx->stream->dst.height *= 2; - calculate_scaling_ratios(pipe_ctx); calculate_viewport(pipe_ctx); @@ -1138,9 +1118,6 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx) pipe_ctx->plane_res.scl_data.h_active = timing->h_addressable + timing->h_border_left + timing->h_border_right; pipe_ctx->plane_res.scl_data.v_active = timing->v_addressable + timing->v_border_top + timing->v_border_bottom; - if (pipe_ctx->stream->timing.flags.INTERLACE) - pipe_ctx->plane_res.scl_data.v_active *= 2; - /* Taps calculations */ if (pipe_ctx->plane_res.xfm != NULL) @@ -1185,9 +1162,6 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx) plane_state->dst_rect.x, plane_state->dst_rect.y); - if (pipe_ctx->stream->timing.flags.INTERLACE) - pipe_ctx->stream->dst.height /= 2; - return res; } @@ -2071,7 +2045,7 @@ void dc_resource_state_construct( const struct dc *dc, struct dc_state *dst_ctx) { - dst_ctx->dis_clk = dc->res_pool->dccg; + dst_ctx->dccg = dc->res_pool->clk_mgr; } enum dc_status dc_validate_global_state( diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c index 2ac848a106ba..e113439aaa86 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c @@ -106,6 +106,7 @@ static void construct(struct dc_stream_state *stream, stream->out_transfer_func = dc_create_transfer_func(); stream->out_transfer_func->type = TF_TYPE_BYPASS; + stream->out_transfer_func->ctx = stream->ctx; } static void destruct(struct dc_stream_state *stream) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_surface.c b/drivers/gpu/drm/amd/display/dc/core/dc_surface.c index 8fb3aefd195c..c60c9b4c3075 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_surface.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_surface.c @@ -44,6 +44,7 @@ static void construct(struct dc_context *ctx, struct dc_plane_state *plane_state plane_state->in_transfer_func = dc_create_transfer_func(); plane_state->in_transfer_func->type = TF_TYPE_BYPASS; + plane_state->in_transfer_func->ctx = ctx; } static void destruct(struct dc_plane_state *plane_state) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index b57fa61b3034..d16a20c84792 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -38,7 +38,7 @@ #include "inc/compressor.h" #include "dml/display_mode_lib.h" -#define DC_VER "3.1.68" +#define DC_VER "3.2.04" #define MAX_SURFACES 3 #define MAX_STREAMS 6 @@ -250,8 +250,6 @@ struct dc_debug_options { bool disable_dmcu; bool disable_psr; bool force_abm_enable; - bool disable_hbup_pg; - bool disable_dpp_pg; bool disable_stereo_support; bool vsr_support; bool performance_trace; @@ -305,11 +303,6 @@ struct dc { struct hw_sequencer_funcs hwss; struct dce_hwseq *hwseq; - /* temp store of dm_pp_display_configuration - * to compare to see if display config changed - */ - struct dm_pp_display_configuration prev_display_config; - bool optimized_required; /* FBC compressor */ diff --git a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h index 8130b95ccc53..a8b3cedf9431 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h @@ -86,6 +86,10 @@ struct dc_vbios_funcs { bool (*is_accelerated_mode)( struct dc_bios *bios); + bool (*is_active_display)( + struct dc_bios *bios, + enum signal_type signal, + const struct connector_device_tag_info *device_tag); void (*set_scratch_critical_state)( struct dc_bios *bios, bool state); @@ -141,6 +145,7 @@ struct dc_vbios_funcs { }; struct bios_registers { + uint32_t BIOS_SCRATCH_0; uint32_t BIOS_SCRATCH_3; uint32_t BIOS_SCRATCH_6; }; diff --git a/drivers/gpu/drm/amd/display/dc/dc_link.h b/drivers/gpu/drm/amd/display/dc/dc_link.h index 3bfdccceb524..8738f27a8708 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_link.h +++ b/drivers/gpu/drm/amd/display/dc/dc_link.h @@ -138,9 +138,14 @@ static inline struct dc_link *dc_get_link_at_index(struct dc *dc, uint32_t link_ return dc->links[link_index]; } -/* Set backlight level of an embedded panel (eDP, LVDS). */ -bool dc_link_set_backlight_level(const struct dc_link *dc_link, uint32_t level, - uint32_t frame_ramp, const struct dc_stream_state *stream); +/* Set backlight level of an embedded panel (eDP, LVDS). + * backlight_pwm_u16_16 is unsigned 32 bit with 16 bit integer + * and 16 bit fractional, where 1.0 is max backlight value. + */ +bool dc_link_set_backlight_level(const struct dc_link *dc_link, + uint32_t backlight_pwm_u16_16, + uint32_t frame_ramp, + const struct dc_stream_state *stream); int dc_link_get_backlight_level(const struct dc_link *dc_link); diff --git a/drivers/gpu/drm/amd/display/dc/dce/Makefile b/drivers/gpu/drm/amd/display/dc/dce/Makefile index 8f7f0e8b341f..6d7b64a743ca 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dce/Makefile @@ -28,7 +28,7 @@ DCE = dce_audio.o dce_stream_encoder.o dce_link_encoder.o dce_hwseq.o \ dce_mem_input.o dce_clock_source.o dce_scl_filters.o dce_transform.o \ -dce_clocks.o dce_opp.o dce_dmcu.o dce_abm.o dce_ipp.o dce_aux.o \ +dce_clk_mgr.o dce_opp.o dce_dmcu.o dce_abm.o dce_ipp.o dce_aux.o \ dce_i2c.o dce_i2c_hw.o dce_i2c_sw.o AMD_DAL_DCE = $(addprefix $(AMDDALPATH)/dc/dce/,$(DCE)) diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c index 29294db1a96b..2a342eae80fd 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c @@ -54,7 +54,7 @@ #define MCP_DISABLE_ABM_IMMEDIATELY 255 -static unsigned int get_current_backlight_16_bit(struct dce_abm *abm_dce) +static unsigned int calculate_16_bit_backlight_from_pwm(struct dce_abm *abm_dce) { uint64_t current_backlight; uint32_t round_result; @@ -103,45 +103,21 @@ static unsigned int get_current_backlight_16_bit(struct dce_abm *abm_dce) return (uint32_t)(current_backlight); } -static void driver_set_backlight_level(struct dce_abm *abm_dce, uint32_t level) +static void driver_set_backlight_level(struct dce_abm *abm_dce, + uint32_t backlight_pwm_u16_16) { - uint32_t backlight_24bit; - uint32_t backlight_17bit; uint32_t backlight_16bit; uint32_t masked_pwm_period; - uint8_t rounding_bit; uint8_t bit_count; uint64_t active_duty_cycle; uint32_t pwm_period_bitcnt; /* - * 1. Convert 8-bit value to 17 bit U1.16 format - * (1 integer, 16 fractional bits) - */ - - /* 1.1 multiply 8 bit value by 0x10101 to get a 24 bit value, - * effectively multiplying value by 256/255 - * eg. for a level of 0xEF, backlight_24bit = 0xEF * 0x10101 = 0xEFEFEF - */ - backlight_24bit = level * 0x10101; - - /* 1.2 The upper 16 bits of the 24 bit value is the fraction, lower 8 - * used for rounding, take most significant bit of fraction for - * rounding, e.g. for 0xEFEFEF, rounding bit is 1 - */ - rounding_bit = (backlight_24bit >> 7) & 1; - - /* 1.3 Add the upper 16 bits of the 24 bit value with the rounding bit - * resulting in a 17 bit value e.g. 0xEFF0 = (0xEFEFEF >> 8) + 1 - */ - backlight_17bit = (backlight_24bit >> 8) + rounding_bit; - - /* - * 2. Find 16 bit backlight active duty cycle, where 0 <= backlight + * 1. Find 16 bit backlight active duty cycle, where 0 <= backlight * active duty cycle <= backlight period */ - /* 2.1 Apply bitmask for backlight period value based on value of BITCNT + /* 1.1 Apply bitmask for backlight period value based on value of BITCNT */ REG_GET_2(BL_PWM_PERIOD_CNTL, BL_PWM_PERIOD_BITCNT, &pwm_period_bitcnt, @@ -155,13 +131,13 @@ static void driver_set_backlight_level(struct dce_abm *abm_dce, uint32_t level) /* e.g. maskedPwmPeriod = 0x24 when bitCount is 6 */ masked_pwm_period = masked_pwm_period & ((1 << bit_count) - 1); - /* 2.2 Calculate integer active duty cycle required upper 16 bits + /* 1.2 Calculate integer active duty cycle required upper 16 bits * contain integer component, lower 16 bits contain fractional component * of active duty cycle e.g. 0x21BDC0 = 0xEFF0 * 0x24 */ - active_duty_cycle = backlight_17bit * masked_pwm_period; + active_duty_cycle = backlight_pwm_u16_16 * masked_pwm_period; - /* 2.3 Calculate 16 bit active duty cycle from integer and fractional + /* 1.3 Calculate 16 bit active duty cycle from integer and fractional * components shift by bitCount then mask 16 bits and add rounding bit * from MSB of fraction e.g. 0x86F7 = ((0x21BDC0 >> 6) & 0xFFF) + 0 */ @@ -170,23 +146,23 @@ static void driver_set_backlight_level(struct dce_abm *abm_dce, uint32_t level) backlight_16bit += (active_duty_cycle >> (bit_count - 1)) & 0x1; /* - * 3. Program register with updated value + * 2. Program register with updated value */ - /* 3.1 Lock group 2 backlight registers */ + /* 2.1 Lock group 2 backlight registers */ REG_UPDATE_2(BL_PWM_GRP1_REG_LOCK, BL_PWM_GRP1_IGNORE_MASTER_LOCK_EN, 1, BL_PWM_GRP1_REG_LOCK, 1); - // 3.2 Write new active duty cycle + // 2.2 Write new active duty cycle REG_UPDATE(BL_PWM_CNTL, BL_ACTIVE_INT_FRAC_CNT, backlight_16bit); - /* 3.3 Unlock group 2 backlight registers */ + /* 2.3 Unlock group 2 backlight registers */ REG_UPDATE(BL_PWM_GRP1_REG_LOCK, BL_PWM_GRP1_REG_LOCK, 0); - /* 5.4.4 Wait for pending bit to be cleared */ + /* 3 Wait for pending bit to be cleared */ REG_WAIT(BL_PWM_GRP1_REG_LOCK, BL_PWM_GRP1_REG_UPDATE_PENDING, 0, 1, 10000); @@ -194,16 +170,21 @@ static void driver_set_backlight_level(struct dce_abm *abm_dce, uint32_t level) static void dmcu_set_backlight_level( struct dce_abm *abm_dce, - uint32_t level, + uint32_t backlight_pwm_u16_16, uint32_t frame_ramp, uint32_t controller_id) { - unsigned int backlight_16_bit = (level * 0x10101) >> 8; - unsigned int backlight_17_bit = backlight_16_bit + - (((backlight_16_bit & 0x80) >> 7) & 1); + unsigned int backlight_8_bit = 0; uint32_t rampingBoundary = 0xFFFF; uint32_t s2; + if (backlight_pwm_u16_16 & 0x10000) + // Check for max backlight condition + backlight_8_bit = 0xFF; + else + // Take MSB of fractional part since backlight is not max + backlight_8_bit = (backlight_pwm_u16_16 >> 8) & 0xFF; + /* set ramping boundary */ REG_WRITE(MASTER_COMM_DATA_REG1, rampingBoundary); @@ -220,7 +201,7 @@ static void dmcu_set_backlight_level( 0, 1, 80000); /* setDMCUParam_BL */ - REG_UPDATE(BL1_PWM_USER_LEVEL, BL1_PWM_USER_LEVEL, backlight_17_bit); + REG_UPDATE(BL1_PWM_USER_LEVEL, BL1_PWM_USER_LEVEL, backlight_pwm_u16_16); /* write ramp */ if (controller_id == 0) @@ -237,9 +218,9 @@ static void dmcu_set_backlight_level( s2 = REG_READ(BIOS_SCRATCH_2); s2 &= ~ATOM_S2_CURRENT_BL_LEVEL_MASK; - level &= (ATOM_S2_CURRENT_BL_LEVEL_MASK >> + backlight_8_bit &= (ATOM_S2_CURRENT_BL_LEVEL_MASK >> ATOM_S2_CURRENT_BL_LEVEL_SHIFT); - s2 |= (level << ATOM_S2_CURRENT_BL_LEVEL_SHIFT); + s2 |= (backlight_8_bit << ATOM_S2_CURRENT_BL_LEVEL_SHIFT); REG_WRITE(BIOS_SCRATCH_2, s2); } @@ -247,7 +228,7 @@ static void dmcu_set_backlight_level( static void dce_abm_init(struct abm *abm) { struct dce_abm *abm_dce = TO_DCE_ABM(abm); - unsigned int backlight = get_current_backlight_16_bit(abm_dce); + unsigned int backlight = calculate_16_bit_backlight_from_pwm(abm_dce); REG_WRITE(DC_ABM1_HG_SAMPLE_RATE, 0x103); REG_WRITE(DC_ABM1_HG_SAMPLE_RATE, 0x101); @@ -284,12 +265,26 @@ static void dce_abm_init(struct abm *abm) ABM1_BL_REG_READ_MISSED_FRAME_CLEAR, 1); } -static unsigned int dce_abm_get_current_backlight_8_bit(struct abm *abm) +static unsigned int dce_abm_get_current_backlight(struct abm *abm) { struct dce_abm *abm_dce = TO_DCE_ABM(abm); unsigned int backlight = REG_READ(BL1_PWM_CURRENT_ABM_LEVEL); - return (backlight >> 8); + /* return backlight in hardware format which is unsigned 17 bits, with + * 1 bit integer and 16 bit fractional + */ + return backlight; +} + +static unsigned int dce_abm_get_target_backlight(struct abm *abm) +{ + struct dce_abm *abm_dce = TO_DCE_ABM(abm); + unsigned int backlight = REG_READ(BL1_PWM_TARGET_ABM_LEVEL); + + /* return backlight in hardware format which is unsigned 17 bits, with + * 1 bit integer and 16 bit fractional + */ + return backlight; } static bool dce_abm_set_level(struct abm *abm, uint32_t level) @@ -396,9 +391,9 @@ static bool dce_abm_init_backlight(struct abm *abm) return true; } -static bool dce_abm_set_backlight_level( +static bool dce_abm_set_backlight_level_pwm( struct abm *abm, - unsigned int backlight_level, + unsigned int backlight_pwm_u16_16, unsigned int frame_ramp, unsigned int controller_id, bool use_smooth_brightness) @@ -406,16 +401,16 @@ static bool dce_abm_set_backlight_level( struct dce_abm *abm_dce = TO_DCE_ABM(abm); DC_LOG_BACKLIGHT("New Backlight level: %d (0x%X)\n", - backlight_level, backlight_level); + backlight_pwm_u16_16, backlight_pwm_u16_16); /* If DMCU is in reset state, DMCU is uninitialized */ if (use_smooth_brightness) dmcu_set_backlight_level(abm_dce, - backlight_level, + backlight_pwm_u16_16, frame_ramp, controller_id); else - driver_set_backlight_level(abm_dce, backlight_level); + driver_set_backlight_level(abm_dce, backlight_pwm_u16_16); return true; } @@ -424,8 +419,9 @@ static const struct abm_funcs dce_funcs = { .abm_init = dce_abm_init, .set_abm_level = dce_abm_set_level, .init_backlight = dce_abm_init_backlight, - .set_backlight_level = dce_abm_set_backlight_level, - .get_current_backlight_8_bit = dce_abm_get_current_backlight_8_bit, + .set_backlight_level_pwm = dce_abm_set_backlight_level_pwm, + .get_current_backlight = dce_abm_get_current_backlight, + .get_target_backlight = dce_abm_get_target_backlight, .set_abm_immediate_disable = dce_abm_immediate_disable }; diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c new file mode 100644 index 000000000000..9a28a04417d1 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.c @@ -0,0 +1,879 @@ +/* + * Copyright 2012-16 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. + * + * Authors: AMD + * + */ + +#include "dce_clk_mgr.h" + +#include "reg_helper.h" +#include "dmcu.h" +#include "core_types.h" +#include "dal_asic_id.h" + +#define TO_DCE_CLK_MGR(clocks)\ + container_of(clocks, struct dce_clk_mgr, base) + +#define REG(reg) \ + (clk_mgr_dce->regs->reg) + +#undef FN +#define FN(reg_name, field_name) \ + clk_mgr_dce->clk_mgr_shift->field_name, clk_mgr_dce->clk_mgr_mask->field_name + +#define CTX \ + clk_mgr_dce->base.ctx +#define DC_LOGGER \ + clk_mgr->ctx->logger + +/* Max clock values for each state indexed by "enum clocks_state": */ +static const struct state_dependent_clocks dce80_max_clks_by_state[] = { +/* ClocksStateInvalid - should not be used */ +{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, +/* ClocksStateUltraLow - not expected to be used for DCE 8.0 */ +{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, +/* ClocksStateLow */ +{ .display_clk_khz = 352000, .pixel_clk_khz = 330000}, +/* ClocksStateNominal */ +{ .display_clk_khz = 600000, .pixel_clk_khz = 400000 }, +/* ClocksStatePerformance */ +{ .display_clk_khz = 600000, .pixel_clk_khz = 400000 } }; + +static const struct state_dependent_clocks dce110_max_clks_by_state[] = { +/*ClocksStateInvalid - should not be used*/ +{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, +/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/ +{ .display_clk_khz = 352000, .pixel_clk_khz = 330000 }, +/*ClocksStateLow*/ +{ .display_clk_khz = 352000, .pixel_clk_khz = 330000 }, +/*ClocksStateNominal*/ +{ .display_clk_khz = 467000, .pixel_clk_khz = 400000 }, +/*ClocksStatePerformance*/ +{ .display_clk_khz = 643000, .pixel_clk_khz = 400000 } }; + +static const struct state_dependent_clocks dce112_max_clks_by_state[] = { +/*ClocksStateInvalid - should not be used*/ +{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, +/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/ +{ .display_clk_khz = 389189, .pixel_clk_khz = 346672 }, +/*ClocksStateLow*/ +{ .display_clk_khz = 459000, .pixel_clk_khz = 400000 }, +/*ClocksStateNominal*/ +{ .display_clk_khz = 667000, .pixel_clk_khz = 600000 }, +/*ClocksStatePerformance*/ +{ .display_clk_khz = 1132000, .pixel_clk_khz = 600000 } }; + +static const struct state_dependent_clocks dce120_max_clks_by_state[] = { +/*ClocksStateInvalid - should not be used*/ +{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, +/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/ +{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, +/*ClocksStateLow*/ +{ .display_clk_khz = 460000, .pixel_clk_khz = 400000 }, +/*ClocksStateNominal*/ +{ .display_clk_khz = 670000, .pixel_clk_khz = 600000 }, +/*ClocksStatePerformance*/ +{ .display_clk_khz = 1133000, .pixel_clk_khz = 600000 } }; + +static int dentist_get_divider_from_did(int did) +{ + if (did < DENTIST_BASE_DID_1) + did = DENTIST_BASE_DID_1; + if (did > DENTIST_MAX_DID) + did = DENTIST_MAX_DID; + + if (did < DENTIST_BASE_DID_2) { + return DENTIST_DIVIDER_RANGE_1_START + DENTIST_DIVIDER_RANGE_1_STEP + * (did - DENTIST_BASE_DID_1); + } else if (did < DENTIST_BASE_DID_3) { + return DENTIST_DIVIDER_RANGE_2_START + DENTIST_DIVIDER_RANGE_2_STEP + * (did - DENTIST_BASE_DID_2); + } else if (did < DENTIST_BASE_DID_4) { + return DENTIST_DIVIDER_RANGE_3_START + DENTIST_DIVIDER_RANGE_3_STEP + * (did - DENTIST_BASE_DID_3); + } else { + return DENTIST_DIVIDER_RANGE_4_START + DENTIST_DIVIDER_RANGE_4_STEP + * (did - DENTIST_BASE_DID_4); + } +} + +/* SW will adjust DP REF Clock average value for all purposes + * (DP DTO / DP Audio DTO and DP GTC) + if clock is spread for all cases: + -if SS enabled on DP Ref clock and HW de-spreading enabled with SW + calculations for DS_INCR/DS_MODULO (this is planned to be default case) + -if SS enabled on DP Ref clock and HW de-spreading enabled with HW + calculations (not planned to be used, but average clock should still + be valid) + -if SS enabled on DP Ref clock and HW de-spreading disabled + (should not be case with CIK) then SW should program all rates + generated according to average value (case as with previous ASICs) + */ +static int clk_mgr_adjust_dp_ref_freq_for_ss(struct dce_clk_mgr *clk_mgr_dce, int dp_ref_clk_khz) +{ + if (clk_mgr_dce->ss_on_dprefclk && clk_mgr_dce->dprefclk_ss_divider != 0) { + struct fixed31_32 ss_percentage = dc_fixpt_div_int( + dc_fixpt_from_fraction(clk_mgr_dce->dprefclk_ss_percentage, + clk_mgr_dce->dprefclk_ss_divider), 200); + struct fixed31_32 adj_dp_ref_clk_khz; + + ss_percentage = dc_fixpt_sub(dc_fixpt_one, ss_percentage); + adj_dp_ref_clk_khz = dc_fixpt_mul_int(ss_percentage, dp_ref_clk_khz); + dp_ref_clk_khz = dc_fixpt_floor(adj_dp_ref_clk_khz); + } + return dp_ref_clk_khz; +} + +static int dce_get_dp_ref_freq_khz(struct clk_mgr *clk_mgr) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + int dprefclk_wdivider; + int dprefclk_src_sel; + int dp_ref_clk_khz = 600000; + int target_div; + + /* ASSERT DP Reference Clock source is from DFS*/ + REG_GET(DPREFCLK_CNTL, DPREFCLK_SRC_SEL, &dprefclk_src_sel); + ASSERT(dprefclk_src_sel == 0); + + /* Read the mmDENTIST_DISPCLK_CNTL to get the currently + * programmed DID DENTIST_DPREFCLK_WDIVIDER*/ + REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DPREFCLK_WDIVIDER, &dprefclk_wdivider); + + /* Convert DENTIST_DPREFCLK_WDIVIDERto actual divider*/ + target_div = dentist_get_divider_from_did(dprefclk_wdivider); + + /* Calculate the current DFS clock, in kHz.*/ + dp_ref_clk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR + * clk_mgr_dce->dentist_vco_freq_khz) / target_div; + + return clk_mgr_adjust_dp_ref_freq_for_ss(clk_mgr_dce, dp_ref_clk_khz); +} + +int dce12_get_dp_ref_freq_khz(struct clk_mgr *clk_mgr) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + + return clk_mgr_adjust_dp_ref_freq_for_ss(clk_mgr_dce, clk_mgr_dce->dprefclk_khz); +} + +/* unit: in_khz before mode set, get pixel clock from context. ASIC register + * may not be programmed yet + */ +static uint32_t get_max_pixel_clock_for_all_paths(struct dc_state *context) +{ + uint32_t max_pix_clk = 0; + int i; + + for (i = 0; i < MAX_PIPES; i++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; + + if (pipe_ctx->stream == NULL) + continue; + + /* do not check under lay */ + if (pipe_ctx->top_pipe) + continue; + + if (pipe_ctx->stream_res.pix_clk_params.requested_pix_clk > max_pix_clk) + max_pix_clk = pipe_ctx->stream_res.pix_clk_params.requested_pix_clk; + + /* raise clock state for HBR3/2 if required. Confirmed with HW DCE/DPCS + * logic for HBR3 still needs Nominal (0.8V) on VDDC rail + */ + if (dc_is_dp_signal(pipe_ctx->stream->signal) && + pipe_ctx->stream_res.pix_clk_params.requested_sym_clk > max_pix_clk) + max_pix_clk = pipe_ctx->stream_res.pix_clk_params.requested_sym_clk; + } + + return max_pix_clk; +} + +static enum dm_pp_clocks_state dce_get_required_clocks_state( + struct clk_mgr *clk_mgr, + struct dc_state *context) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + int i; + enum dm_pp_clocks_state low_req_clk; + int max_pix_clk = get_max_pixel_clock_for_all_paths(context); + + /* Iterate from highest supported to lowest valid state, and update + * lowest RequiredState with the lowest state that satisfies + * all required clocks + */ + for (i = clk_mgr_dce->max_clks_state; i >= DM_PP_CLOCKS_STATE_ULTRA_LOW; i--) + if (context->bw.dce.dispclk_khz > + clk_mgr_dce->max_clks_by_state[i].display_clk_khz + || max_pix_clk > + clk_mgr_dce->max_clks_by_state[i].pixel_clk_khz) + break; + + low_req_clk = i + 1; + if (low_req_clk > clk_mgr_dce->max_clks_state) { + /* set max clock state for high phyclock, invalid on exceeding display clock */ + if (clk_mgr_dce->max_clks_by_state[clk_mgr_dce->max_clks_state].display_clk_khz + < context->bw.dce.dispclk_khz) + low_req_clk = DM_PP_CLOCKS_STATE_INVALID; + else + low_req_clk = clk_mgr_dce->max_clks_state; + } + + return low_req_clk; +} + +static int dce_set_clock( + struct clk_mgr *clk_mgr, + int requested_clk_khz) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + struct bp_pixel_clock_parameters pxl_clk_params = { 0 }; + struct dc_bios *bp = clk_mgr->ctx->dc_bios; + int actual_clock = requested_clk_khz; + struct dmcu *dmcu = clk_mgr_dce->base.ctx->dc->res_pool->dmcu; + + /* Make sure requested clock isn't lower than minimum threshold*/ + if (requested_clk_khz > 0) + requested_clk_khz = max(requested_clk_khz, + clk_mgr_dce->dentist_vco_freq_khz / 64); + + /* Prepare to program display clock*/ + pxl_clk_params.target_pixel_clock = requested_clk_khz; + pxl_clk_params.pll_id = CLOCK_SOURCE_ID_DFS; + + if (clk_mgr_dce->dfs_bypass_active) + pxl_clk_params.flags.SET_DISPCLK_DFS_BYPASS = true; + + bp->funcs->program_display_engine_pll(bp, &pxl_clk_params); + + if (clk_mgr_dce->dfs_bypass_active) { + /* Cache the fixed display clock*/ + clk_mgr_dce->dfs_bypass_disp_clk = + pxl_clk_params.dfs_bypass_display_clock; + actual_clock = pxl_clk_params.dfs_bypass_display_clock; + } + + /* from power down, we need mark the clock state as ClocksStateNominal + * from HWReset, so when resume we will call pplib voltage regulator.*/ + if (requested_clk_khz == 0) + clk_mgr_dce->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL; + + dmcu->funcs->set_psr_wait_loop(dmcu, actual_clock / 1000 / 7); + + return actual_clock; +} + +int dce112_set_clock(struct clk_mgr *clk_mgr, int requested_clk_khz) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + struct bp_set_dce_clock_parameters dce_clk_params; + struct dc_bios *bp = clk_mgr->ctx->dc_bios; + struct dc *core_dc = clk_mgr->ctx->dc; + struct dmcu *dmcu = core_dc->res_pool->dmcu; + int actual_clock = requested_clk_khz; + /* Prepare to program display clock*/ + memset(&dce_clk_params, 0, sizeof(dce_clk_params)); + + /* Make sure requested clock isn't lower than minimum threshold*/ + if (requested_clk_khz > 0) + requested_clk_khz = max(requested_clk_khz, + clk_mgr_dce->dentist_vco_freq_khz / 62); + + dce_clk_params.target_clock_frequency = requested_clk_khz; + dce_clk_params.pll_id = CLOCK_SOURCE_ID_DFS; + dce_clk_params.clock_type = DCECLOCK_TYPE_DISPLAY_CLOCK; + + bp->funcs->set_dce_clock(bp, &dce_clk_params); + actual_clock = dce_clk_params.target_clock_frequency; + + /* from power down, we need mark the clock state as ClocksStateNominal + * from HWReset, so when resume we will call pplib voltage regulator.*/ + if (requested_clk_khz == 0) + clk_mgr_dce->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL; + + /*Program DP ref Clock*/ + /*VBIOS will determine DPREFCLK frequency, so we don't set it*/ + dce_clk_params.target_clock_frequency = 0; + dce_clk_params.clock_type = DCECLOCK_TYPE_DPREFCLK; + if (!ASICREV_IS_VEGA20_P(clk_mgr->ctx->asic_id.hw_internal_rev)) + dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = + (dce_clk_params.pll_id == + CLOCK_SOURCE_COMBO_DISPLAY_PLL0); + else + dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = false; + + bp->funcs->set_dce_clock(bp, &dce_clk_params); + + if (!IS_FPGA_MAXIMUS_DC(core_dc->ctx->dce_environment)) { + if (clk_mgr_dce->dfs_bypass_disp_clk != actual_clock) + dmcu->funcs->set_psr_wait_loop(dmcu, + actual_clock / 1000 / 7); + } + + clk_mgr_dce->dfs_bypass_disp_clk = actual_clock; + return actual_clock; +} + +static void dce_clock_read_integrated_info(struct dce_clk_mgr *clk_mgr_dce) +{ + struct dc_debug_options *debug = &clk_mgr_dce->base.ctx->dc->debug; + struct dc_bios *bp = clk_mgr_dce->base.ctx->dc_bios; + struct integrated_info info = { { { 0 } } }; + struct dc_firmware_info fw_info = { { 0 } }; + int i; + + if (bp->integrated_info) + info = *bp->integrated_info; + + clk_mgr_dce->dentist_vco_freq_khz = info.dentist_vco_freq; + if (clk_mgr_dce->dentist_vco_freq_khz == 0) { + bp->funcs->get_firmware_info(bp, &fw_info); + clk_mgr_dce->dentist_vco_freq_khz = + fw_info.smu_gpu_pll_output_freq; + if (clk_mgr_dce->dentist_vco_freq_khz == 0) + clk_mgr_dce->dentist_vco_freq_khz = 3600000; + } + + /*update the maximum display clock for each power state*/ + for (i = 0; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) { + enum dm_pp_clocks_state clk_state = DM_PP_CLOCKS_STATE_INVALID; + + switch (i) { + case 0: + clk_state = DM_PP_CLOCKS_STATE_ULTRA_LOW; + break; + + case 1: + clk_state = DM_PP_CLOCKS_STATE_LOW; + break; + + case 2: + clk_state = DM_PP_CLOCKS_STATE_NOMINAL; + break; + + case 3: + clk_state = DM_PP_CLOCKS_STATE_PERFORMANCE; + break; + + default: + clk_state = DM_PP_CLOCKS_STATE_INVALID; + break; + } + + /*Do not allow bad VBIOS/SBIOS to override with invalid values, + * check for > 100MHz*/ + if (info.disp_clk_voltage[i].max_supported_clk >= 100000) + clk_mgr_dce->max_clks_by_state[clk_state].display_clk_khz = + info.disp_clk_voltage[i].max_supported_clk; + } + + if (!debug->disable_dfs_bypass && bp->integrated_info) + if (bp->integrated_info->gpu_cap_info & DFS_BYPASS_ENABLE) + clk_mgr_dce->dfs_bypass_enabled = true; +} + +void dce_clock_read_ss_info(struct dce_clk_mgr *clk_mgr_dce) +{ + struct dc_bios *bp = clk_mgr_dce->base.ctx->dc_bios; + int ss_info_num = bp->funcs->get_ss_entry_number( + bp, AS_SIGNAL_TYPE_GPU_PLL); + + if (ss_info_num) { + struct spread_spectrum_info info = { { 0 } }; + enum bp_result result = bp->funcs->get_spread_spectrum_info( + bp, AS_SIGNAL_TYPE_GPU_PLL, 0, &info); + + /* Based on VBIOS, VBIOS will keep entry for GPU PLL SS + * even if SS not enabled and in that case + * SSInfo.spreadSpectrumPercentage !=0 would be sign + * that SS is enabled + */ + if (result == BP_RESULT_OK && + info.spread_spectrum_percentage != 0) { + clk_mgr_dce->ss_on_dprefclk = true; + clk_mgr_dce->dprefclk_ss_divider = info.spread_percentage_divider; + + if (info.type.CENTER_MODE == 0) { + /* TODO: Currently for DP Reference clock we + * need only SS percentage for + * downspread */ + clk_mgr_dce->dprefclk_ss_percentage = + info.spread_spectrum_percentage; + } + + return; + } + + result = bp->funcs->get_spread_spectrum_info( + bp, AS_SIGNAL_TYPE_DISPLAY_PORT, 0, &info); + + /* Based on VBIOS, VBIOS will keep entry for DPREFCLK SS + * even if SS not enabled and in that case + * SSInfo.spreadSpectrumPercentage !=0 would be sign + * that SS is enabled + */ + if (result == BP_RESULT_OK && + info.spread_spectrum_percentage != 0) { + clk_mgr_dce->ss_on_dprefclk = true; + clk_mgr_dce->dprefclk_ss_divider = info.spread_percentage_divider; + + if (info.type.CENTER_MODE == 0) { + /* Currently for DP Reference clock we + * need only SS percentage for + * downspread */ + clk_mgr_dce->dprefclk_ss_percentage = + info.spread_spectrum_percentage; + } + } + } +} + +void dce110_fill_display_configs( + const struct dc_state *context, + struct dm_pp_display_configuration *pp_display_cfg) +{ + int j; + int num_cfgs = 0; + + for (j = 0; j < context->stream_count; j++) { + int k; + + const struct dc_stream_state *stream = context->streams[j]; + struct dm_pp_single_disp_config *cfg = + &pp_display_cfg->disp_configs[num_cfgs]; + const struct pipe_ctx *pipe_ctx = NULL; + + for (k = 0; k < MAX_PIPES; k++) + if (stream == context->res_ctx.pipe_ctx[k].stream) { + pipe_ctx = &context->res_ctx.pipe_ctx[k]; + break; + } + + ASSERT(pipe_ctx != NULL); + + /* only notify active stream */ + if (stream->dpms_off) + continue; + + num_cfgs++; + cfg->signal = pipe_ctx->stream->signal; + cfg->pipe_idx = pipe_ctx->stream_res.tg->inst; + cfg->src_height = stream->src.height; + cfg->src_width = stream->src.width; + cfg->ddi_channel_mapping = + stream->sink->link->ddi_channel_mapping.raw; + cfg->transmitter = + stream->sink->link->link_enc->transmitter; + cfg->link_settings.lane_count = + stream->sink->link->cur_link_settings.lane_count; + cfg->link_settings.link_rate = + stream->sink->link->cur_link_settings.link_rate; + cfg->link_settings.link_spread = + stream->sink->link->cur_link_settings.link_spread; + cfg->sym_clock = stream->phy_pix_clk; + /* Round v_refresh*/ + cfg->v_refresh = stream->timing.pix_clk_khz * 1000; + cfg->v_refresh /= stream->timing.h_total; + cfg->v_refresh = (cfg->v_refresh + stream->timing.v_total / 2) + / stream->timing.v_total; + } + + pp_display_cfg->display_count = num_cfgs; +} + +static uint32_t dce110_get_min_vblank_time_us(const struct dc_state *context) +{ + uint8_t j; + uint32_t min_vertical_blank_time = -1; + + for (j = 0; j < context->stream_count; j++) { + struct dc_stream_state *stream = context->streams[j]; + uint32_t vertical_blank_in_pixels = 0; + uint32_t vertical_blank_time = 0; + + vertical_blank_in_pixels = stream->timing.h_total * + (stream->timing.v_total + - stream->timing.v_addressable); + + vertical_blank_time = vertical_blank_in_pixels + * 1000 / stream->timing.pix_clk_khz; + + if (min_vertical_blank_time > vertical_blank_time) + min_vertical_blank_time = vertical_blank_time; + } + + return min_vertical_blank_time; +} + +static int determine_sclk_from_bounding_box( + const struct dc *dc, + int required_sclk) +{ + int i; + + /* + * Some asics do not give us sclk levels, so we just report the actual + * required sclk + */ + if (dc->sclk_lvls.num_levels == 0) + return required_sclk; + + for (i = 0; i < dc->sclk_lvls.num_levels; i++) { + if (dc->sclk_lvls.clocks_in_khz[i] >= required_sclk) + return dc->sclk_lvls.clocks_in_khz[i]; + } + /* + * even maximum level could not satisfy requirement, this + * is unexpected at this stage, should have been caught at + * validation time + */ + ASSERT(0); + return dc->sclk_lvls.clocks_in_khz[dc->sclk_lvls.num_levels - 1]; +} + +static void dce_pplib_apply_display_requirements( + struct dc *dc, + struct dc_state *context) +{ + struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg; + + pp_display_cfg->avail_mclk_switch_time_us = dce110_get_min_vblank_time_us(context); + + dce110_fill_display_configs(context, pp_display_cfg); + + if (memcmp(&dc->current_state->pp_display_cfg, pp_display_cfg, sizeof(*pp_display_cfg)) != 0) + dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg); +} + +static void dce11_pplib_apply_display_requirements( + struct dc *dc, + struct dc_state *context) +{ + struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg; + + pp_display_cfg->all_displays_in_sync = + context->bw.dce.all_displays_in_sync; + pp_display_cfg->nb_pstate_switch_disable = + context->bw.dce.nbp_state_change_enable == false; + pp_display_cfg->cpu_cc6_disable = + context->bw.dce.cpuc_state_change_enable == false; + pp_display_cfg->cpu_pstate_disable = + context->bw.dce.cpup_state_change_enable == false; + pp_display_cfg->cpu_pstate_separation_time = + context->bw.dce.blackout_recovery_time_us; + + pp_display_cfg->min_memory_clock_khz = context->bw.dce.yclk_khz + / MEMORY_TYPE_MULTIPLIER_CZ; + + pp_display_cfg->min_engine_clock_khz = determine_sclk_from_bounding_box( + dc, + context->bw.dce.sclk_khz); + + pp_display_cfg->min_engine_clock_deep_sleep_khz + = context->bw.dce.sclk_deep_sleep_khz; + + pp_display_cfg->avail_mclk_switch_time_us = + dce110_get_min_vblank_time_us(context); + /* TODO: dce11.2*/ + pp_display_cfg->avail_mclk_switch_time_in_disp_active_us = 0; + + pp_display_cfg->disp_clk_khz = dc->res_pool->clk_mgr->clks.dispclk_khz; + + dce110_fill_display_configs(context, pp_display_cfg); + + /* TODO: is this still applicable?*/ + if (pp_display_cfg->display_count == 1) { + const struct dc_crtc_timing *timing = + &context->streams[0]->timing; + + pp_display_cfg->crtc_index = + pp_display_cfg->disp_configs[0].pipe_idx; + pp_display_cfg->line_time_in_us = timing->h_total * 1000 / timing->pix_clk_khz; + } + + if (memcmp(&dc->current_state->pp_display_cfg, pp_display_cfg, sizeof(*pp_display_cfg)) != 0) + dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg); +} + +static void dce_update_clocks(struct clk_mgr *clk_mgr, + struct dc_state *context, + bool safe_to_lower) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + struct dm_pp_power_level_change_request level_change_req; + int unpatched_disp_clk = context->bw.dce.dispclk_khz; + + /*TODO: W/A for dal3 linux, investigate why this works */ + if (!clk_mgr_dce->dfs_bypass_active) + context->bw.dce.dispclk_khz = context->bw.dce.dispclk_khz * 115 / 100; + + level_change_req.power_level = dce_get_required_clocks_state(clk_mgr, context); + /* get max clock state from PPLIB */ + if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower) + || level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) { + if (dm_pp_apply_power_level_change_request(clk_mgr->ctx, &level_change_req)) + clk_mgr_dce->cur_min_clks_state = level_change_req.power_level; + } + + if (should_set_clock(safe_to_lower, context->bw.dce.dispclk_khz, clk_mgr->clks.dispclk_khz)) { + context->bw.dce.dispclk_khz = dce_set_clock(clk_mgr, context->bw.dce.dispclk_khz); + clk_mgr->clks.dispclk_khz = context->bw.dce.dispclk_khz; + } + dce_pplib_apply_display_requirements(clk_mgr->ctx->dc, context); + + context->bw.dce.dispclk_khz = unpatched_disp_clk; +} + +static void dce11_update_clocks(struct clk_mgr *clk_mgr, + struct dc_state *context, + bool safe_to_lower) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + struct dm_pp_power_level_change_request level_change_req; + + level_change_req.power_level = dce_get_required_clocks_state(clk_mgr, context); + /* get max clock state from PPLIB */ + if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower) + || level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) { + if (dm_pp_apply_power_level_change_request(clk_mgr->ctx, &level_change_req)) + clk_mgr_dce->cur_min_clks_state = level_change_req.power_level; + } + + if (should_set_clock(safe_to_lower, context->bw.dce.dispclk_khz, clk_mgr->clks.dispclk_khz)) { + context->bw.dce.dispclk_khz = dce_set_clock(clk_mgr, context->bw.dce.dispclk_khz); + clk_mgr->clks.dispclk_khz = context->bw.dce.dispclk_khz; + } + dce11_pplib_apply_display_requirements(clk_mgr->ctx->dc, context); +} + +static void dce112_update_clocks(struct clk_mgr *clk_mgr, + struct dc_state *context, + bool safe_to_lower) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + struct dm_pp_power_level_change_request level_change_req; + + level_change_req.power_level = dce_get_required_clocks_state(clk_mgr, context); + /* get max clock state from PPLIB */ + if ((level_change_req.power_level < clk_mgr_dce->cur_min_clks_state && safe_to_lower) + || level_change_req.power_level > clk_mgr_dce->cur_min_clks_state) { + if (dm_pp_apply_power_level_change_request(clk_mgr->ctx, &level_change_req)) + clk_mgr_dce->cur_min_clks_state = level_change_req.power_level; + } + + if (should_set_clock(safe_to_lower, context->bw.dce.dispclk_khz, clk_mgr->clks.dispclk_khz)) { + context->bw.dce.dispclk_khz = dce112_set_clock(clk_mgr, context->bw.dce.dispclk_khz); + clk_mgr->clks.dispclk_khz = context->bw.dce.dispclk_khz; + } + dce11_pplib_apply_display_requirements(clk_mgr->ctx->dc, context); +} + +static void dce12_update_clocks(struct clk_mgr *clk_mgr, + struct dc_state *context, + bool safe_to_lower) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(clk_mgr); + struct dm_pp_clock_for_voltage_req clock_voltage_req = {0}; + int max_pix_clk = get_max_pixel_clock_for_all_paths(context); + int unpatched_disp_clk = context->bw.dce.dispclk_khz; + + /*TODO: W/A for dal3 linux, investigate why this works */ + if (!clk_mgr_dce->dfs_bypass_active) + context->bw.dce.dispclk_khz = context->bw.dce.dispclk_khz * 115 / 100; + + if (should_set_clock(safe_to_lower, context->bw.dce.dispclk_khz, clk_mgr->clks.dispclk_khz)) { + clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAY_CLK; + clock_voltage_req.clocks_in_khz = context->bw.dce.dispclk_khz; + context->bw.dce.dispclk_khz = dce112_set_clock(clk_mgr, context->bw.dce.dispclk_khz); + clk_mgr->clks.dispclk_khz = context->bw.dce.dispclk_khz; + + dm_pp_apply_clock_for_voltage_request(clk_mgr->ctx, &clock_voltage_req); + } + + if (should_set_clock(safe_to_lower, max_pix_clk, clk_mgr->clks.phyclk_khz)) { + clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAYPHYCLK; + clock_voltage_req.clocks_in_khz = max_pix_clk; + clk_mgr->clks.phyclk_khz = max_pix_clk; + + dm_pp_apply_clock_for_voltage_request(clk_mgr->ctx, &clock_voltage_req); + } + dce11_pplib_apply_display_requirements(clk_mgr->ctx->dc, context); + + context->bw.dce.dispclk_khz = unpatched_disp_clk; +} + +static const struct clk_mgr_funcs dce120_funcs = { + .get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz, + .update_clocks = dce12_update_clocks +}; + +static const struct clk_mgr_funcs dce112_funcs = { + .get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz, + .update_clocks = dce112_update_clocks +}; + +static const struct clk_mgr_funcs dce110_funcs = { + .get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz, + .update_clocks = dce11_update_clocks, +}; + +static const struct clk_mgr_funcs dce_funcs = { + .get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz, + .update_clocks = dce_update_clocks +}; + +static void dce_clk_mgr_construct( + struct dce_clk_mgr *clk_mgr_dce, + struct dc_context *ctx, + const struct clk_mgr_registers *regs, + const struct clk_mgr_shift *clk_shift, + const struct clk_mgr_mask *clk_mask) +{ + struct clk_mgr *base = &clk_mgr_dce->base; + struct dm_pp_static_clock_info static_clk_info = {0}; + + base->ctx = ctx; + base->funcs = &dce_funcs; + + clk_mgr_dce->regs = regs; + clk_mgr_dce->clk_mgr_shift = clk_shift; + clk_mgr_dce->clk_mgr_mask = clk_mask; + + clk_mgr_dce->dfs_bypass_disp_clk = 0; + + clk_mgr_dce->dprefclk_ss_percentage = 0; + clk_mgr_dce->dprefclk_ss_divider = 1000; + clk_mgr_dce->ss_on_dprefclk = false; + + + if (dm_pp_get_static_clocks(ctx, &static_clk_info)) + clk_mgr_dce->max_clks_state = static_clk_info.max_clocks_state; + else + clk_mgr_dce->max_clks_state = DM_PP_CLOCKS_STATE_NOMINAL; + clk_mgr_dce->cur_min_clks_state = DM_PP_CLOCKS_STATE_INVALID; + + dce_clock_read_integrated_info(clk_mgr_dce); + dce_clock_read_ss_info(clk_mgr_dce); +} + +struct clk_mgr *dce_clk_mgr_create( + struct dc_context *ctx, + const struct clk_mgr_registers *regs, + const struct clk_mgr_shift *clk_shift, + const struct clk_mgr_mask *clk_mask) +{ + struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL); + + if (clk_mgr_dce == NULL) { + BREAK_TO_DEBUGGER(); + return NULL; + } + + memcpy(clk_mgr_dce->max_clks_by_state, + dce80_max_clks_by_state, + sizeof(dce80_max_clks_by_state)); + + dce_clk_mgr_construct( + clk_mgr_dce, ctx, regs, clk_shift, clk_mask); + + return &clk_mgr_dce->base; +} + +struct clk_mgr *dce110_clk_mgr_create( + struct dc_context *ctx, + const struct clk_mgr_registers *regs, + const struct clk_mgr_shift *clk_shift, + const struct clk_mgr_mask *clk_mask) +{ + struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL); + + if (clk_mgr_dce == NULL) { + BREAK_TO_DEBUGGER(); + return NULL; + } + + memcpy(clk_mgr_dce->max_clks_by_state, + dce110_max_clks_by_state, + sizeof(dce110_max_clks_by_state)); + + dce_clk_mgr_construct( + clk_mgr_dce, ctx, regs, clk_shift, clk_mask); + + clk_mgr_dce->base.funcs = &dce110_funcs; + + return &clk_mgr_dce->base; +} + +struct clk_mgr *dce112_clk_mgr_create( + struct dc_context *ctx, + const struct clk_mgr_registers *regs, + const struct clk_mgr_shift *clk_shift, + const struct clk_mgr_mask *clk_mask) +{ + struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL); + + if (clk_mgr_dce == NULL) { + BREAK_TO_DEBUGGER(); + return NULL; + } + + memcpy(clk_mgr_dce->max_clks_by_state, + dce112_max_clks_by_state, + sizeof(dce112_max_clks_by_state)); + + dce_clk_mgr_construct( + clk_mgr_dce, ctx, regs, clk_shift, clk_mask); + + clk_mgr_dce->base.funcs = &dce112_funcs; + + return &clk_mgr_dce->base; +} + +struct clk_mgr *dce120_clk_mgr_create(struct dc_context *ctx) +{ + struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL); + + if (clk_mgr_dce == NULL) { + BREAK_TO_DEBUGGER(); + return NULL; + } + + memcpy(clk_mgr_dce->max_clks_by_state, + dce120_max_clks_by_state, + sizeof(dce120_max_clks_by_state)); + + dce_clk_mgr_construct( + clk_mgr_dce, ctx, NULL, NULL, NULL); + + clk_mgr_dce->dprefclk_khz = 600000; + clk_mgr_dce->base.funcs = &dce120_funcs; + + return &clk_mgr_dce->base; +} + +void dce_clk_mgr_destroy(struct clk_mgr **clk_mgr) +{ + struct dce_clk_mgr *clk_mgr_dce = TO_DCE_CLK_MGR(*clk_mgr); + + kfree(clk_mgr_dce); + *clk_mgr = NULL; +} diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.h b/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.h index 34fdb386c884..046077797416 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_clk_mgr.h @@ -24,10 +24,13 @@ */ -#ifndef _DCE_CLOCKS_H_ -#define _DCE_CLOCKS_H_ +#ifndef _DCE_CLK_MGR_H_ +#define _DCE_CLK_MGR_H_ -#include "display_clock.h" +#include "clk_mgr.h" +#include "dccg.h" + +#define MEMORY_TYPE_MULTIPLIER_CZ 4 #define CLK_COMMON_REG_LIST_DCE_BASE() \ .DPREFCLK_CNTL = mmDPREFCLK_CNTL, \ @@ -53,24 +56,31 @@ type DENTIST_DISPCLK_WDIVIDER; \ type DENTIST_DISPCLK_CHG_DONE; -struct dccg_shift { +struct clk_mgr_shift { CLK_REG_FIELD_LIST(uint8_t) }; -struct dccg_mask { +struct clk_mgr_mask { CLK_REG_FIELD_LIST(uint32_t) }; -struct dccg_registers { +struct clk_mgr_registers { uint32_t DPREFCLK_CNTL; uint32_t DENTIST_DISPCLK_CNTL; }; -struct dce_dccg { - struct dccg base; - const struct dccg_registers *regs; - const struct dccg_shift *clk_shift; - const struct dccg_mask *clk_mask; +struct state_dependent_clocks { + int display_clk_khz; + int pixel_clk_khz; +}; + +struct dce_clk_mgr { + struct clk_mgr base; + const struct clk_mgr_registers *regs; + const struct clk_mgr_shift *clk_mgr_shift; + const struct clk_mgr_mask *clk_mgr_mask; + + struct dccg *dccg; struct state_dependent_clocks max_clks_by_state[DM_PP_CLOCKS_MAX_STATES]; @@ -91,33 +101,68 @@ struct dce_dccg { /* DPREFCLK SS percentage Divider (100 or 1000) */ int dprefclk_ss_divider; int dprefclk_khz; + + enum dm_pp_clocks_state max_clks_state; + enum dm_pp_clocks_state cur_min_clks_state; }; +/* Starting DID for each range */ +enum dentist_base_divider_id { + DENTIST_BASE_DID_1 = 0x08, + DENTIST_BASE_DID_2 = 0x40, + DENTIST_BASE_DID_3 = 0x60, + DENTIST_BASE_DID_4 = 0x7e, + DENTIST_MAX_DID = 0x7f +}; -struct dccg *dce_dccg_create( - struct dc_context *ctx, - const struct dccg_registers *regs, - const struct dccg_shift *clk_shift, - const struct dccg_mask *clk_mask); +/* Starting point and step size for each divider range.*/ +enum dentist_divider_range { + DENTIST_DIVIDER_RANGE_1_START = 8, /* 2.00 */ + DENTIST_DIVIDER_RANGE_1_STEP = 1, /* 0.25 */ + DENTIST_DIVIDER_RANGE_2_START = 64, /* 16.00 */ + DENTIST_DIVIDER_RANGE_2_STEP = 2, /* 0.50 */ + DENTIST_DIVIDER_RANGE_3_START = 128, /* 32.00 */ + DENTIST_DIVIDER_RANGE_3_STEP = 4, /* 1.00 */ + DENTIST_DIVIDER_RANGE_4_START = 248, /* 62.00 */ + DENTIST_DIVIDER_RANGE_4_STEP = 264, /* 66.00 */ + DENTIST_DIVIDER_RANGE_SCALE_FACTOR = 4 +}; + +static inline bool should_set_clock(bool safe_to_lower, int calc_clk, int cur_clk) +{ + return ((safe_to_lower && calc_clk < cur_clk) || calc_clk > cur_clk); +} + +void dce_clock_read_ss_info(struct dce_clk_mgr *dccg_dce); + +int dce12_get_dp_ref_freq_khz(struct clk_mgr *dccg); -struct dccg *dce110_dccg_create( +void dce110_fill_display_configs( + const struct dc_state *context, + struct dm_pp_display_configuration *pp_display_cfg); + +int dce112_set_clock(struct clk_mgr *dccg, int requested_clk_khz); + +struct clk_mgr *dce_clk_mgr_create( struct dc_context *ctx, - const struct dccg_registers *regs, - const struct dccg_shift *clk_shift, - const struct dccg_mask *clk_mask); + const struct clk_mgr_registers *regs, + const struct clk_mgr_shift *clk_shift, + const struct clk_mgr_mask *clk_mask); -struct dccg *dce112_dccg_create( +struct clk_mgr *dce110_clk_mgr_create( struct dc_context *ctx, - const struct dccg_registers *regs, - const struct dccg_shift *clk_shift, - const struct dccg_mask *clk_mask); + const struct clk_mgr_registers *regs, + const struct clk_mgr_shift *clk_shift, + const struct clk_mgr_mask *clk_mask); -struct dccg *dce120_dccg_create(struct dc_context *ctx); +struct clk_mgr *dce112_clk_mgr_create( + struct dc_context *ctx, + const struct clk_mgr_registers *regs, + const struct clk_mgr_shift *clk_shift, + const struct clk_mgr_mask *clk_mask); -#ifdef CONFIG_DRM_AMD_DC_DCN1_0 -struct dccg *dcn1_dccg_create(struct dc_context *ctx); -#endif +struct clk_mgr *dce120_clk_mgr_create(struct dc_context *ctx); -void dce_dccg_destroy(struct dccg **dccg); +void dce_clk_mgr_destroy(struct clk_mgr **clk_mgr); -#endif /* _DCE_CLOCKS_H_ */ +#endif /* _DCE_CLK_MGR_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c b/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c deleted file mode 100644 index d89a097ba936..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_clocks.c +++ /dev/null @@ -1,947 +0,0 @@ -/* - * Copyright 2012-16 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. - * - * Authors: AMD - * - */ - -#include "dce_clocks.h" -#include "dm_services.h" -#include "reg_helper.h" -#include "fixed31_32.h" -#include "bios_parser_interface.h" -#include "dc.h" -#include "dmcu.h" -#if defined(CONFIG_DRM_AMD_DC_DCN1_0) -#include "dcn_calcs.h" -#endif -#include "core_types.h" -#include "dc_types.h" -#include "dal_asic_id.h" - -#define TO_DCE_CLOCKS(clocks)\ - container_of(clocks, struct dce_dccg, base) - -#define REG(reg) \ - (clk_dce->regs->reg) - -#undef FN -#define FN(reg_name, field_name) \ - clk_dce->clk_shift->field_name, clk_dce->clk_mask->field_name - -#define CTX \ - clk_dce->base.ctx -#define DC_LOGGER \ - clk->ctx->logger - -/* Max clock values for each state indexed by "enum clocks_state": */ -static const struct state_dependent_clocks dce80_max_clks_by_state[] = { -/* ClocksStateInvalid - should not be used */ -{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, -/* ClocksStateUltraLow - not expected to be used for DCE 8.0 */ -{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, -/* ClocksStateLow */ -{ .display_clk_khz = 352000, .pixel_clk_khz = 330000}, -/* ClocksStateNominal */ -{ .display_clk_khz = 600000, .pixel_clk_khz = 400000 }, -/* ClocksStatePerformance */ -{ .display_clk_khz = 600000, .pixel_clk_khz = 400000 } }; - -static const struct state_dependent_clocks dce110_max_clks_by_state[] = { -/*ClocksStateInvalid - should not be used*/ -{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, -/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/ -{ .display_clk_khz = 352000, .pixel_clk_khz = 330000 }, -/*ClocksStateLow*/ -{ .display_clk_khz = 352000, .pixel_clk_khz = 330000 }, -/*ClocksStateNominal*/ -{ .display_clk_khz = 467000, .pixel_clk_khz = 400000 }, -/*ClocksStatePerformance*/ -{ .display_clk_khz = 643000, .pixel_clk_khz = 400000 } }; - -static const struct state_dependent_clocks dce112_max_clks_by_state[] = { -/*ClocksStateInvalid - should not be used*/ -{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, -/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/ -{ .display_clk_khz = 389189, .pixel_clk_khz = 346672 }, -/*ClocksStateLow*/ -{ .display_clk_khz = 459000, .pixel_clk_khz = 400000 }, -/*ClocksStateNominal*/ -{ .display_clk_khz = 667000, .pixel_clk_khz = 600000 }, -/*ClocksStatePerformance*/ -{ .display_clk_khz = 1132000, .pixel_clk_khz = 600000 } }; - -static const struct state_dependent_clocks dce120_max_clks_by_state[] = { -/*ClocksStateInvalid - should not be used*/ -{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, -/*ClocksStateUltraLow - currently by HW design team not supposed to be used*/ -{ .display_clk_khz = 0, .pixel_clk_khz = 0 }, -/*ClocksStateLow*/ -{ .display_clk_khz = 460000, .pixel_clk_khz = 400000 }, -/*ClocksStateNominal*/ -{ .display_clk_khz = 670000, .pixel_clk_khz = 600000 }, -/*ClocksStatePerformance*/ -{ .display_clk_khz = 1133000, .pixel_clk_khz = 600000 } }; - -/* Starting DID for each range */ -enum dentist_base_divider_id { - DENTIST_BASE_DID_1 = 0x08, - DENTIST_BASE_DID_2 = 0x40, - DENTIST_BASE_DID_3 = 0x60, - DENTIST_BASE_DID_4 = 0x7e, - DENTIST_MAX_DID = 0x7f -}; - -/* Starting point and step size for each divider range.*/ -enum dentist_divider_range { - DENTIST_DIVIDER_RANGE_1_START = 8, /* 2.00 */ - DENTIST_DIVIDER_RANGE_1_STEP = 1, /* 0.25 */ - DENTIST_DIVIDER_RANGE_2_START = 64, /* 16.00 */ - DENTIST_DIVIDER_RANGE_2_STEP = 2, /* 0.50 */ - DENTIST_DIVIDER_RANGE_3_START = 128, /* 32.00 */ - DENTIST_DIVIDER_RANGE_3_STEP = 4, /* 1.00 */ - DENTIST_DIVIDER_RANGE_4_START = 248, /* 62.00 */ - DENTIST_DIVIDER_RANGE_4_STEP = 264, /* 66.00 */ - DENTIST_DIVIDER_RANGE_SCALE_FACTOR = 4 -}; - -static int dentist_get_divider_from_did(int did) -{ - if (did < DENTIST_BASE_DID_1) - did = DENTIST_BASE_DID_1; - if (did > DENTIST_MAX_DID) - did = DENTIST_MAX_DID; - - if (did < DENTIST_BASE_DID_2) { - return DENTIST_DIVIDER_RANGE_1_START + DENTIST_DIVIDER_RANGE_1_STEP - * (did - DENTIST_BASE_DID_1); - } else if (did < DENTIST_BASE_DID_3) { - return DENTIST_DIVIDER_RANGE_2_START + DENTIST_DIVIDER_RANGE_2_STEP - * (did - DENTIST_BASE_DID_2); - } else if (did < DENTIST_BASE_DID_4) { - return DENTIST_DIVIDER_RANGE_3_START + DENTIST_DIVIDER_RANGE_3_STEP - * (did - DENTIST_BASE_DID_3); - } else { - return DENTIST_DIVIDER_RANGE_4_START + DENTIST_DIVIDER_RANGE_4_STEP - * (did - DENTIST_BASE_DID_4); - } -} - -/* SW will adjust DP REF Clock average value for all purposes - * (DP DTO / DP Audio DTO and DP GTC) - if clock is spread for all cases: - -if SS enabled on DP Ref clock and HW de-spreading enabled with SW - calculations for DS_INCR/DS_MODULO (this is planned to be default case) - -if SS enabled on DP Ref clock and HW de-spreading enabled with HW - calculations (not planned to be used, but average clock should still - be valid) - -if SS enabled on DP Ref clock and HW de-spreading disabled - (should not be case with CIK) then SW should program all rates - generated according to average value (case as with previous ASICs) - */ -static int dccg_adjust_dp_ref_freq_for_ss(struct dce_dccg *clk_dce, int dp_ref_clk_khz) -{ - if (clk_dce->ss_on_dprefclk && clk_dce->dprefclk_ss_divider != 0) { - struct fixed31_32 ss_percentage = dc_fixpt_div_int( - dc_fixpt_from_fraction(clk_dce->dprefclk_ss_percentage, - clk_dce->dprefclk_ss_divider), 200); - struct fixed31_32 adj_dp_ref_clk_khz; - - ss_percentage = dc_fixpt_sub(dc_fixpt_one, ss_percentage); - adj_dp_ref_clk_khz = dc_fixpt_mul_int(ss_percentage, dp_ref_clk_khz); - dp_ref_clk_khz = dc_fixpt_floor(adj_dp_ref_clk_khz); - } - return dp_ref_clk_khz; -} - -static int dce_get_dp_ref_freq_khz(struct dccg *clk) -{ - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(clk); - int dprefclk_wdivider; - int dprefclk_src_sel; - int dp_ref_clk_khz = 600000; - int target_div; - - /* ASSERT DP Reference Clock source is from DFS*/ - REG_GET(DPREFCLK_CNTL, DPREFCLK_SRC_SEL, &dprefclk_src_sel); - ASSERT(dprefclk_src_sel == 0); - - /* Read the mmDENTIST_DISPCLK_CNTL to get the currently - * programmed DID DENTIST_DPREFCLK_WDIVIDER*/ - REG_GET(DENTIST_DISPCLK_CNTL, DENTIST_DPREFCLK_WDIVIDER, &dprefclk_wdivider); - - /* Convert DENTIST_DPREFCLK_WDIVIDERto actual divider*/ - target_div = dentist_get_divider_from_did(dprefclk_wdivider); - - /* Calculate the current DFS clock, in kHz.*/ - dp_ref_clk_khz = (DENTIST_DIVIDER_RANGE_SCALE_FACTOR - * clk_dce->dentist_vco_freq_khz) / target_div; - - return dccg_adjust_dp_ref_freq_for_ss(clk_dce, dp_ref_clk_khz); -} - -static int dce12_get_dp_ref_freq_khz(struct dccg *clk) -{ - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(clk); - - return dccg_adjust_dp_ref_freq_for_ss(clk_dce, clk_dce->dprefclk_khz); -} - -static enum dm_pp_clocks_state dce_get_required_clocks_state( - struct dccg *clk, - struct dc_clocks *req_clocks) -{ - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(clk); - int i; - enum dm_pp_clocks_state low_req_clk; - - /* Iterate from highest supported to lowest valid state, and update - * lowest RequiredState with the lowest state that satisfies - * all required clocks - */ - for (i = clk->max_clks_state; i >= DM_PP_CLOCKS_STATE_ULTRA_LOW; i--) - if (req_clocks->dispclk_khz > - clk_dce->max_clks_by_state[i].display_clk_khz - || req_clocks->phyclk_khz > - clk_dce->max_clks_by_state[i].pixel_clk_khz) - break; - - low_req_clk = i + 1; - if (low_req_clk > clk->max_clks_state) { - /* set max clock state for high phyclock, invalid on exceeding display clock */ - if (clk_dce->max_clks_by_state[clk->max_clks_state].display_clk_khz - < req_clocks->dispclk_khz) - low_req_clk = DM_PP_CLOCKS_STATE_INVALID; - else - low_req_clk = clk->max_clks_state; - } - - return low_req_clk; -} - -static int dce_set_clock( - struct dccg *clk, - int requested_clk_khz) -{ - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(clk); - struct bp_pixel_clock_parameters pxl_clk_params = { 0 }; - struct dc_bios *bp = clk->ctx->dc_bios; - int actual_clock = requested_clk_khz; - - /* Make sure requested clock isn't lower than minimum threshold*/ - if (requested_clk_khz > 0) - requested_clk_khz = max(requested_clk_khz, - clk_dce->dentist_vco_freq_khz / 64); - - /* Prepare to program display clock*/ - pxl_clk_params.target_pixel_clock = requested_clk_khz; - pxl_clk_params.pll_id = CLOCK_SOURCE_ID_DFS; - - if (clk_dce->dfs_bypass_active) - pxl_clk_params.flags.SET_DISPCLK_DFS_BYPASS = true; - - bp->funcs->program_display_engine_pll(bp, &pxl_clk_params); - - if (clk_dce->dfs_bypass_active) { - /* Cache the fixed display clock*/ - clk_dce->dfs_bypass_disp_clk = - pxl_clk_params.dfs_bypass_display_clock; - actual_clock = pxl_clk_params.dfs_bypass_display_clock; - } - - /* from power down, we need mark the clock state as ClocksStateNominal - * from HWReset, so when resume we will call pplib voltage regulator.*/ - if (requested_clk_khz == 0) - clk->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL; - return actual_clock; -} - -static int dce_psr_set_clock( - struct dccg *clk, - int requested_clk_khz) -{ - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(clk); - struct dc_context *ctx = clk_dce->base.ctx; - struct dc *core_dc = ctx->dc; - struct dmcu *dmcu = core_dc->res_pool->dmcu; - int actual_clk_khz = requested_clk_khz; - - actual_clk_khz = dce_set_clock(clk, requested_clk_khz); - - dmcu->funcs->set_psr_wait_loop(dmcu, actual_clk_khz / 1000 / 7); - return actual_clk_khz; -} - -static int dce112_set_clock( - struct dccg *clk, - int requested_clk_khz) -{ - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(clk); - struct bp_set_dce_clock_parameters dce_clk_params; - struct dc_bios *bp = clk->ctx->dc_bios; - struct dc *core_dc = clk->ctx->dc; - struct dmcu *dmcu = core_dc->res_pool->dmcu; - int actual_clock = requested_clk_khz; - /* Prepare to program display clock*/ - memset(&dce_clk_params, 0, sizeof(dce_clk_params)); - - /* Make sure requested clock isn't lower than minimum threshold*/ - if (requested_clk_khz > 0) - requested_clk_khz = max(requested_clk_khz, - clk_dce->dentist_vco_freq_khz / 62); - - dce_clk_params.target_clock_frequency = requested_clk_khz; - dce_clk_params.pll_id = CLOCK_SOURCE_ID_DFS; - dce_clk_params.clock_type = DCECLOCK_TYPE_DISPLAY_CLOCK; - - bp->funcs->set_dce_clock(bp, &dce_clk_params); - actual_clock = dce_clk_params.target_clock_frequency; - - /* from power down, we need mark the clock state as ClocksStateNominal - * from HWReset, so when resume we will call pplib voltage regulator.*/ - if (requested_clk_khz == 0) - clk->cur_min_clks_state = DM_PP_CLOCKS_STATE_NOMINAL; - - /*Program DP ref Clock*/ - /*VBIOS will determine DPREFCLK frequency, so we don't set it*/ - dce_clk_params.target_clock_frequency = 0; - dce_clk_params.clock_type = DCECLOCK_TYPE_DPREFCLK; - if (!ASICREV_IS_VEGA20_P(clk->ctx->asic_id.hw_internal_rev)) - dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = - (dce_clk_params.pll_id == - CLOCK_SOURCE_COMBO_DISPLAY_PLL0); - else - dce_clk_params.flags.USE_GENLOCK_AS_SOURCE_FOR_DPREFCLK = false; - - bp->funcs->set_dce_clock(bp, &dce_clk_params); - - if (!IS_FPGA_MAXIMUS_DC(core_dc->ctx->dce_environment)) { - if (clk_dce->dfs_bypass_disp_clk != actual_clock) - dmcu->funcs->set_psr_wait_loop(dmcu, - actual_clock / 1000 / 7); - } - - clk_dce->dfs_bypass_disp_clk = actual_clock; - return actual_clock; -} - -static void dce_clock_read_integrated_info(struct dce_dccg *clk_dce) -{ - struct dc_debug_options *debug = &clk_dce->base.ctx->dc->debug; - struct dc_bios *bp = clk_dce->base.ctx->dc_bios; - struct integrated_info info = { { { 0 } } }; - struct dc_firmware_info fw_info = { { 0 } }; - int i; - - if (bp->integrated_info) - info = *bp->integrated_info; - - clk_dce->dentist_vco_freq_khz = info.dentist_vco_freq; - if (clk_dce->dentist_vco_freq_khz == 0) { - bp->funcs->get_firmware_info(bp, &fw_info); - clk_dce->dentist_vco_freq_khz = - fw_info.smu_gpu_pll_output_freq; - if (clk_dce->dentist_vco_freq_khz == 0) - clk_dce->dentist_vco_freq_khz = 3600000; - } - - /*update the maximum display clock for each power state*/ - for (i = 0; i < NUMBER_OF_DISP_CLK_VOLTAGE; ++i) { - enum dm_pp_clocks_state clk_state = DM_PP_CLOCKS_STATE_INVALID; - - switch (i) { - case 0: - clk_state = DM_PP_CLOCKS_STATE_ULTRA_LOW; - break; - - case 1: - clk_state = DM_PP_CLOCKS_STATE_LOW; - break; - - case 2: - clk_state = DM_PP_CLOCKS_STATE_NOMINAL; - break; - - case 3: - clk_state = DM_PP_CLOCKS_STATE_PERFORMANCE; - break; - - default: - clk_state = DM_PP_CLOCKS_STATE_INVALID; - break; - } - - /*Do not allow bad VBIOS/SBIOS to override with invalid values, - * check for > 100MHz*/ - if (info.disp_clk_voltage[i].max_supported_clk >= 100000) - clk_dce->max_clks_by_state[clk_state].display_clk_khz = - info.disp_clk_voltage[i].max_supported_clk; - } - - if (!debug->disable_dfs_bypass && bp->integrated_info) - if (bp->integrated_info->gpu_cap_info & DFS_BYPASS_ENABLE) - clk_dce->dfs_bypass_enabled = true; -} - -static void dce_clock_read_ss_info(struct dce_dccg *clk_dce) -{ - struct dc_bios *bp = clk_dce->base.ctx->dc_bios; - int ss_info_num = bp->funcs->get_ss_entry_number( - bp, AS_SIGNAL_TYPE_GPU_PLL); - - if (ss_info_num) { - struct spread_spectrum_info info = { { 0 } }; - enum bp_result result = bp->funcs->get_spread_spectrum_info( - bp, AS_SIGNAL_TYPE_GPU_PLL, 0, &info); - - /* Based on VBIOS, VBIOS will keep entry for GPU PLL SS - * even if SS not enabled and in that case - * SSInfo.spreadSpectrumPercentage !=0 would be sign - * that SS is enabled - */ - if (result == BP_RESULT_OK && - info.spread_spectrum_percentage != 0) { - clk_dce->ss_on_dprefclk = true; - clk_dce->dprefclk_ss_divider = info.spread_percentage_divider; - - if (info.type.CENTER_MODE == 0) { - /* TODO: Currently for DP Reference clock we - * need only SS percentage for - * downspread */ - clk_dce->dprefclk_ss_percentage = - info.spread_spectrum_percentage; - } - - return; - } - - result = bp->funcs->get_spread_spectrum_info( - bp, AS_SIGNAL_TYPE_DISPLAY_PORT, 0, &info); - - /* Based on VBIOS, VBIOS will keep entry for DPREFCLK SS - * even if SS not enabled and in that case - * SSInfo.spreadSpectrumPercentage !=0 would be sign - * that SS is enabled - */ - if (result == BP_RESULT_OK && - info.spread_spectrum_percentage != 0) { - clk_dce->ss_on_dprefclk = true; - clk_dce->dprefclk_ss_divider = info.spread_percentage_divider; - - if (info.type.CENTER_MODE == 0) { - /* Currently for DP Reference clock we - * need only SS percentage for - * downspread */ - clk_dce->dprefclk_ss_percentage = - info.spread_spectrum_percentage; - } - } - } -} - -static inline bool should_set_clock(bool safe_to_lower, int calc_clk, int cur_clk) -{ - return ((safe_to_lower && calc_clk < cur_clk) || calc_clk > cur_clk); -} - -static void dce12_update_clocks(struct dccg *dccg, - struct dc_clocks *new_clocks, - bool safe_to_lower) -{ - struct dm_pp_clock_for_voltage_req clock_voltage_req = {0}; - - /* TODO: Investigate why this is needed to fix display corruption. */ - new_clocks->dispclk_khz = new_clocks->dispclk_khz * 115 / 100; - - if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, dccg->clks.dispclk_khz)) { - clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAY_CLK; - clock_voltage_req.clocks_in_khz = new_clocks->dispclk_khz; - new_clocks->dispclk_khz = dccg->funcs->set_dispclk(dccg, new_clocks->dispclk_khz); - dccg->clks.dispclk_khz = new_clocks->dispclk_khz; - - dm_pp_apply_clock_for_voltage_request(dccg->ctx, &clock_voltage_req); - } - - if (should_set_clock(safe_to_lower, new_clocks->phyclk_khz, dccg->clks.phyclk_khz)) { - clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DISPLAYPHYCLK; - clock_voltage_req.clocks_in_khz = new_clocks->phyclk_khz; - dccg->clks.phyclk_khz = new_clocks->phyclk_khz; - - dm_pp_apply_clock_for_voltage_request(dccg->ctx, &clock_voltage_req); - } -} - -#ifdef CONFIG_DRM_AMD_DC_DCN1_0 -static int dcn1_determine_dppclk_threshold(struct dccg *dccg, struct dc_clocks *new_clocks) -{ - bool request_dpp_div = new_clocks->dispclk_khz > new_clocks->dppclk_khz; - bool dispclk_increase = new_clocks->dispclk_khz > dccg->clks.dispclk_khz; - int disp_clk_threshold = new_clocks->max_supported_dppclk_khz; - bool cur_dpp_div = dccg->clks.dispclk_khz > dccg->clks.dppclk_khz; - - /* increase clock, looking for div is 0 for current, request div is 1*/ - if (dispclk_increase) { - /* already divided by 2, no need to reach target clk with 2 steps*/ - if (cur_dpp_div) - return new_clocks->dispclk_khz; - - /* request disp clk is lower than maximum supported dpp clk, - * no need to reach target clk with two steps. - */ - if (new_clocks->dispclk_khz <= disp_clk_threshold) - return new_clocks->dispclk_khz; - - /* target dpp clk not request divided by 2, still within threshold */ - if (!request_dpp_div) - return new_clocks->dispclk_khz; - - } else { - /* decrease clock, looking for current dppclk divided by 2, - * request dppclk not divided by 2. - */ - - /* current dpp clk not divided by 2, no need to ramp*/ - if (!cur_dpp_div) - return new_clocks->dispclk_khz; - - /* current disp clk is lower than current maximum dpp clk, - * no need to ramp - */ - if (dccg->clks.dispclk_khz <= disp_clk_threshold) - return new_clocks->dispclk_khz; - - /* request dpp clk need to be divided by 2 */ - if (request_dpp_div) - return new_clocks->dispclk_khz; - } - - return disp_clk_threshold; -} - -static void dcn1_ramp_up_dispclk_with_dpp(struct dccg *dccg, struct dc_clocks *new_clocks) -{ - struct dc *dc = dccg->ctx->dc; - int dispclk_to_dpp_threshold = dcn1_determine_dppclk_threshold(dccg, new_clocks); - bool request_dpp_div = new_clocks->dispclk_khz > new_clocks->dppclk_khz; - int i; - - /* set disp clk to dpp clk threshold */ - dccg->funcs->set_dispclk(dccg, dispclk_to_dpp_threshold); - - /* update request dpp clk division option */ - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; - - if (!pipe_ctx->plane_state) - continue; - - pipe_ctx->plane_res.dpp->funcs->dpp_dppclk_control( - pipe_ctx->plane_res.dpp, - request_dpp_div, - true); - } - - /* If target clk not same as dppclk threshold, set to target clock */ - if (dispclk_to_dpp_threshold != new_clocks->dispclk_khz) - dccg->funcs->set_dispclk(dccg, new_clocks->dispclk_khz); - - dccg->clks.dispclk_khz = new_clocks->dispclk_khz; - dccg->clks.dppclk_khz = new_clocks->dppclk_khz; - dccg->clks.max_supported_dppclk_khz = new_clocks->max_supported_dppclk_khz; -} - -static void dcn1_update_clocks(struct dccg *dccg, - struct dc_clocks *new_clocks, - bool safe_to_lower) -{ - struct dc *dc = dccg->ctx->dc; - struct pp_smu_display_requirement_rv *smu_req_cur = - &dc->res_pool->pp_smu_req; - struct pp_smu_display_requirement_rv smu_req = *smu_req_cur; - struct pp_smu_funcs_rv *pp_smu = dc->res_pool->pp_smu; - struct dm_pp_clock_for_voltage_req clock_voltage_req = {0}; - bool send_request_to_increase = false; - bool send_request_to_lower = false; - - if (new_clocks->phyclk_khz) - smu_req.display_count = 1; - else - smu_req.display_count = 0; - - if (new_clocks->dispclk_khz > dccg->clks.dispclk_khz - || new_clocks->phyclk_khz > dccg->clks.phyclk_khz - || new_clocks->fclk_khz > dccg->clks.fclk_khz - || new_clocks->dcfclk_khz > dccg->clks.dcfclk_khz) - send_request_to_increase = true; - - if (should_set_clock(safe_to_lower, new_clocks->phyclk_khz, dccg->clks.phyclk_khz)) { - dccg->clks.phyclk_khz = new_clocks->phyclk_khz; - - send_request_to_lower = true; - } - - if (should_set_clock(safe_to_lower, new_clocks->fclk_khz, dccg->clks.fclk_khz)) { - dccg->clks.fclk_khz = new_clocks->fclk_khz; - clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_FCLK; - clock_voltage_req.clocks_in_khz = new_clocks->fclk_khz; - smu_req.hard_min_fclk_khz = new_clocks->fclk_khz; - - dm_pp_apply_clock_for_voltage_request(dccg->ctx, &clock_voltage_req); - send_request_to_lower = true; - } - - if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, dccg->clks.dcfclk_khz)) { - dccg->clks.dcfclk_khz = new_clocks->dcfclk_khz; - smu_req.hard_min_dcefclk_khz = new_clocks->dcfclk_khz; - - send_request_to_lower = true; - } - - if (should_set_clock(safe_to_lower, - new_clocks->dcfclk_deep_sleep_khz, dccg->clks.dcfclk_deep_sleep_khz)) { - dccg->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz; - smu_req.min_deep_sleep_dcefclk_mhz = new_clocks->dcfclk_deep_sleep_khz; - - send_request_to_lower = true; - } - - /* make sure dcf clk is before dpp clk to - * make sure we have enough voltage to run dpp clk - */ - if (send_request_to_increase) { - /*use dcfclk to request voltage*/ - clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DCFCLK; - clock_voltage_req.clocks_in_khz = dcn_find_dcfclk_suits_all(dc, new_clocks); - dm_pp_apply_clock_for_voltage_request(dccg->ctx, &clock_voltage_req); - if (pp_smu->set_display_requirement) - pp_smu->set_display_requirement(&pp_smu->pp_smu, &smu_req); - } - - /* dcn1 dppclk is tied to dispclk */ - /* program dispclk on = as a w/a for sleep resume clock ramping issues */ - if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, dccg->clks.dispclk_khz) - || new_clocks->dispclk_khz == dccg->clks.dispclk_khz) { - dcn1_ramp_up_dispclk_with_dpp(dccg, new_clocks); - dccg->clks.dispclk_khz = new_clocks->dispclk_khz; - - send_request_to_lower = true; - } - - if (!send_request_to_increase && send_request_to_lower) { - /*use dcfclk to request voltage*/ - clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DCFCLK; - clock_voltage_req.clocks_in_khz = dcn_find_dcfclk_suits_all(dc, new_clocks); - dm_pp_apply_clock_for_voltage_request(dccg->ctx, &clock_voltage_req); - if (pp_smu->set_display_requirement) - pp_smu->set_display_requirement(&pp_smu->pp_smu, &smu_req); - } - - - *smu_req_cur = smu_req; -} -#endif - -static void dce_update_clocks(struct dccg *dccg, - struct dc_clocks *new_clocks, - bool safe_to_lower) -{ - struct dm_pp_power_level_change_request level_change_req; - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(dccg); - - /* TODO: Investigate why this is needed to fix display corruption. */ - if (!clk_dce->dfs_bypass_active) - new_clocks->dispclk_khz = new_clocks->dispclk_khz * 115 / 100; - - level_change_req.power_level = dce_get_required_clocks_state(dccg, new_clocks); - /* get max clock state from PPLIB */ - if ((level_change_req.power_level < dccg->cur_min_clks_state && safe_to_lower) - || level_change_req.power_level > dccg->cur_min_clks_state) { - if (dm_pp_apply_power_level_change_request(dccg->ctx, &level_change_req)) - dccg->cur_min_clks_state = level_change_req.power_level; - } - - if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, dccg->clks.dispclk_khz)) { - new_clocks->dispclk_khz = dccg->funcs->set_dispclk(dccg, new_clocks->dispclk_khz); - dccg->clks.dispclk_khz = new_clocks->dispclk_khz; - } -} - -static bool dce_update_dfs_bypass( - struct dccg *dccg, - struct dc *dc, - struct dc_state *context, - int requested_clock_khz) -{ - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(dccg); - struct resource_context *res_ctx = &context->res_ctx; - enum signal_type signal_type = SIGNAL_TYPE_NONE; - bool was_active = clk_dce->dfs_bypass_active; - int i; - - /* Disable DFS bypass by default. */ - clk_dce->dfs_bypass_active = false; - - /* Check that DFS bypass is available. */ - if (!clk_dce->dfs_bypass_enabled) - goto update; - - /* Check if the requested display clock is below the threshold. */ - if (requested_clock_khz >= 400000) - goto update; - - /* DFS-bypass should only be enabled on single stream setups */ - if (context->stream_count != 1) - goto update; - - /* Check that the stream's signal type is an embedded panel */ - for (i = 0; i < dc->res_pool->pipe_count; i++) { - if (res_ctx->pipe_ctx[i].stream) { - struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; - - signal_type = pipe_ctx->stream->sink->link->connector_signal; - break; - } - } - - if (signal_type == SIGNAL_TYPE_EDP || - signal_type == SIGNAL_TYPE_LVDS) - clk_dce->dfs_bypass_active = true; - -update: - /* Update the clock state. We don't need to respect safe_to_lower - * because DFS bypass should always be greater than the current - * display clock frequency. - */ - if (was_active != clk_dce->dfs_bypass_active) { - dccg->clks.dispclk_khz = - dccg->funcs->set_dispclk(dccg, dccg->clks.dispclk_khz); - return true; - } - - return false; -} - -#ifdef CONFIG_DRM_AMD_DC_DCN1_0 -static const struct display_clock_funcs dcn1_funcs = { - .get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz, - .set_dispclk = dce112_set_clock, - .update_clocks = dcn1_update_clocks -}; -#endif - -static const struct display_clock_funcs dce120_funcs = { - .get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz, - .set_dispclk = dce112_set_clock, - .update_clocks = dce12_update_clocks -}; - -static const struct display_clock_funcs dce112_funcs = { - .get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz, - .set_dispclk = dce112_set_clock, - .update_clocks = dce_update_clocks -}; - -static const struct display_clock_funcs dce110_funcs = { - .get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz, - .set_dispclk = dce_psr_set_clock, - .update_clocks = dce_update_clocks, - .update_dfs_bypass = dce_update_dfs_bypass -}; - -static const struct display_clock_funcs dce_funcs = { - .get_dp_ref_clk_frequency = dce_get_dp_ref_freq_khz, - .set_dispclk = dce_set_clock, - .update_clocks = dce_update_clocks -}; - -static void dce_dccg_construct( - struct dce_dccg *clk_dce, - struct dc_context *ctx, - const struct dccg_registers *regs, - const struct dccg_shift *clk_shift, - const struct dccg_mask *clk_mask) -{ - struct dccg *base = &clk_dce->base; - - base->ctx = ctx; - base->funcs = &dce_funcs; - - clk_dce->regs = regs; - clk_dce->clk_shift = clk_shift; - clk_dce->clk_mask = clk_mask; - - clk_dce->dfs_bypass_disp_clk = 0; - - clk_dce->dprefclk_ss_percentage = 0; - clk_dce->dprefclk_ss_divider = 1000; - clk_dce->ss_on_dprefclk = false; - - base->max_clks_state = DM_PP_CLOCKS_STATE_NOMINAL; - base->cur_min_clks_state = DM_PP_CLOCKS_STATE_INVALID; - - dce_clock_read_integrated_info(clk_dce); - dce_clock_read_ss_info(clk_dce); -} - -struct dccg *dce_dccg_create( - struct dc_context *ctx, - const struct dccg_registers *regs, - const struct dccg_shift *clk_shift, - const struct dccg_mask *clk_mask) -{ - struct dce_dccg *clk_dce = kzalloc(sizeof(*clk_dce), GFP_KERNEL); - - if (clk_dce == NULL) { - BREAK_TO_DEBUGGER(); - return NULL; - } - - memcpy(clk_dce->max_clks_by_state, - dce80_max_clks_by_state, - sizeof(dce80_max_clks_by_state)); - - dce_dccg_construct( - clk_dce, ctx, regs, clk_shift, clk_mask); - - return &clk_dce->base; -} - -struct dccg *dce110_dccg_create( - struct dc_context *ctx, - const struct dccg_registers *regs, - const struct dccg_shift *clk_shift, - const struct dccg_mask *clk_mask) -{ - struct dce_dccg *clk_dce = kzalloc(sizeof(*clk_dce), GFP_KERNEL); - - if (clk_dce == NULL) { - BREAK_TO_DEBUGGER(); - return NULL; - } - - memcpy(clk_dce->max_clks_by_state, - dce110_max_clks_by_state, - sizeof(dce110_max_clks_by_state)); - - dce_dccg_construct( - clk_dce, ctx, regs, clk_shift, clk_mask); - - clk_dce->base.funcs = &dce110_funcs; - - return &clk_dce->base; -} - -struct dccg *dce112_dccg_create( - struct dc_context *ctx, - const struct dccg_registers *regs, - const struct dccg_shift *clk_shift, - const struct dccg_mask *clk_mask) -{ - struct dce_dccg *clk_dce = kzalloc(sizeof(*clk_dce), GFP_KERNEL); - - if (clk_dce == NULL) { - BREAK_TO_DEBUGGER(); - return NULL; - } - - memcpy(clk_dce->max_clks_by_state, - dce112_max_clks_by_state, - sizeof(dce112_max_clks_by_state)); - - dce_dccg_construct( - clk_dce, ctx, regs, clk_shift, clk_mask); - - clk_dce->base.funcs = &dce112_funcs; - - return &clk_dce->base; -} - -struct dccg *dce120_dccg_create(struct dc_context *ctx) -{ - struct dce_dccg *clk_dce = kzalloc(sizeof(*clk_dce), GFP_KERNEL); - - if (clk_dce == NULL) { - BREAK_TO_DEBUGGER(); - return NULL; - } - - memcpy(clk_dce->max_clks_by_state, - dce120_max_clks_by_state, - sizeof(dce120_max_clks_by_state)); - - dce_dccg_construct( - clk_dce, ctx, NULL, NULL, NULL); - - clk_dce->dprefclk_khz = 600000; - clk_dce->base.funcs = &dce120_funcs; - - return &clk_dce->base; -} - -#ifdef CONFIG_DRM_AMD_DC_DCN1_0 -struct dccg *dcn1_dccg_create(struct dc_context *ctx) -{ - struct dc_debug_options *debug = &ctx->dc->debug; - struct dc_bios *bp = ctx->dc_bios; - struct dc_firmware_info fw_info = { { 0 } }; - struct dce_dccg *clk_dce = kzalloc(sizeof(*clk_dce), GFP_KERNEL); - - if (clk_dce == NULL) { - BREAK_TO_DEBUGGER(); - return NULL; - } - - clk_dce->base.ctx = ctx; - clk_dce->base.funcs = &dcn1_funcs; - - clk_dce->dfs_bypass_disp_clk = 0; - - clk_dce->dprefclk_ss_percentage = 0; - clk_dce->dprefclk_ss_divider = 1000; - clk_dce->ss_on_dprefclk = false; - - clk_dce->dprefclk_khz = 600000; - if (bp->integrated_info) - clk_dce->dentist_vco_freq_khz = bp->integrated_info->dentist_vco_freq; - if (clk_dce->dentist_vco_freq_khz == 0) { - bp->funcs->get_firmware_info(bp, &fw_info); - clk_dce->dentist_vco_freq_khz = fw_info.smu_gpu_pll_output_freq; - if (clk_dce->dentist_vco_freq_khz == 0) - clk_dce->dentist_vco_freq_khz = 3600000; - } - - if (!debug->disable_dfs_bypass && bp->integrated_info) - if (bp->integrated_info->gpu_cap_info & DFS_BYPASS_ENABLE) - clk_dce->dfs_bypass_enabled = true; - - dce_clock_read_ss_info(clk_dce); - - return &clk_dce->base; -} -#endif - -void dce_dccg_destroy(struct dccg **dccg) -{ - struct dce_dccg *clk_dce = TO_DCE_CLOCKS(*dccg); - - kfree(clk_dce); - *dccg = NULL; -} diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_hwseq.h b/drivers/gpu/drm/amd/display/dc/dce/dce_hwseq.h index 64dc75378541..c83a7f05f14c 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_hwseq.h @@ -233,6 +233,16 @@ struct dce_hwseq_registers { uint32_t DOMAIN5_PG_CONFIG; uint32_t DOMAIN6_PG_CONFIG; uint32_t DOMAIN7_PG_CONFIG; + uint32_t DOMAIN8_PG_CONFIG; + uint32_t DOMAIN9_PG_CONFIG; + uint32_t DOMAIN10_PG_CONFIG; + uint32_t DOMAIN11_PG_CONFIG; + uint32_t DOMAIN16_PG_CONFIG; + uint32_t DOMAIN17_PG_CONFIG; + uint32_t DOMAIN18_PG_CONFIG; + uint32_t DOMAIN19_PG_CONFIG; + uint32_t DOMAIN20_PG_CONFIG; + uint32_t DOMAIN21_PG_CONFIG; uint32_t DOMAIN0_PG_STATUS; uint32_t DOMAIN1_PG_STATUS; uint32_t DOMAIN2_PG_STATUS; @@ -241,6 +251,16 @@ struct dce_hwseq_registers { uint32_t DOMAIN5_PG_STATUS; uint32_t DOMAIN6_PG_STATUS; uint32_t DOMAIN7_PG_STATUS; + uint32_t DOMAIN8_PG_STATUS; + uint32_t DOMAIN9_PG_STATUS; + uint32_t DOMAIN10_PG_STATUS; + uint32_t DOMAIN11_PG_STATUS; + uint32_t DOMAIN16_PG_STATUS; + uint32_t DOMAIN17_PG_STATUS; + uint32_t DOMAIN18_PG_STATUS; + uint32_t DOMAIN19_PG_STATUS; + uint32_t DOMAIN20_PG_STATUS; + uint32_t DOMAIN21_PG_STATUS; uint32_t DIO_MEM_PWR_CTRL; uint32_t DCCG_GATE_DISABLE_CNTL; uint32_t DCCG_GATE_DISABLE_CNTL2; @@ -262,6 +282,8 @@ struct dce_hwseq_registers { uint32_t D2VGA_CONTROL; uint32_t D3VGA_CONTROL; uint32_t D4VGA_CONTROL; + uint32_t D5VGA_CONTROL; + uint32_t D6VGA_CONTROL; uint32_t VGA_TEST_CONTROL; /* MMHUB registers. read only. temporary hack */ uint32_t VM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32; @@ -489,6 +511,26 @@ struct dce_hwseq_registers { type DOMAIN6_POWER_GATE; \ type DOMAIN7_POWER_FORCEON; \ type DOMAIN7_POWER_GATE; \ + type DOMAIN8_POWER_FORCEON; \ + type DOMAIN8_POWER_GATE; \ + type DOMAIN9_POWER_FORCEON; \ + type DOMAIN9_POWER_GATE; \ + type DOMAIN10_POWER_FORCEON; \ + type DOMAIN10_POWER_GATE; \ + type DOMAIN11_POWER_FORCEON; \ + type DOMAIN11_POWER_GATE; \ + type DOMAIN16_POWER_FORCEON; \ + type DOMAIN16_POWER_GATE; \ + type DOMAIN17_POWER_FORCEON; \ + type DOMAIN17_POWER_GATE; \ + type DOMAIN18_POWER_FORCEON; \ + type DOMAIN18_POWER_GATE; \ + type DOMAIN19_POWER_FORCEON; \ + type DOMAIN19_POWER_GATE; \ + type DOMAIN20_POWER_FORCEON; \ + type DOMAIN20_POWER_GATE; \ + type DOMAIN21_POWER_FORCEON; \ + type DOMAIN21_POWER_GATE; \ type DOMAIN0_PGFSM_PWR_STATUS; \ type DOMAIN1_PGFSM_PWR_STATUS; \ type DOMAIN2_PGFSM_PWR_STATUS; \ @@ -497,6 +539,16 @@ struct dce_hwseq_registers { type DOMAIN5_PGFSM_PWR_STATUS; \ type DOMAIN6_PGFSM_PWR_STATUS; \ type DOMAIN7_PGFSM_PWR_STATUS; \ + type DOMAIN8_PGFSM_PWR_STATUS; \ + type DOMAIN9_PGFSM_PWR_STATUS; \ + type DOMAIN10_PGFSM_PWR_STATUS; \ + type DOMAIN11_PGFSM_PWR_STATUS; \ + type DOMAIN16_PGFSM_PWR_STATUS; \ + type DOMAIN17_PGFSM_PWR_STATUS; \ + type DOMAIN18_PGFSM_PWR_STATUS; \ + type DOMAIN19_PGFSM_PWR_STATUS; \ + type DOMAIN20_PGFSM_PWR_STATUS; \ + type DOMAIN21_PGFSM_PWR_STATUS; \ type DCFCLK_GATE_DIS; \ type DCHUBBUB_GLOBAL_TIMER_REFDIV; \ type VGA_TEST_ENABLE; \ diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c index 366bc8c2c643..3e18ea84b1f9 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c @@ -645,7 +645,7 @@ static bool dce110_link_encoder_validate_hdmi_output( return false; /* DCE11 HW does not support 420 */ - if (!enc110->base.features.ycbcr420_supported && + if (!enc110->base.features.hdmi_ycbcr420_supported && crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420) return false; diff --git a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.c index 74c05e878807..bc50a8e25f4f 100644 --- a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.c @@ -105,74 +105,18 @@ bool dce100_enable_display_power_gating( return false; } -static void dce100_pplib_apply_display_requirements( - struct dc *dc, - struct dc_state *context) -{ - struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg; - - pp_display_cfg->avail_mclk_switch_time_us = - dce110_get_min_vblank_time_us(context); - /*pp_display_cfg->min_memory_clock_khz = context->bw.dce.yclk_khz - / MEMORY_TYPE_MULTIPLIER;*/ - - dce110_fill_display_configs(context, pp_display_cfg); - - if (memcmp(&dc->prev_display_config, pp_display_cfg, sizeof( - struct dm_pp_display_configuration)) != 0) - dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg); - - dc->prev_display_config = *pp_display_cfg; -} - -/* unit: in_khz before mode set, get pixel clock from context. ASIC register - * may not be programmed yet - */ -static uint32_t get_max_pixel_clock_for_all_paths( - struct dc *dc, - struct dc_state *context) -{ - uint32_t max_pix_clk = 0; - int i; - - for (i = 0; i < MAX_PIPES; i++) { - struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; - - if (pipe_ctx->stream == NULL) - continue; - - /* do not check under lay */ - if (pipe_ctx->top_pipe) - continue; - - if (pipe_ctx->stream_res.pix_clk_params.requested_pix_clk > max_pix_clk) - max_pix_clk = - pipe_ctx->stream_res.pix_clk_params.requested_pix_clk; - } - return max_pix_clk; -} - -void dce100_set_bandwidth( +void dce100_prepare_bandwidth( struct dc *dc, - struct dc_state *context, - bool decrease_allowed) + struct dc_state *context) { - struct dc_clocks req_clks; - - req_clks.dispclk_khz = context->bw.dce.dispclk_khz * 115 / 100; - req_clks.phyclk_khz = get_max_pixel_clock_for_all_paths(dc, context); - dce110_set_safe_displaymarks(&context->res_ctx, dc->res_pool); - dc->res_pool->dccg->funcs->update_clocks( - dc->res_pool->dccg, - &req_clks, - decrease_allowed); - - dce100_pplib_apply_display_requirements(dc, context); + dc->res_pool->clk_mgr->funcs->update_clocks( + dc->res_pool->clk_mgr, + context, + false); } - /**************************************************************************/ void dce100_hw_sequencer_construct(struct dc *dc) @@ -180,8 +124,7 @@ void dce100_hw_sequencer_construct(struct dc *dc) dce110_hw_sequencer_construct(dc); dc->hwss.enable_display_power_gating = dce100_enable_display_power_gating; - dc->hwss.set_bandwidth = dce100_set_bandwidth; - dc->hwss.pplib_apply_display_requirements = - dce100_pplib_apply_display_requirements; + dc->hwss.prepare_bandwidth = dce100_prepare_bandwidth; + dc->hwss.optimize_bandwidth = dce100_prepare_bandwidth; } diff --git a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h index c6ec0ed6ec3d..acd418515346 100644 --- a/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/dce100/dce100_hw_sequencer.h @@ -33,10 +33,9 @@ struct dc_state; void dce100_hw_sequencer_construct(struct dc *dc); -void dce100_set_bandwidth( +void dce100_prepare_bandwidth( struct dc *dc, - struct dc_state *context, - bool decrease_allowed); + struct dc_state *context); bool dce100_enable_display_power_gating(struct dc *dc, uint8_t controller_id, struct dc_bios *dcb, diff --git a/drivers/gpu/drm/amd/display/dc/dce100/dce100_resource.c b/drivers/gpu/drm/amd/display/dc/dce100/dce100_resource.c index 14754a87156c..6ae51a5dfc04 100644 --- a/drivers/gpu/drm/amd/display/dc/dce100/dce100_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce100/dce100_resource.c @@ -36,11 +36,11 @@ #include "dce/dce_link_encoder.h" #include "dce/dce_stream_encoder.h" +#include "dce/dce_clk_mgr.h" #include "dce/dce_mem_input.h" #include "dce/dce_ipp.h" #include "dce/dce_transform.h" #include "dce/dce_opp.h" -#include "dce/dce_clocks.h" #include "dce/dce_clock_source.h" #include "dce/dce_audio.h" #include "dce/dce_hwseq.h" @@ -137,15 +137,15 @@ static const struct dce110_timing_generator_offsets dce100_tg_offsets[] = { .reg_name = mm ## block ## id ## _ ## reg_name -static const struct dccg_registers disp_clk_regs = { +static const struct clk_mgr_registers disp_clk_regs = { CLK_COMMON_REG_LIST_DCE_BASE() }; -static const struct dccg_shift disp_clk_shift = { +static const struct clk_mgr_shift disp_clk_shift = { CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(__SHIFT) }; -static const struct dccg_mask disp_clk_mask = { +static const struct clk_mgr_mask disp_clk_mask = { CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(_MASK) }; @@ -722,8 +722,8 @@ static void destruct(struct dce110_resource_pool *pool) dce_aud_destroy(&pool->base.audios[i]); } - if (pool->base.dccg != NULL) - dce_dccg_destroy(&pool->base.dccg); + if (pool->base.clk_mgr != NULL) + dce_clk_mgr_destroy(&pool->base.clk_mgr); if (pool->base.abm != NULL) dce_abm_destroy(&pool->base.abm); @@ -767,7 +767,7 @@ bool dce100_validate_bandwidth( if (at_least_one_pipe) { /* TODO implement when needed but for now hardcode max value*/ context->bw.dce.dispclk_khz = 681000; - context->bw.dce.yclk_khz = 250000 * MEMORY_TYPE_MULTIPLIER; + context->bw.dce.yclk_khz = 250000 * MEMORY_TYPE_MULTIPLIER_CZ; } else { context->bw.dce.dispclk_khz = 0; context->bw.dce.yclk_khz = 0; @@ -860,7 +860,6 @@ static bool construct( struct dc_context *ctx = dc->ctx; struct dc_firmware_info info; struct dc_bios *bp; - struct dm_pp_static_clock_info static_clk_info = {0}; ctx->dc_bios->regs = &bios_regs; @@ -908,11 +907,11 @@ static bool construct( } } - pool->base.dccg = dce_dccg_create(ctx, + pool->base.clk_mgr = dce_clk_mgr_create(ctx, &disp_clk_regs, &disp_clk_shift, &disp_clk_mask); - if (pool->base.dccg == NULL) { + if (pool->base.clk_mgr == NULL) { dm_error("DC: failed to create display clock!\n"); BREAK_TO_DEBUGGER(); goto res_create_fail; @@ -938,12 +937,6 @@ static bool construct( goto res_create_fail; } - /* get static clock information for PPLIB or firmware, save - * max_clock_state - */ - if (dm_pp_get_static_clocks(ctx, &static_clk_info)) - pool->base.dccg->max_clks_state = - static_clk_info.max_clocks_state; { struct irq_service_init_data init_data; init_data.ctx = dc->ctx; diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c index b459867a05b2..9724a17e352b 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c @@ -548,14 +548,14 @@ dce110_translate_regamma_to_hw_format(const struct dc_transfer_func *output_tf, regamma_params->hw_points_num = hw_points; - i = 1; - for (k = 0; k < 16 && i < 16; k++) { + k = 0; + for (i = 1; i < 16; i++) { if (seg_distr[k] != -1) { regamma_params->arr_curve_points[k].segments_num = seg_distr[k]; regamma_params->arr_curve_points[i].offset = regamma_params->arr_curve_points[k].offset + (1 << seg_distr[k]); } - i++; + k++; } if (seg_distr[k] != -1) @@ -1085,7 +1085,6 @@ void dce110_unblank_stream(struct pipe_ctx *pipe_ctx, if (link->local_sink && link->local_sink->sink_signal == SIGNAL_TYPE_EDP) { link->dc->hwss.edp_backlight_control(link, true); - stream->bl_pwm_level = EDP_BACKLIGHT_RAMP_DISABLE_LEVEL; } } void dce110_blank_stream(struct pipe_ctx *pipe_ctx) @@ -1192,8 +1191,8 @@ static void build_audio_output( if (pipe_ctx->stream->signal == SIGNAL_TYPE_DISPLAY_PORT || pipe_ctx->stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST) { audio_output->pll_info.dp_dto_source_clock_in_khz = - state->dis_clk->funcs->get_dp_ref_clk_frequency( - state->dis_clk); + state->dccg->funcs->get_dp_ref_clk_frequency( + state->dccg); } audio_output->pll_info.feed_back_divider = @@ -1547,6 +1546,7 @@ void dce110_enable_accelerated_mode(struct dc *dc, struct dc_state *context) int i; struct dc_link *edp_link_to_turnoff = NULL; struct dc_link *edp_link = get_link_for_edp(dc); + struct dc_bios *bios = dc->ctx->dc_bios; bool can_edp_fast_boot_optimize = false; bool apply_edp_fast_boot_optimization = false; @@ -1573,6 +1573,20 @@ void dce110_enable_accelerated_mode(struct dc *dc, struct dc_state *context) if (context->streams[i]->signal == SIGNAL_TYPE_EDP) { context->streams[i]->apply_edp_fast_boot_optimization = true; apply_edp_fast_boot_optimization = true; + + /* When after S4 and S5, vbios may post edp and previous dpms_off + * doesn't make sense. + * Update dpms_off state to align hw and sw state via check + * vBios scratch register. + */ + if (bios->funcs->is_active_display) { + const struct connector_device_tag_info *device_tag = &(edp_link->device_tag); + + if (bios->funcs->is_active_display(bios, + context->streams[i]->signal, + device_tag)) + context->streams[i]->dpms_off = false; + } } } } @@ -1748,34 +1762,6 @@ static void set_static_screen_control(struct pipe_ctx **pipe_ctx, set_static_screen_control(pipe_ctx[i]->stream_res.tg, value); } -/* unit: in_khz before mode set, get pixel clock from context. ASIC register - * may not be programmed yet - */ -static uint32_t get_max_pixel_clock_for_all_paths( - struct dc *dc, - struct dc_state *context) -{ - uint32_t max_pix_clk = 0; - int i; - - for (i = 0; i < MAX_PIPES; i++) { - struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; - - if (pipe_ctx->stream == NULL) - continue; - - /* do not check under lay */ - if (pipe_ctx->top_pipe) - continue; - - if (pipe_ctx->stream_res.pix_clk_params.requested_pix_clk > max_pix_clk) - max_pix_clk = - pipe_ctx->stream_res.pix_clk_params.requested_pix_clk; - } - - return max_pix_clk; -} - /* * Check if FBC can be enabled */ @@ -2385,191 +2371,33 @@ static void init_hw(struct dc *dc) } -void dce110_fill_display_configs( - const struct dc_state *context, - struct dm_pp_display_configuration *pp_display_cfg) -{ - int j; - int num_cfgs = 0; - - for (j = 0; j < context->stream_count; j++) { - int k; - - const struct dc_stream_state *stream = context->streams[j]; - struct dm_pp_single_disp_config *cfg = - &pp_display_cfg->disp_configs[num_cfgs]; - const struct pipe_ctx *pipe_ctx = NULL; - for (k = 0; k < MAX_PIPES; k++) - if (stream == context->res_ctx.pipe_ctx[k].stream) { - pipe_ctx = &context->res_ctx.pipe_ctx[k]; - break; - } - - ASSERT(pipe_ctx != NULL); - - /* only notify active stream */ - if (stream->dpms_off) - continue; - - num_cfgs++; - cfg->signal = pipe_ctx->stream->signal; - cfg->pipe_idx = pipe_ctx->stream_res.tg->inst; - cfg->src_height = stream->src.height; - cfg->src_width = stream->src.width; - cfg->ddi_channel_mapping = - stream->sink->link->ddi_channel_mapping.raw; - cfg->transmitter = - stream->sink->link->link_enc->transmitter; - cfg->link_settings.lane_count = - stream->sink->link->cur_link_settings.lane_count; - cfg->link_settings.link_rate = - stream->sink->link->cur_link_settings.link_rate; - cfg->link_settings.link_spread = - stream->sink->link->cur_link_settings.link_spread; - cfg->sym_clock = stream->phy_pix_clk; - /* Round v_refresh*/ - cfg->v_refresh = stream->timing.pix_clk_khz * 1000; - cfg->v_refresh /= stream->timing.h_total; - cfg->v_refresh = (cfg->v_refresh + stream->timing.v_total / 2) - / stream->timing.v_total; - } - - pp_display_cfg->display_count = num_cfgs; -} - -uint32_t dce110_get_min_vblank_time_us(const struct dc_state *context) -{ - uint8_t j; - uint32_t min_vertical_blank_time = -1; - - for (j = 0; j < context->stream_count; j++) { - struct dc_stream_state *stream = context->streams[j]; - uint32_t vertical_blank_in_pixels = 0; - uint32_t vertical_blank_time = 0; - - vertical_blank_in_pixels = stream->timing.h_total * - (stream->timing.v_total - - stream->timing.v_addressable); - - vertical_blank_time = vertical_blank_in_pixels - * 1000 / stream->timing.pix_clk_khz; - - if (min_vertical_blank_time > vertical_blank_time) - min_vertical_blank_time = vertical_blank_time; - } - - return min_vertical_blank_time; -} - -static int determine_sclk_from_bounding_box( - const struct dc *dc, - int required_sclk) -{ - int i; - - /* - * Some asics do not give us sclk levels, so we just report the actual - * required sclk - */ - if (dc->sclk_lvls.num_levels == 0) - return required_sclk; - - for (i = 0; i < dc->sclk_lvls.num_levels; i++) { - if (dc->sclk_lvls.clocks_in_khz[i] >= required_sclk) - return dc->sclk_lvls.clocks_in_khz[i]; - } - /* - * even maximum level could not satisfy requirement, this - * is unexpected at this stage, should have been caught at - * validation time - */ - ASSERT(0); - return dc->sclk_lvls.clocks_in_khz[dc->sclk_lvls.num_levels - 1]; -} - -static void pplib_apply_display_requirements( - struct dc *dc, - struct dc_state *context) +void dce110_prepare_bandwidth( + struct dc *dc, + struct dc_state *context) { - struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg; - - pp_display_cfg->all_displays_in_sync = - context->bw.dce.all_displays_in_sync; - pp_display_cfg->nb_pstate_switch_disable = - context->bw.dce.nbp_state_change_enable == false; - pp_display_cfg->cpu_cc6_disable = - context->bw.dce.cpuc_state_change_enable == false; - pp_display_cfg->cpu_pstate_disable = - context->bw.dce.cpup_state_change_enable == false; - pp_display_cfg->cpu_pstate_separation_time = - context->bw.dce.blackout_recovery_time_us; + struct clk_mgr *dccg = dc->res_pool->clk_mgr; - pp_display_cfg->min_memory_clock_khz = context->bw.dce.yclk_khz - / MEMORY_TYPE_MULTIPLIER; + dce110_set_safe_displaymarks(&context->res_ctx, dc->res_pool); - pp_display_cfg->min_engine_clock_khz = determine_sclk_from_bounding_box( - dc, - context->bw.dce.sclk_khz); - - pp_display_cfg->min_engine_clock_deep_sleep_khz - = context->bw.dce.sclk_deep_sleep_khz; - - pp_display_cfg->avail_mclk_switch_time_us = - dce110_get_min_vblank_time_us(context); - /* TODO: dce11.2*/ - pp_display_cfg->avail_mclk_switch_time_in_disp_active_us = 0; - - pp_display_cfg->disp_clk_khz = dc->res_pool->dccg->clks.dispclk_khz; - - dce110_fill_display_configs(context, pp_display_cfg); - - /* TODO: is this still applicable?*/ - if (pp_display_cfg->display_count == 1) { - const struct dc_crtc_timing *timing = - &context->streams[0]->timing; - - pp_display_cfg->crtc_index = - pp_display_cfg->disp_configs[0].pipe_idx; - pp_display_cfg->line_time_in_us = timing->h_total * 1000 - / timing->pix_clk_khz; - } - - if (memcmp(&dc->prev_display_config, pp_display_cfg, sizeof( - struct dm_pp_display_configuration)) != 0) - dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg); - - dc->prev_display_config = *pp_display_cfg; + dccg->funcs->update_clocks( + dccg, + context, + false); } -static void dce110_set_bandwidth( +void dce110_optimize_bandwidth( struct dc *dc, - struct dc_state *context, - bool decrease_allowed) + struct dc_state *context) { - struct dc_clocks req_clks; - struct dccg *dccg = dc->res_pool->dccg; - - req_clks.dispclk_khz = context->bw.dce.dispclk_khz; - req_clks.phyclk_khz = get_max_pixel_clock_for_all_paths(dc, context); - - if (decrease_allowed) - dce110_set_displaymarks(dc, context); - else - dce110_set_safe_displaymarks(&context->res_ctx, dc->res_pool); + struct clk_mgr *dccg = dc->res_pool->clk_mgr; - if (dccg->funcs->update_dfs_bypass) - dccg->funcs->update_dfs_bypass( - dccg, - dc, - context, - req_clks.dispclk_khz); + dce110_set_displaymarks(dc, context); dccg->funcs->update_clocks( dccg, - &req_clks, - decrease_allowed); - pplib_apply_display_requirements(dc, context); + context, + true); } static void dce110_program_front_end_for_pipe( @@ -2774,28 +2602,6 @@ static void dce110_wait_for_mpcc_disconnect( /* do nothing*/ } -static void program_csc_matrix(struct pipe_ctx *pipe_ctx, - enum dc_color_space colorspace, - uint16_t *matrix) -{ - int i; - struct out_csc_color_matrix tbl_entry; - - if (pipe_ctx->stream->csc_color_matrix.enable_adjustment - == true) { - enum dc_color_space color_space = - pipe_ctx->stream->output_color_space; - - //uint16_t matrix[12]; - for (i = 0; i < 12; i++) - tbl_entry.regval[i] = pipe_ctx->stream->csc_color_matrix.matrix[i]; - - tbl_entry.color_space = color_space; - //tbl_entry.regval = matrix; - pipe_ctx->plane_res.xfm->funcs->opp_set_csc_adjustment(pipe_ctx->plane_res.xfm, &tbl_entry); - } -} - void dce110_set_cursor_position(struct pipe_ctx *pipe_ctx) { struct dc_cursor_position pos_cpy = pipe_ctx->stream->cursor_position; @@ -2844,13 +2650,8 @@ void dce110_set_cursor_attribute(struct pipe_ctx *pipe_ctx) pipe_ctx->plane_res.xfm, attributes); } -static void ready_shared_resources(struct dc *dc, struct dc_state *context) {} - -static void optimize_shared_resources(struct dc *dc) {} - static const struct hw_sequencer_funcs dce110_funcs = { .program_gamut_remap = program_gamut_remap, - .program_csc_matrix = program_csc_matrix, .init_hw = init_hw, .apply_ctx_to_hw = dce110_apply_ctx_to_hw, .apply_ctx_for_surface = dce110_apply_ctx_for_surface, @@ -2873,7 +2674,8 @@ static const struct hw_sequencer_funcs dce110_funcs = { .enable_display_power_gating = dce110_enable_display_power_gating, .disable_plane = dce110_power_down_fe, .pipe_control_lock = dce_pipe_control_lock, - .set_bandwidth = dce110_set_bandwidth, + .prepare_bandwidth = dce110_prepare_bandwidth, + .optimize_bandwidth = dce110_optimize_bandwidth, .set_drr = set_drr, .get_position = get_position, .set_static_screen_control = set_static_screen_control, @@ -2882,9 +2684,6 @@ static const struct hw_sequencer_funcs dce110_funcs = { .setup_stereo = NULL, .set_avmute = dce110_set_avmute, .wait_for_mpcc_disconnect = dce110_wait_for_mpcc_disconnect, - .ready_shared_resources = ready_shared_resources, - .optimize_shared_resources = optimize_shared_resources, - .pplib_apply_display_requirements = pplib_apply_display_requirements, .edp_backlight_control = hwss_edp_backlight_control, .edp_power_control = hwss_edp_power_control, .edp_wait_for_hpd_ready = hwss_edp_wait_for_hpd_ready, diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.h index d6db3dbd9015..cd3e36d52a52 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.h @@ -40,7 +40,6 @@ enum dc_status dce110_apply_ctx_to_hw( struct dc_state *context); - void dce110_enable_stream(struct pipe_ctx *pipe_ctx); void dce110_disable_stream(struct pipe_ctx *pipe_ctx, int option); @@ -64,11 +63,13 @@ void dce110_set_safe_displaymarks( struct resource_context *res_ctx, const struct resource_pool *pool); -void dce110_fill_display_configs( - const struct dc_state *context, - struct dm_pp_display_configuration *pp_display_cfg); +void dce110_prepare_bandwidth( + struct dc *dc, + struct dc_state *context); -uint32_t dce110_get_min_vblank_time_us(const struct dc_state *context); +void dce110_optimize_bandwidth( + struct dc *dc, + struct dc_state *context); void dp_receiver_power_ctrl(struct dc_link *link, bool on); diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_resource.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_resource.c index 7c9fd9052ee2..e33d11785b1f 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_resource.c @@ -31,6 +31,7 @@ #include "resource.h" #include "dce110/dce110_resource.h" +#include "dce/dce_clk_mgr.h" #include "include/irq_service_interface.h" #include "dce/dce_audio.h" #include "dce110/dce110_timing_generator.h" @@ -45,7 +46,6 @@ #include "dce110/dce110_transform_v.h" #include "dce/dce_opp.h" #include "dce110/dce110_opp_v.h" -#include "dce/dce_clocks.h" #include "dce/dce_clock_source.h" #include "dce/dce_hwseq.h" #include "dce110/dce110_hw_sequencer.h" @@ -148,15 +148,15 @@ static const struct dce110_timing_generator_offsets dce110_tg_offsets[] = { #define SRI(reg_name, block, id)\ .reg_name = mm ## block ## id ## _ ## reg_name -static const struct dccg_registers disp_clk_regs = { +static const struct clk_mgr_registers disp_clk_regs = { CLK_COMMON_REG_LIST_DCE_BASE() }; -static const struct dccg_shift disp_clk_shift = { +static const struct clk_mgr_shift disp_clk_shift = { CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(__SHIFT) }; -static const struct dccg_mask disp_clk_mask = { +static const struct clk_mgr_mask disp_clk_mask = { CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(_MASK) }; @@ -760,8 +760,8 @@ static void destruct(struct dce110_resource_pool *pool) if (pool->base.dmcu != NULL) dce_dmcu_destroy(&pool->base.dmcu); - if (pool->base.dccg != NULL) - dce_dccg_destroy(&pool->base.dccg); + if (pool->base.clk_mgr != NULL) + dce_clk_mgr_destroy(&pool->base.clk_mgr); if (pool->base.irqs != NULL) { dal_irq_service_destroy(&pool->base.irqs); @@ -1173,12 +1173,12 @@ static void bw_calcs_data_update_from_pplib(struct dc *dc) &clks); dc->bw_vbios->low_yclk = bw_frc_to_fixed( - clks.clocks_in_khz[0] * MEMORY_TYPE_MULTIPLIER, 1000); + clks.clocks_in_khz[0] * MEMORY_TYPE_MULTIPLIER_CZ, 1000); dc->bw_vbios->mid_yclk = bw_frc_to_fixed( - clks.clocks_in_khz[clks.num_levels>>1] * MEMORY_TYPE_MULTIPLIER, + clks.clocks_in_khz[clks.num_levels>>1] * MEMORY_TYPE_MULTIPLIER_CZ, 1000); dc->bw_vbios->high_yclk = bw_frc_to_fixed( - clks.clocks_in_khz[clks.num_levels-1] * MEMORY_TYPE_MULTIPLIER, + clks.clocks_in_khz[clks.num_levels-1] * MEMORY_TYPE_MULTIPLIER_CZ, 1000); } @@ -1201,7 +1201,6 @@ static bool construct( struct dc_context *ctx = dc->ctx; struct dc_firmware_info info; struct dc_bios *bp; - struct dm_pp_static_clock_info static_clk_info = {0}; ctx->dc_bios->regs = &bios_regs; @@ -1257,11 +1256,11 @@ static bool construct( } } - pool->base.dccg = dce110_dccg_create(ctx, + pool->base.clk_mgr = dce110_clk_mgr_create(ctx, &disp_clk_regs, &disp_clk_shift, &disp_clk_mask); - if (pool->base.dccg == NULL) { + if (pool->base.clk_mgr == NULL) { dm_error("DC: failed to create display clock!\n"); BREAK_TO_DEBUGGER(); goto res_create_fail; @@ -1287,13 +1286,6 @@ static bool construct( goto res_create_fail; } - /* get static clock information for PPLIB or firmware, save - * max_clock_state - */ - if (dm_pp_get_static_clocks(ctx, &static_clk_info)) - pool->base.dccg->max_clks_state = - static_clk_info.max_clocks_state; - { struct irq_service_init_data init_data; init_data.ctx = dc->ctx; diff --git a/drivers/gpu/drm/amd/display/dc/dce112/dce112_resource.c b/drivers/gpu/drm/amd/display/dc/dce112/dce112_resource.c index 3ce79c208ddf..969d4e72dc94 100644 --- a/drivers/gpu/drm/amd/display/dc/dce112/dce112_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce112/dce112_resource.c @@ -35,6 +35,7 @@ #include "irq/dce110/irq_service_dce110.h" +#include "dce/dce_clk_mgr.h" #include "dce/dce_mem_input.h" #include "dce/dce_transform.h" #include "dce/dce_link_encoder.h" @@ -42,7 +43,6 @@ #include "dce/dce_audio.h" #include "dce/dce_opp.h" #include "dce/dce_ipp.h" -#include "dce/dce_clocks.h" #include "dce/dce_clock_source.h" #include "dce/dce_hwseq.h" @@ -148,15 +148,15 @@ static const struct dce110_timing_generator_offsets dce112_tg_offsets[] = { .reg_name = mm ## block ## id ## _ ## reg_name -static const struct dccg_registers disp_clk_regs = { +static const struct clk_mgr_registers disp_clk_regs = { CLK_COMMON_REG_LIST_DCE_BASE() }; -static const struct dccg_shift disp_clk_shift = { +static const struct clk_mgr_shift disp_clk_shift = { CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(__SHIFT) }; -static const struct dccg_mask disp_clk_mask = { +static const struct clk_mgr_mask disp_clk_mask = { CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(_MASK) }; @@ -551,7 +551,8 @@ static struct transform *dce112_transform_create( static const struct encoder_feature_support link_enc_feature = { .max_hdmi_deep_color = COLOR_DEPTH_121212, .max_hdmi_pixel_clock = 600000, - .ycbcr420_supported = true, + .hdmi_ycbcr420_supported = true, + .dp_ycbcr420_supported = false, .flags.bits.IS_HBR2_CAPABLE = true, .flags.bits.IS_HBR3_CAPABLE = true, .flags.bits.IS_TPS3_CAPABLE = true, @@ -749,8 +750,8 @@ static void destruct(struct dce110_resource_pool *pool) if (pool->base.dmcu != NULL) dce_dmcu_destroy(&pool->base.dmcu); - if (pool->base.dccg != NULL) - dce_dccg_destroy(&pool->base.dccg); + if (pool->base.clk_mgr != NULL) + dce_clk_mgr_destroy(&pool->base.clk_mgr); if (pool->base.irqs != NULL) { dal_irq_service_destroy(&pool->base.irqs); @@ -1015,12 +1016,12 @@ static void bw_calcs_data_update_from_pplib(struct dc *dc) &clks); dc->bw_vbios->low_yclk = bw_frc_to_fixed( - clks.clocks_in_khz[0] * MEMORY_TYPE_MULTIPLIER, 1000); + clks.clocks_in_khz[0] * MEMORY_TYPE_MULTIPLIER_CZ, 1000); dc->bw_vbios->mid_yclk = bw_frc_to_fixed( - clks.clocks_in_khz[clks.num_levels>>1] * MEMORY_TYPE_MULTIPLIER, + clks.clocks_in_khz[clks.num_levels>>1] * MEMORY_TYPE_MULTIPLIER_CZ, 1000); dc->bw_vbios->high_yclk = bw_frc_to_fixed( - clks.clocks_in_khz[clks.num_levels-1] * MEMORY_TYPE_MULTIPLIER, + clks.clocks_in_khz[clks.num_levels-1] * MEMORY_TYPE_MULTIPLIER_CZ, 1000); return; @@ -1056,12 +1057,12 @@ static void bw_calcs_data_update_from_pplib(struct dc *dc) * YCLK = UMACLK*m_memoryTypeMultiplier */ dc->bw_vbios->low_yclk = bw_frc_to_fixed( - mem_clks.data[0].clocks_in_khz * MEMORY_TYPE_MULTIPLIER, 1000); + mem_clks.data[0].clocks_in_khz * MEMORY_TYPE_MULTIPLIER_CZ, 1000); dc->bw_vbios->mid_yclk = bw_frc_to_fixed( - mem_clks.data[mem_clks.num_levels>>1].clocks_in_khz * MEMORY_TYPE_MULTIPLIER, + mem_clks.data[mem_clks.num_levels>>1].clocks_in_khz * MEMORY_TYPE_MULTIPLIER_CZ, 1000); dc->bw_vbios->high_yclk = bw_frc_to_fixed( - mem_clks.data[mem_clks.num_levels-1].clocks_in_khz * MEMORY_TYPE_MULTIPLIER, + mem_clks.data[mem_clks.num_levels-1].clocks_in_khz * MEMORY_TYPE_MULTIPLIER_CZ, 1000); /* Now notify PPLib/SMU about which Watermarks sets they should select @@ -1131,7 +1132,6 @@ static bool construct( { unsigned int i; struct dc_context *ctx = dc->ctx; - struct dm_pp_static_clock_info static_clk_info = {0}; ctx->dc_bios->regs = &bios_regs; @@ -1199,11 +1199,11 @@ static bool construct( } } - pool->base.dccg = dce112_dccg_create(ctx, + pool->base.clk_mgr = dce112_clk_mgr_create(ctx, &disp_clk_regs, &disp_clk_shift, &disp_clk_mask); - if (pool->base.dccg == NULL) { + if (pool->base.clk_mgr == NULL) { dm_error("DC: failed to create display clock!\n"); BREAK_TO_DEBUGGER(); goto res_create_fail; @@ -1229,13 +1229,6 @@ static bool construct( goto res_create_fail; } - /* get static clock information for PPLIB or firmware, save - * max_clock_state - */ - if (dm_pp_get_static_clocks(ctx, &static_clk_info)) - pool->base.dccg->max_clks_state = - static_clk_info.max_clocks_state; - { struct irq_service_init_data init_data; init_data.ctx = dc->ctx; diff --git a/drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c b/drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c index 79ab5f9f9115..f12696674eb0 100644 --- a/drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce120/dce120_resource.c @@ -31,6 +31,7 @@ #include "resource.h" #include "include/irq_service_interface.h" #include "dce120_resource.h" + #include "dce112/dce112_resource.h" #include "dce110/dce110_resource.h" @@ -39,7 +40,6 @@ #include "irq/dce120/irq_service_dce120.h" #include "dce/dce_opp.h" #include "dce/dce_clock_source.h" -#include "dce/dce_clocks.h" #include "dce/dce_ipp.h" #include "dce/dce_mem_input.h" @@ -47,6 +47,7 @@ #include "dce120/dce120_hw_sequencer.h" #include "dce/dce_transform.h" +#include "dce/dce_clk_mgr.h" #include "dce/dce_audio.h" #include "dce/dce_link_encoder.h" #include "dce/dce_stream_encoder.h" @@ -573,8 +574,8 @@ static void destruct(struct dce110_resource_pool *pool) if (pool->base.dmcu != NULL) dce_dmcu_destroy(&pool->base.dmcu); - if (pool->base.dccg != NULL) - dce_dccg_destroy(&pool->base.dccg); + if (pool->base.clk_mgr != NULL) + dce_clk_mgr_destroy(&pool->base.clk_mgr); } static void read_dce_straps( @@ -606,7 +607,8 @@ static struct audio *create_audio( static const struct encoder_feature_support link_enc_feature = { .max_hdmi_deep_color = COLOR_DEPTH_121212, .max_hdmi_pixel_clock = 600000, - .ycbcr420_supported = true, + .hdmi_ycbcr420_supported = true, + .dp_ycbcr420_supported = false, .flags.bits.IS_HBR2_CAPABLE = true, .flags.bits.IS_HBR3_CAPABLE = true, .flags.bits.IS_TPS3_CAPABLE = true, @@ -834,12 +836,12 @@ static void bw_calcs_data_update_from_pplib(struct dc *dc) * YCLK = UMACLK*m_memoryTypeMultiplier */ dc->bw_vbios->low_yclk = bw_frc_to_fixed( - mem_clks.data[0].clocks_in_khz * MEMORY_TYPE_MULTIPLIER, 1000); + mem_clks.data[0].clocks_in_khz * MEMORY_TYPE_MULTIPLIER_CZ, 1000); dc->bw_vbios->mid_yclk = bw_frc_to_fixed( - mem_clks.data[mem_clks.num_levels>>1].clocks_in_khz * MEMORY_TYPE_MULTIPLIER, + mem_clks.data[mem_clks.num_levels>>1].clocks_in_khz * MEMORY_TYPE_MULTIPLIER_CZ, 1000); dc->bw_vbios->high_yclk = bw_frc_to_fixed( - mem_clks.data[mem_clks.num_levels-1].clocks_in_khz * MEMORY_TYPE_MULTIPLIER, + mem_clks.data[mem_clks.num_levels-1].clocks_in_khz * MEMORY_TYPE_MULTIPLIER_CZ, 1000); /* Now notify PPLib/SMU about which Watermarks sets they should select @@ -973,8 +975,8 @@ static bool construct( } } - pool->base.dccg = dce120_dccg_create(ctx); - if (pool->base.dccg == NULL) { + pool->base.clk_mgr = dce120_clk_mgr_create(ctx); + if (pool->base.clk_mgr == NULL) { dm_error("DC: failed to create display clock!\n"); BREAK_TO_DEBUGGER(); goto dccg_create_fail; diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c index 6c6a1a16af19..a60a90e68d91 100644 --- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_hw_sequencer.c @@ -76,6 +76,7 @@ void dce80_hw_sequencer_construct(struct dc *dc) dc->hwss.enable_display_power_gating = dce100_enable_display_power_gating; dc->hwss.pipe_control_lock = dce_pipe_control_lock; - dc->hwss.set_bandwidth = dce100_set_bandwidth; + dc->hwss.prepare_bandwidth = dce100_prepare_bandwidth; + dc->hwss.optimize_bandwidth = dce100_prepare_bandwidth; } diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c index d68f951f9869..6d40b3d54ac1 100644 --- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c @@ -37,6 +37,7 @@ #include "dce110/dce110_timing_generator.h" #include "dce110/dce110_resource.h" #include "dce80/dce80_timing_generator.h" +#include "dce/dce_clk_mgr.h" #include "dce/dce_mem_input.h" #include "dce/dce_link_encoder.h" #include "dce/dce_stream_encoder.h" @@ -44,7 +45,6 @@ #include "dce/dce_ipp.h" #include "dce/dce_transform.h" #include "dce/dce_opp.h" -#include "dce/dce_clocks.h" #include "dce/dce_clock_source.h" #include "dce/dce_audio.h" #include "dce/dce_hwseq.h" @@ -155,15 +155,15 @@ static const struct dce110_timing_generator_offsets dce80_tg_offsets[] = { .reg_name = mm ## block ## id ## _ ## reg_name -static const struct dccg_registers disp_clk_regs = { +static const struct clk_mgr_registers disp_clk_regs = { CLK_COMMON_REG_LIST_DCE_BASE() }; -static const struct dccg_shift disp_clk_shift = { +static const struct clk_mgr_shift disp_clk_shift = { CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(__SHIFT) }; -static const struct dccg_mask disp_clk_mask = { +static const struct clk_mgr_mask disp_clk_mask = { CLK_COMMON_MASK_SH_LIST_DCE_COMMON_BASE(_MASK) }; @@ -779,8 +779,8 @@ static void destruct(struct dce110_resource_pool *pool) } } - if (pool->base.dccg != NULL) - dce_dccg_destroy(&pool->base.dccg); + if (pool->base.clk_mgr != NULL) + dce_clk_mgr_destroy(&pool->base.clk_mgr); if (pool->base.irqs != NULL) { dal_irq_service_destroy(&pool->base.irqs); @@ -793,7 +793,7 @@ bool dce80_validate_bandwidth( { /* TODO implement when needed but for now hardcode max value*/ context->bw.dce.dispclk_khz = 681000; - context->bw.dce.yclk_khz = 250000 * MEMORY_TYPE_MULTIPLIER; + context->bw.dce.yclk_khz = 250000 * MEMORY_TYPE_MULTIPLIER_CZ; return true; } @@ -855,7 +855,6 @@ static bool dce80_construct( struct dc_context *ctx = dc->ctx; struct dc_firmware_info info; struct dc_bios *bp; - struct dm_pp_static_clock_info static_clk_info = {0}; ctx->dc_bios->regs = &bios_regs; @@ -918,11 +917,11 @@ static bool dce80_construct( } } - pool->base.dccg = dce_dccg_create(ctx, + pool->base.clk_mgr = dce_clk_mgr_create(ctx, &disp_clk_regs, &disp_clk_shift, &disp_clk_mask); - if (pool->base.dccg == NULL) { + if (pool->base.clk_mgr == NULL) { dm_error("DC: failed to create display clock!\n"); BREAK_TO_DEBUGGER(); goto res_create_fail; @@ -948,10 +947,6 @@ static bool dce80_construct( goto res_create_fail; } - if (dm_pp_get_static_clocks(ctx, &static_clk_info)) - pool->base.dccg->max_clks_state = - static_clk_info.max_clocks_state; - { struct irq_service_init_data init_data; init_data.ctx = dc->ctx; @@ -1065,7 +1060,6 @@ static bool dce81_construct( struct dc_context *ctx = dc->ctx; struct dc_firmware_info info; struct dc_bios *bp; - struct dm_pp_static_clock_info static_clk_info = {0}; ctx->dc_bios->regs = &bios_regs; @@ -1128,11 +1122,11 @@ static bool dce81_construct( } } - pool->base.dccg = dce_dccg_create(ctx, + pool->base.clk_mgr = dce_clk_mgr_create(ctx, &disp_clk_regs, &disp_clk_shift, &disp_clk_mask); - if (pool->base.dccg == NULL) { + if (pool->base.clk_mgr == NULL) { dm_error("DC: failed to create display clock!\n"); BREAK_TO_DEBUGGER(); goto res_create_fail; @@ -1158,10 +1152,6 @@ static bool dce81_construct( goto res_create_fail; } - if (dm_pp_get_static_clocks(ctx, &static_clk_info)) - pool->base.dccg->max_clks_state = - static_clk_info.max_clocks_state; - { struct irq_service_init_data init_data; init_data.ctx = dc->ctx; @@ -1275,7 +1265,6 @@ static bool dce83_construct( struct dc_context *ctx = dc->ctx; struct dc_firmware_info info; struct dc_bios *bp; - struct dm_pp_static_clock_info static_clk_info = {0}; ctx->dc_bios->regs = &bios_regs; @@ -1334,11 +1323,11 @@ static bool dce83_construct( } } - pool->base.dccg = dce_dccg_create(ctx, + pool->base.clk_mgr = dce_clk_mgr_create(ctx, &disp_clk_regs, &disp_clk_shift, &disp_clk_mask); - if (pool->base.dccg == NULL) { + if (pool->base.clk_mgr == NULL) { dm_error("DC: failed to create display clock!\n"); BREAK_TO_DEBUGGER(); goto res_create_fail; @@ -1364,10 +1353,6 @@ static bool dce83_construct( goto res_create_fail; } - if (dm_pp_get_static_clocks(ctx, &static_clk_info)) - pool->base.dccg->max_clks_state = - static_clk_info.max_clocks_state; - { struct irq_service_init_data init_data; init_data.ctx = dc->ctx; diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/Makefile b/drivers/gpu/drm/amd/display/dc/dcn10/Makefile index 032f872be89c..55f293c8a3c0 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dcn10/Makefile @@ -24,7 +24,7 @@ DCN10 = dcn10_resource.o dcn10_ipp.o dcn10_hw_sequencer.o dcn10_hw_sequencer_debug.o \ dcn10_dpp.o dcn10_opp.o dcn10_optc.o \ - dcn10_hubp.o dcn10_mpc.o \ + dcn10_hubp.o dcn10_mpc.o dcn10_clk_mgr.o \ dcn10_dpp_dscl.o dcn10_dpp_cm.o dcn10_cm_common.o \ dcn10_hubbub.o dcn10_stream_encoder.o dcn10_link_encoder.o diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_clk_mgr.c new file mode 100644 index 000000000000..20f531d27e2b --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_clk_mgr.c @@ -0,0 +1,379 @@ +/* + * Copyright 2018 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. + * + * Authors: AMD + * + */ + +#include "dcn10_clk_mgr.h" + +#include "reg_helper.h" +#include "core_types.h" + +#define TO_DCE_CLK_MGR(clocks)\ + container_of(clocks, struct dce_clk_mgr, base) + +#define REG(reg) \ + (clk_mgr_dce->regs->reg) + +#undef FN +#define FN(reg_name, field_name) \ + clk_mgr_dce->clk_mgr_shift->field_name, clk_mgr_dce->clk_mgr_mask->field_name + +#define CTX \ + clk_mgr_dce->base.ctx +#define DC_LOGGER \ + clk_mgr->ctx->logger + +void dcn1_pplib_apply_display_requirements( + struct dc *dc, + struct dc_state *context) +{ + struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg; + + pp_display_cfg->min_engine_clock_khz = dc->res_pool->clk_mgr->clks.dcfclk_khz; + pp_display_cfg->min_memory_clock_khz = dc->res_pool->clk_mgr->clks.fclk_khz; + pp_display_cfg->min_engine_clock_deep_sleep_khz = dc->res_pool->clk_mgr->clks.dcfclk_deep_sleep_khz; + pp_display_cfg->min_dcfc_deep_sleep_clock_khz = dc->res_pool->clk_mgr->clks.dcfclk_deep_sleep_khz; + pp_display_cfg->min_dcfclock_khz = dc->res_pool->clk_mgr->clks.dcfclk_khz; + pp_display_cfg->disp_clk_khz = dc->res_pool->clk_mgr->clks.dispclk_khz; + dce110_fill_display_configs(context, pp_display_cfg); + + dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg); +} + +static int dcn1_determine_dppclk_threshold(struct clk_mgr *clk_mgr, struct dc_clocks *new_clocks) +{ + bool request_dpp_div = new_clocks->dispclk_khz > new_clocks->dppclk_khz; + bool dispclk_increase = new_clocks->dispclk_khz > clk_mgr->clks.dispclk_khz; + int disp_clk_threshold = new_clocks->max_supported_dppclk_khz; + bool cur_dpp_div = clk_mgr->clks.dispclk_khz > clk_mgr->clks.dppclk_khz; + + /* increase clock, looking for div is 0 for current, request div is 1*/ + if (dispclk_increase) { + /* already divided by 2, no need to reach target clk with 2 steps*/ + if (cur_dpp_div) + return new_clocks->dispclk_khz; + + /* request disp clk is lower than maximum supported dpp clk, + * no need to reach target clk with two steps. + */ + if (new_clocks->dispclk_khz <= disp_clk_threshold) + return new_clocks->dispclk_khz; + + /* target dpp clk not request divided by 2, still within threshold */ + if (!request_dpp_div) + return new_clocks->dispclk_khz; + + } else { + /* decrease clock, looking for current dppclk divided by 2, + * request dppclk not divided by 2. + */ + + /* current dpp clk not divided by 2, no need to ramp*/ + if (!cur_dpp_div) + return new_clocks->dispclk_khz; + + /* current disp clk is lower than current maximum dpp clk, + * no need to ramp + */ + if (clk_mgr->clks.dispclk_khz <= disp_clk_threshold) + return new_clocks->dispclk_khz; + + /* request dpp clk need to be divided by 2 */ + if (request_dpp_div) + return new_clocks->dispclk_khz; + } + + return disp_clk_threshold; +} + +static void dcn1_ramp_up_dispclk_with_dpp(struct clk_mgr *clk_mgr, struct dc_clocks *new_clocks) +{ + struct dc *dc = clk_mgr->ctx->dc; + int dispclk_to_dpp_threshold = dcn1_determine_dppclk_threshold(clk_mgr, new_clocks); + bool request_dpp_div = new_clocks->dispclk_khz > new_clocks->dppclk_khz; + int i; + + /* set disp clk to dpp clk threshold */ + dce112_set_clock(clk_mgr, dispclk_to_dpp_threshold); + + /* update request dpp clk division option */ + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; + + if (!pipe_ctx->plane_state) + continue; + + pipe_ctx->plane_res.dpp->funcs->dpp_dppclk_control( + pipe_ctx->plane_res.dpp, + request_dpp_div, + true); + } + + /* If target clk not same as dppclk threshold, set to target clock */ + if (dispclk_to_dpp_threshold != new_clocks->dispclk_khz) + dce112_set_clock(clk_mgr, new_clocks->dispclk_khz); + + clk_mgr->clks.dispclk_khz = new_clocks->dispclk_khz; + clk_mgr->clks.dppclk_khz = new_clocks->dppclk_khz; + clk_mgr->clks.max_supported_dppclk_khz = new_clocks->max_supported_dppclk_khz; +} + +static int get_active_display_cnt( + struct dc *dc, + struct dc_state *context) +{ + int i, display_count; + + display_count = 0; + for (i = 0; i < context->stream_count; i++) { + const struct dc_stream_state *stream = context->streams[i]; + + /* + * Only notify active stream or virtual stream. + * Need to notify virtual stream to work around + * headless case. HPD does not fire when system is in + * S0i2. + */ + if (!stream->dpms_off || stream->signal == SIGNAL_TYPE_VIRTUAL) + display_count++; + } + + return display_count; +} + +static void notify_deep_sleep_dcfclk_to_smu( + struct pp_smu_funcs_rv *pp_smu, int min_dcef_deep_sleep_clk_khz) +{ + int min_dcef_deep_sleep_clk_mhz; //minimum required DCEF Deep Sleep clock in mhz + /* + * if function pointer not set up, this message is + * sent as part of pplib_apply_display_requirements. + * So just return. + */ + if (!pp_smu || !pp_smu->set_min_deep_sleep_dcfclk) + return; + + min_dcef_deep_sleep_clk_mhz = (min_dcef_deep_sleep_clk_khz + 999) / 1000; //Round up + pp_smu->set_min_deep_sleep_dcfclk(&pp_smu->pp_smu, min_dcef_deep_sleep_clk_mhz); +} + +static void notify_hard_min_dcfclk_to_smu( + struct pp_smu_funcs_rv *pp_smu, int min_dcf_clk_khz) +{ + int min_dcf_clk_mhz; //minimum required DCF clock in mhz + + /* + * if function pointer not set up, this message is + * sent as part of pplib_apply_display_requirements. + * So just return. + */ + if (!pp_smu || !pp_smu->set_hard_min_dcfclk_by_freq) + return; + + min_dcf_clk_mhz = min_dcf_clk_khz / 1000; + + pp_smu->set_hard_min_dcfclk_by_freq(&pp_smu->pp_smu, min_dcf_clk_mhz); +} + +static void notify_hard_min_fclk_to_smu( + struct pp_smu_funcs_rv *pp_smu, int min_f_clk_khz) +{ + int min_f_clk_mhz; //minimum required F clock in mhz + + /* + * if function pointer not set up, this message is + * sent as part of pplib_apply_display_requirements. + * So just return. + */ + if (!pp_smu || !pp_smu->set_hard_min_fclk_by_freq) + return; + + min_f_clk_mhz = min_f_clk_khz / 1000; + + pp_smu->set_hard_min_fclk_by_freq(&pp_smu->pp_smu, min_f_clk_mhz); +} + +static void dcn1_update_clocks(struct clk_mgr *clk_mgr, + struct dc_state *context, + bool safe_to_lower) +{ + struct dc *dc = clk_mgr->ctx->dc; + struct dc_clocks *new_clocks = &context->bw.dcn.clk; + struct pp_smu_display_requirement_rv *smu_req_cur = + &dc->res_pool->pp_smu_req; + struct pp_smu_display_requirement_rv smu_req = *smu_req_cur; + struct pp_smu_funcs_rv *pp_smu = dc->res_pool->pp_smu; + struct dm_pp_clock_for_voltage_req clock_voltage_req = {0}; + bool send_request_to_increase = false; + bool send_request_to_lower = false; + int display_count; + + bool enter_display_off = false; + + display_count = get_active_display_cnt(dc, context); + + if (display_count == 0) + enter_display_off = true; + + if (enter_display_off == safe_to_lower) { + /* + * Notify SMU active displays + * if function pointer not set up, this message is + * sent as part of pplib_apply_display_requirements. + */ + if (pp_smu->set_display_count) + pp_smu->set_display_count(&pp_smu->pp_smu, display_count); + else + smu_req.display_count = display_count; + + } + + if (new_clocks->dispclk_khz > clk_mgr->clks.dispclk_khz + || new_clocks->phyclk_khz > clk_mgr->clks.phyclk_khz + || new_clocks->fclk_khz > clk_mgr->clks.fclk_khz + || new_clocks->dcfclk_khz > clk_mgr->clks.dcfclk_khz) + send_request_to_increase = true; + + if (should_set_clock(safe_to_lower, new_clocks->phyclk_khz, clk_mgr->clks.phyclk_khz)) { + clk_mgr->clks.phyclk_khz = new_clocks->phyclk_khz; + + send_request_to_lower = true; + } + + // F Clock + if (should_set_clock(safe_to_lower, new_clocks->fclk_khz, clk_mgr->clks.fclk_khz)) { + clk_mgr->clks.fclk_khz = new_clocks->fclk_khz; + clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_FCLK; + clock_voltage_req.clocks_in_khz = new_clocks->fclk_khz; + smu_req.hard_min_fclk_mhz = new_clocks->fclk_khz / 1000; + + notify_hard_min_fclk_to_smu(pp_smu, new_clocks->fclk_khz); + + send_request_to_lower = true; + } + + //DCF Clock + if (should_set_clock(safe_to_lower, new_clocks->dcfclk_khz, clk_mgr->clks.dcfclk_khz)) { + clk_mgr->clks.dcfclk_khz = new_clocks->dcfclk_khz; + smu_req.hard_min_dcefclk_mhz = new_clocks->dcfclk_khz / 1000; + + send_request_to_lower = true; + } + + if (should_set_clock(safe_to_lower, + new_clocks->dcfclk_deep_sleep_khz, clk_mgr->clks.dcfclk_deep_sleep_khz)) { + clk_mgr->clks.dcfclk_deep_sleep_khz = new_clocks->dcfclk_deep_sleep_khz; + smu_req.min_deep_sleep_dcefclk_mhz = new_clocks->dcfclk_deep_sleep_khz / 1000; + + send_request_to_lower = true; + } + + /* make sure dcf clk is before dpp clk to + * make sure we have enough voltage to run dpp clk + */ + if (send_request_to_increase) { + /*use dcfclk to request voltage*/ + clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DCFCLK; + clock_voltage_req.clocks_in_khz = dcn_find_dcfclk_suits_all(dc, new_clocks); + + notify_hard_min_dcfclk_to_smu(pp_smu, clock_voltage_req.clocks_in_khz); + + if (pp_smu->set_display_requirement) + pp_smu->set_display_requirement(&pp_smu->pp_smu, &smu_req); + + notify_deep_sleep_dcfclk_to_smu(pp_smu, clk_mgr->clks.dcfclk_deep_sleep_khz); + dcn1_pplib_apply_display_requirements(dc, context); + } + + /* dcn1 dppclk is tied to dispclk */ + /* program dispclk on = as a w/a for sleep resume clock ramping issues */ + if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr->clks.dispclk_khz) + || new_clocks->dispclk_khz == clk_mgr->clks.dispclk_khz) { + dcn1_ramp_up_dispclk_with_dpp(clk_mgr, new_clocks); + clk_mgr->clks.dispclk_khz = new_clocks->dispclk_khz; + + send_request_to_lower = true; + } + + if (!send_request_to_increase && send_request_to_lower) { + /*use dcfclk to request voltage*/ + clock_voltage_req.clk_type = DM_PP_CLOCK_TYPE_DCFCLK; + clock_voltage_req.clocks_in_khz = dcn_find_dcfclk_suits_all(dc, new_clocks); + + notify_hard_min_dcfclk_to_smu(pp_smu, clock_voltage_req.clocks_in_khz); + + if (pp_smu->set_display_requirement) + pp_smu->set_display_requirement(&pp_smu->pp_smu, &smu_req); + + notify_deep_sleep_dcfclk_to_smu(pp_smu, clk_mgr->clks.dcfclk_deep_sleep_khz); + dcn1_pplib_apply_display_requirements(dc, context); + } + + + *smu_req_cur = smu_req; +} + +static const struct clk_mgr_funcs dcn1_funcs = { + .get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz, + .update_clocks = dcn1_update_clocks +}; + +struct clk_mgr *dcn1_clk_mgr_create(struct dc_context *ctx) +{ + struct dc_debug_options *debug = &ctx->dc->debug; + struct dc_bios *bp = ctx->dc_bios; + struct dc_firmware_info fw_info = { { 0 } }; + struct dce_clk_mgr *clk_mgr_dce = kzalloc(sizeof(*clk_mgr_dce), GFP_KERNEL); + + if (clk_mgr_dce == NULL) { + BREAK_TO_DEBUGGER(); + return NULL; + } + + clk_mgr_dce->base.ctx = ctx; + clk_mgr_dce->base.funcs = &dcn1_funcs; + + clk_mgr_dce->dfs_bypass_disp_clk = 0; + + clk_mgr_dce->dprefclk_ss_percentage = 0; + clk_mgr_dce->dprefclk_ss_divider = 1000; + clk_mgr_dce->ss_on_dprefclk = false; + + clk_mgr_dce->dprefclk_khz = 600000; + if (bp->integrated_info) + clk_mgr_dce->dentist_vco_freq_khz = bp->integrated_info->dentist_vco_freq; + if (clk_mgr_dce->dentist_vco_freq_khz == 0) { + bp->funcs->get_firmware_info(bp, &fw_info); + clk_mgr_dce->dentist_vco_freq_khz = fw_info.smu_gpu_pll_output_freq; + if (clk_mgr_dce->dentist_vco_freq_khz == 0) + clk_mgr_dce->dentist_vco_freq_khz = 3600000; + } + + if (!debug->disable_dfs_bypass && bp->integrated_info) + if (bp->integrated_info->gpu_cap_info & DFS_BYPASS_ENABLE) + clk_mgr_dce->dfs_bypass_enabled = true; + + dce_clock_read_ss_info(clk_mgr_dce); + + return &clk_mgr_dce->base; +} diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_clk_mgr.h new file mode 100644 index 000000000000..9dbaf6578006 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_clk_mgr.h @@ -0,0 +1,37 @@ +/* + * Copyright 2018 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. + * + * Authors: AMD + * + */ + +#ifndef __DCN10_CLK_MGR_H__ +#define __DCN10_CLK_MGR_H__ + +#include "../dce/dce_clk_mgr.h" + +void dcn1_pplib_apply_display_requirements( + struct dc *dc, + struct dc_state *context); + +struct clk_mgr *dcn1_clk_mgr_create(struct dc_context *ctx); + +#endif //__DCN10_CLK_MGR_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c index 5d95a997fd9f..3eea44092a04 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.c @@ -71,39 +71,39 @@ void cm_helper_program_xfer_func( unsigned int i = 0; REG_SET_2(reg->start_cntl_b, 0, - exp_region_start, params->arr_points[0].custom_float_x, + exp_region_start, params->corner_points[0].blue.custom_float_x, exp_resion_start_segment, 0); REG_SET_2(reg->start_cntl_g, 0, - exp_region_start, params->arr_points[0].custom_float_x, + exp_region_start, params->corner_points[0].green.custom_float_x, exp_resion_start_segment, 0); REG_SET_2(reg->start_cntl_r, 0, - exp_region_start, params->arr_points[0].custom_float_x, + exp_region_start, params->corner_points[0].red.custom_float_x, exp_resion_start_segment, 0); REG_SET(reg->start_slope_cntl_b, 0, - field_region_linear_slope, params->arr_points[0].custom_float_slope); + field_region_linear_slope, params->corner_points[0].blue.custom_float_slope); REG_SET(reg->start_slope_cntl_g, 0, - field_region_linear_slope, params->arr_points[0].custom_float_slope); + field_region_linear_slope, params->corner_points[0].green.custom_float_slope); REG_SET(reg->start_slope_cntl_r, 0, - field_region_linear_slope, params->arr_points[0].custom_float_slope); + field_region_linear_slope, params->corner_points[0].red.custom_float_slope); REG_SET(reg->start_end_cntl1_b, 0, - field_region_end, params->arr_points[1].custom_float_x); + field_region_end, params->corner_points[1].blue.custom_float_x); REG_SET_2(reg->start_end_cntl2_b, 0, - field_region_end_slope, params->arr_points[1].custom_float_slope, - field_region_end_base, params->arr_points[1].custom_float_y); + field_region_end_slope, params->corner_points[1].blue.custom_float_slope, + field_region_end_base, params->corner_points[1].blue.custom_float_y); REG_SET(reg->start_end_cntl1_g, 0, - field_region_end, params->arr_points[1].custom_float_x); + field_region_end, params->corner_points[1].green.custom_float_x); REG_SET_2(reg->start_end_cntl2_g, 0, - field_region_end_slope, params->arr_points[1].custom_float_slope, - field_region_end_base, params->arr_points[1].custom_float_y); + field_region_end_slope, params->corner_points[1].green.custom_float_slope, + field_region_end_base, params->corner_points[1].green.custom_float_y); REG_SET(reg->start_end_cntl1_r, 0, - field_region_end, params->arr_points[1].custom_float_x); + field_region_end, params->corner_points[1].red.custom_float_x); REG_SET_2(reg->start_end_cntl2_r, 0, - field_region_end_slope, params->arr_points[1].custom_float_slope, - field_region_end_base, params->arr_points[1].custom_float_y); + field_region_end_slope, params->corner_points[1].red.custom_float_slope, + field_region_end_base, params->corner_points[1].red.custom_float_y); for (reg_region_cur = reg->region_start; reg_region_cur <= reg->region_end; @@ -127,7 +127,7 @@ void cm_helper_program_xfer_func( bool cm_helper_convert_to_custom_float( struct pwl_result_data *rgb_resulted, - struct curve_points *arr_points, + struct curve_points3 *corner_points, uint32_t hw_points_num, bool fixpoint) { @@ -141,20 +141,53 @@ bool cm_helper_convert_to_custom_float( fmt.mantissa_bits = 12; fmt.sign = false; - if (!convert_to_custom_float_format(arr_points[0].x, &fmt, - &arr_points[0].custom_float_x)) { + /* corner_points[0] - beginning base, slope offset for R,G,B + * corner_points[1] - end base, slope offset for R,G,B + */ + if (!convert_to_custom_float_format(corner_points[0].red.x, &fmt, + &corner_points[0].red.custom_float_x)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[0].green.x, &fmt, + &corner_points[0].green.custom_float_x)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[0].blue.x, &fmt, + &corner_points[0].blue.custom_float_x)) { BREAK_TO_DEBUGGER(); return false; } - if (!convert_to_custom_float_format(arr_points[0].offset, &fmt, - &arr_points[0].custom_float_offset)) { + if (!convert_to_custom_float_format(corner_points[0].red.offset, &fmt, + &corner_points[0].red.custom_float_offset)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[0].green.offset, &fmt, + &corner_points[0].green.custom_float_offset)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[0].blue.offset, &fmt, + &corner_points[0].blue.custom_float_offset)) { BREAK_TO_DEBUGGER(); return false; } - if (!convert_to_custom_float_format(arr_points[0].slope, &fmt, - &arr_points[0].custom_float_slope)) { + if (!convert_to_custom_float_format(corner_points[0].red.slope, &fmt, + &corner_points[0].red.custom_float_slope)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[0].green.slope, &fmt, + &corner_points[0].green.custom_float_slope)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[0].blue.slope, &fmt, + &corner_points[0].blue.custom_float_slope)) { BREAK_TO_DEBUGGER(); return false; } @@ -162,22 +195,59 @@ bool cm_helper_convert_to_custom_float( fmt.mantissa_bits = 10; fmt.sign = false; - if (!convert_to_custom_float_format(arr_points[1].x, &fmt, - &arr_points[1].custom_float_x)) { + if (!convert_to_custom_float_format(corner_points[1].red.x, &fmt, + &corner_points[1].red.custom_float_x)) { BREAK_TO_DEBUGGER(); return false; } - - if (fixpoint == true) - arr_points[1].custom_float_y = dc_fixpt_clamp_u0d14(arr_points[1].y); - else if (!convert_to_custom_float_format(arr_points[1].y, &fmt, - &arr_points[1].custom_float_y)) { + if (!convert_to_custom_float_format(corner_points[1].green.x, &fmt, + &corner_points[1].green.custom_float_x)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[1].blue.x, &fmt, + &corner_points[1].blue.custom_float_x)) { BREAK_TO_DEBUGGER(); return false; } - if (!convert_to_custom_float_format(arr_points[1].slope, &fmt, - &arr_points[1].custom_float_slope)) { + if (fixpoint == true) { + corner_points[1].red.custom_float_y = + dc_fixpt_clamp_u0d14(corner_points[1].red.y); + corner_points[1].green.custom_float_y = + dc_fixpt_clamp_u0d14(corner_points[1].green.y); + corner_points[1].blue.custom_float_y = + dc_fixpt_clamp_u0d14(corner_points[1].blue.y); + } else { + if (!convert_to_custom_float_format(corner_points[1].red.y, + &fmt, &corner_points[1].red.custom_float_y)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[1].green.y, + &fmt, &corner_points[1].green.custom_float_y)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[1].blue.y, + &fmt, &corner_points[1].blue.custom_float_y)) { + BREAK_TO_DEBUGGER(); + return false; + } + } + + if (!convert_to_custom_float_format(corner_points[1].red.slope, &fmt, + &corner_points[1].red.custom_float_slope)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[1].green.slope, &fmt, + &corner_points[1].green.custom_float_slope)) { + BREAK_TO_DEBUGGER(); + return false; + } + if (!convert_to_custom_float_format(corner_points[1].blue.slope, &fmt, + &corner_points[1].blue.custom_float_slope)) { BREAK_TO_DEBUGGER(); return false; } @@ -242,15 +312,10 @@ bool cm_helper_translate_curve_to_hw_format( const struct dc_transfer_func *output_tf, struct pwl_params *lut_params, bool fixpoint) { - struct curve_points *arr_points; + struct curve_points3 *corner_points; struct pwl_result_data *rgb_resulted; struct pwl_result_data *rgb; struct pwl_result_data *rgb_plus_1; - struct fixed31_32 y_r; - struct fixed31_32 y_g; - struct fixed31_32 y_b; - struct fixed31_32 y1_min; - struct fixed31_32 y3_max; int32_t region_start, region_end; int32_t i; @@ -261,14 +326,14 @@ bool cm_helper_translate_curve_to_hw_format( PERF_TRACE(); - arr_points = lut_params->arr_points; + corner_points = lut_params->corner_points; rgb_resulted = lut_params->rgb_resulted; hw_points = 0; memset(lut_params, 0, sizeof(struct pwl_params)); memset(seg_distr, 0, sizeof(seg_distr)); - if (output_tf->tf == TRANSFER_FUNCTION_PQ) { + if (output_tf->tf == TRANSFER_FUNCTION_PQ || output_tf->tf == TRANSFER_FUNCTION_GAMMA22) { /* 32 segments * segments are from 2^-25 to 2^7 */ @@ -327,31 +392,37 @@ bool cm_helper_translate_curve_to_hw_format( rgb_resulted[hw_points - 1].green = output_tf->tf_pts.green[start_index]; rgb_resulted[hw_points - 1].blue = output_tf->tf_pts.blue[start_index]; - arr_points[0].x = dc_fixpt_pow(dc_fixpt_from_int(2), + // All 3 color channels have same x + corner_points[0].red.x = dc_fixpt_pow(dc_fixpt_from_int(2), dc_fixpt_from_int(region_start)); - arr_points[1].x = dc_fixpt_pow(dc_fixpt_from_int(2), - dc_fixpt_from_int(region_end)); + corner_points[0].green.x = corner_points[0].red.x; + corner_points[0].blue.x = corner_points[0].red.x; - y_r = rgb_resulted[0].red; - y_g = rgb_resulted[0].green; - y_b = rgb_resulted[0].blue; + corner_points[1].red.x = dc_fixpt_pow(dc_fixpt_from_int(2), + dc_fixpt_from_int(region_end)); + corner_points[1].green.x = corner_points[1].red.x; + corner_points[1].blue.x = corner_points[1].red.x; - y1_min = dc_fixpt_min(y_r, dc_fixpt_min(y_g, y_b)); + corner_points[0].red.y = rgb_resulted[0].red; + corner_points[0].green.y = rgb_resulted[0].green; + corner_points[0].blue.y = rgb_resulted[0].blue; - arr_points[0].y = y1_min; - arr_points[0].slope = dc_fixpt_div(arr_points[0].y, arr_points[0].x); - y_r = rgb_resulted[hw_points - 1].red; - y_g = rgb_resulted[hw_points - 1].green; - y_b = rgb_resulted[hw_points - 1].blue; + corner_points[0].red.slope = dc_fixpt_div(corner_points[0].red.y, + corner_points[0].red.x); + corner_points[0].green.slope = dc_fixpt_div(corner_points[0].green.y, + corner_points[0].green.x); + corner_points[0].blue.slope = dc_fixpt_div(corner_points[0].blue.y, + corner_points[0].blue.x); /* see comment above, m_arrPoints[1].y should be the Y value for the * region end (m_numOfHwPoints), not last HW point(m_numOfHwPoints - 1) */ - y3_max = dc_fixpt_max(y_r, dc_fixpt_max(y_g, y_b)); - - arr_points[1].y = y3_max; - - arr_points[1].slope = dc_fixpt_zero; + corner_points[1].red.y = rgb_resulted[hw_points - 1].red; + corner_points[1].green.y = rgb_resulted[hw_points - 1].green; + corner_points[1].blue.y = rgb_resulted[hw_points - 1].blue; + corner_points[1].red.slope = dc_fixpt_zero; + corner_points[1].green.slope = dc_fixpt_zero; + corner_points[1].blue.slope = dc_fixpt_zero; if (output_tf->tf == TRANSFER_FUNCTION_PQ) { /* for PQ, we want to have a straight line from last HW X point, @@ -360,9 +431,15 @@ bool cm_helper_translate_curve_to_hw_format( const struct fixed31_32 end_value = dc_fixpt_from_int(125); - arr_points[1].slope = dc_fixpt_div( - dc_fixpt_sub(dc_fixpt_one, arr_points[1].y), - dc_fixpt_sub(end_value, arr_points[1].x)); + corner_points[1].red.slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, corner_points[1].red.y), + dc_fixpt_sub(end_value, corner_points[1].red.x)); + corner_points[1].green.slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, corner_points[1].green.y), + dc_fixpt_sub(end_value, corner_points[1].green.x)); + corner_points[1].blue.slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, corner_points[1].blue.y), + dc_fixpt_sub(end_value, corner_points[1].blue.x)); } lut_params->hw_points_num = hw_points; @@ -411,7 +488,7 @@ bool cm_helper_translate_curve_to_hw_format( ++i; } cm_helper_convert_to_custom_float(rgb_resulted, - lut_params->arr_points, + lut_params->corner_points, hw_points, fixpoint); return true; @@ -424,15 +501,10 @@ bool cm_helper_translate_curve_to_degamma_hw_format( const struct dc_transfer_func *output_tf, struct pwl_params *lut_params) { - struct curve_points *arr_points; + struct curve_points3 *corner_points; struct pwl_result_data *rgb_resulted; struct pwl_result_data *rgb; struct pwl_result_data *rgb_plus_1; - struct fixed31_32 y_r; - struct fixed31_32 y_g; - struct fixed31_32 y_b; - struct fixed31_32 y1_min; - struct fixed31_32 y3_max; int32_t region_start, region_end; int32_t i; @@ -443,7 +515,7 @@ bool cm_helper_translate_curve_to_degamma_hw_format( PERF_TRACE(); - arr_points = lut_params->arr_points; + corner_points = lut_params->corner_points; rgb_resulted = lut_params->rgb_resulted; hw_points = 0; @@ -489,31 +561,28 @@ bool cm_helper_translate_curve_to_degamma_hw_format( rgb_resulted[hw_points - 1].green = output_tf->tf_pts.green[start_index]; rgb_resulted[hw_points - 1].blue = output_tf->tf_pts.blue[start_index]; - arr_points[0].x = dc_fixpt_pow(dc_fixpt_from_int(2), + corner_points[0].red.x = dc_fixpt_pow(dc_fixpt_from_int(2), dc_fixpt_from_int(region_start)); - arr_points[1].x = dc_fixpt_pow(dc_fixpt_from_int(2), + corner_points[0].green.x = corner_points[0].red.x; + corner_points[0].blue.x = corner_points[0].red.x; + corner_points[1].red.x = dc_fixpt_pow(dc_fixpt_from_int(2), dc_fixpt_from_int(region_end)); + corner_points[1].green.x = corner_points[1].red.x; + corner_points[1].blue.x = corner_points[1].red.x; - y_r = rgb_resulted[0].red; - y_g = rgb_resulted[0].green; - y_b = rgb_resulted[0].blue; - - y1_min = dc_fixpt_min(y_r, dc_fixpt_min(y_g, y_b)); - - arr_points[0].y = y1_min; - arr_points[0].slope = dc_fixpt_div(arr_points[0].y, arr_points[0].x); - y_r = rgb_resulted[hw_points - 1].red; - y_g = rgb_resulted[hw_points - 1].green; - y_b = rgb_resulted[hw_points - 1].blue; + corner_points[0].red.y = rgb_resulted[0].red; + corner_points[0].green.y = rgb_resulted[0].green; + corner_points[0].blue.y = rgb_resulted[0].blue; /* see comment above, m_arrPoints[1].y should be the Y value for the * region end (m_numOfHwPoints), not last HW point(m_numOfHwPoints - 1) */ - y3_max = dc_fixpt_max(y_r, dc_fixpt_max(y_g, y_b)); - - arr_points[1].y = y3_max; - - arr_points[1].slope = dc_fixpt_zero; + corner_points[1].red.y = rgb_resulted[hw_points - 1].red; + corner_points[1].green.y = rgb_resulted[hw_points - 1].green; + corner_points[1].blue.y = rgb_resulted[hw_points - 1].blue; + corner_points[1].red.slope = dc_fixpt_zero; + corner_points[1].green.slope = dc_fixpt_zero; + corner_points[1].blue.slope = dc_fixpt_zero; if (output_tf->tf == TRANSFER_FUNCTION_PQ) { /* for PQ, we want to have a straight line from last HW X point, @@ -522,9 +591,15 @@ bool cm_helper_translate_curve_to_degamma_hw_format( const struct fixed31_32 end_value = dc_fixpt_from_int(125); - arr_points[1].slope = dc_fixpt_div( - dc_fixpt_sub(dc_fixpt_one, arr_points[1].y), - dc_fixpt_sub(end_value, arr_points[1].x)); + corner_points[1].red.slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, corner_points[1].red.y), + dc_fixpt_sub(end_value, corner_points[1].red.x)); + corner_points[1].green.slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, corner_points[1].green.y), + dc_fixpt_sub(end_value, corner_points[1].green.x)); + corner_points[1].blue.slope = dc_fixpt_div( + dc_fixpt_sub(dc_fixpt_one, corner_points[1].blue.y), + dc_fixpt_sub(end_value, corner_points[1].blue.x)); } lut_params->hw_points_num = hw_points; @@ -564,7 +639,7 @@ bool cm_helper_translate_curve_to_degamma_hw_format( ++i; } cm_helper_convert_to_custom_float(rgb_resulted, - lut_params->arr_points, + lut_params->corner_points, hw_points, false); return true; diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.h index 7a531b02871f..5ae4d69391a5 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_cm_common.h @@ -98,7 +98,7 @@ void cm_helper_program_xfer_func( bool cm_helper_convert_to_custom_float( struct pwl_result_data *rgb_resulted, - struct curve_points *arr_points, + struct curve_points3 *corner_points, uint32_t hw_points_num, bool fixpoint); diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c index 193184affefb..87495dea45ec 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c @@ -45,6 +45,7 @@ #include "dcn10_hubbub.h" #include "dcn10_cm_common.h" #include "dc_link_dp.h" +#include "dccg.h" #define DC_LOGGER_INIT(logger) @@ -786,7 +787,7 @@ static bool dcn10_hw_wa_force_recovery(struct dc *dc) &dc->current_state->res_ctx.pipe_ctx[i]; if (pipe_ctx != NULL) { hubp = pipe_ctx->plane_res.hubp; - if (hubp != NULL) { + if (hubp != NULL && hubp->funcs->hubp_get_underflow_status) { if (hubp->funcs->hubp_get_underflow_status(hubp) != 0) { /* one pipe underflow, we will reset all the pipes*/ need_recover = true; @@ -812,7 +813,7 @@ static bool dcn10_hw_wa_force_recovery(struct dc *dc) if (pipe_ctx != NULL) { hubp = pipe_ctx->plane_res.hubp; /*DCHUBP_CNTL:HUBP_BLANK_EN=1*/ - if (hubp != NULL) + if (hubp != NULL && hubp->funcs->set_hubp_blank_en) hubp->funcs->set_hubp_blank_en(hubp, true); } } @@ -825,7 +826,7 @@ static bool dcn10_hw_wa_force_recovery(struct dc *dc) if (pipe_ctx != NULL) { hubp = pipe_ctx->plane_res.hubp; /*DCHUBP_CNTL:HUBP_DISABLE=1*/ - if (hubp != NULL) + if (hubp != NULL && hubp->funcs->hubp_disable_control) hubp->funcs->hubp_disable_control(hubp, true); } } @@ -835,7 +836,7 @@ static bool dcn10_hw_wa_force_recovery(struct dc *dc) if (pipe_ctx != NULL) { hubp = pipe_ctx->plane_res.hubp; /*DCHUBP_CNTL:HUBP_DISABLE=0*/ - if (hubp != NULL) + if (hubp != NULL && hubp->funcs->hubp_disable_control) hubp->funcs->hubp_disable_control(hubp, true); } } @@ -847,7 +848,7 @@ static bool dcn10_hw_wa_force_recovery(struct dc *dc) if (pipe_ctx != NULL) { hubp = pipe_ctx->plane_res.hubp; /*DCHUBP_CNTL:HUBP_BLANK_EN=0*/ - if (hubp != NULL) + if (hubp != NULL && hubp->funcs->set_hubp_blank_en) hubp->funcs->set_hubp_blank_en(hubp, true); } } @@ -1126,7 +1127,7 @@ static void dcn10_init_hw(struct dc *dc) enable_power_gating_plane(dc->hwseq, true); - memset(&dc->res_pool->dccg->clks, 0, sizeof(dc->res_pool->dccg->clks)); + memset(&dc->res_pool->clk_mgr->clks, 0, sizeof(dc->res_pool->clk_mgr->clks)); } static void reset_hw_ctx_wrap( @@ -1603,7 +1604,7 @@ static void mmhub_read_vm_context0_settings(struct dcn10_hubp *hubp1, } -static void dcn10_program_pte_vm(struct dce_hwseq *hws, struct hubp *hubp) +void dcn10_program_pte_vm(struct dce_hwseq *hws, struct hubp *hubp) { struct dcn10_hubp *hubp1 = TO_DCN10_HUBP(hubp); struct vm_system_aperture_param apt = { {{ 0 } } }; @@ -1703,33 +1704,22 @@ static void program_gamut_remap(struct pipe_ctx *pipe_ctx) pipe_ctx->plane_res.dpp->funcs->dpp_set_gamut_remap(pipe_ctx->plane_res.dpp, &adjust); } - -static void program_csc_matrix(struct pipe_ctx *pipe_ctx, +static void dcn10_program_output_csc(struct dc *dc, + struct pipe_ctx *pipe_ctx, enum dc_color_space colorspace, - uint16_t *matrix) + uint16_t *matrix, + int opp_id) { if (pipe_ctx->stream->csc_color_matrix.enable_adjustment == true) { - if (pipe_ctx->plane_res.dpp->funcs->dpp_set_csc_adjustment != NULL) - pipe_ctx->plane_res.dpp->funcs->dpp_set_csc_adjustment(pipe_ctx->plane_res.dpp, matrix); + if (pipe_ctx->plane_res.dpp->funcs->dpp_set_csc_adjustment != NULL) + pipe_ctx->plane_res.dpp->funcs->dpp_set_csc_adjustment(pipe_ctx->plane_res.dpp, matrix); } else { if (pipe_ctx->plane_res.dpp->funcs->dpp_set_csc_default != NULL) pipe_ctx->plane_res.dpp->funcs->dpp_set_csc_default(pipe_ctx->plane_res.dpp, colorspace); } } -static void dcn10_program_output_csc(struct dc *dc, - struct pipe_ctx *pipe_ctx, - enum dc_color_space colorspace, - uint16_t *matrix, - int opp_id) -{ - if (pipe_ctx->plane_res.dpp->funcs->dpp_set_csc_adjustment != NULL) - program_csc_matrix(pipe_ctx, - colorspace, - matrix); -} - -static bool is_lower_pipe_tree_visible(struct pipe_ctx *pipe_ctx) +bool is_lower_pipe_tree_visible(struct pipe_ctx *pipe_ctx) { if (pipe_ctx->plane_state->visible) return true; @@ -1738,7 +1728,7 @@ static bool is_lower_pipe_tree_visible(struct pipe_ctx *pipe_ctx) return false; } -static bool is_upper_pipe_tree_visible(struct pipe_ctx *pipe_ctx) +bool is_upper_pipe_tree_visible(struct pipe_ctx *pipe_ctx) { if (pipe_ctx->plane_state->visible) return true; @@ -1747,7 +1737,7 @@ static bool is_upper_pipe_tree_visible(struct pipe_ctx *pipe_ctx) return false; } -static bool is_pipe_tree_visible(struct pipe_ctx *pipe_ctx) +bool is_pipe_tree_visible(struct pipe_ctx *pipe_ctx) { if (pipe_ctx->plane_state->visible) return true; @@ -1943,10 +1933,6 @@ static void dcn10_update_mpcc(struct dc *dc, struct pipe_ctx *pipe_ctx) struct mpc *mpc = dc->res_pool->mpc; struct mpc_tree *mpc_tree_params = &(pipe_ctx->stream_res.opp->mpc_tree_params); - - - /* TODO: proper fix once fpga works */ - if (dc->debug.visual_confirm == VISUAL_CONFIRM_HDR) { dcn10_get_hdr_visual_confirm_color( pipe_ctx, &blnd_cfg.black_color); @@ -2026,8 +2012,6 @@ static void update_scaler(struct pipe_ctx *pipe_ctx) bool per_pixel_alpha = pipe_ctx->plane_state->per_pixel_alpha && pipe_ctx->bottom_pipe; - /* TODO: proper fix once fpga works */ - pipe_ctx->plane_res.scl_data.lb_params.alpha_en = per_pixel_alpha; pipe_ctx->plane_res.scl_data.lb_params.depth = LB_PIXEL_DEPTH_30BPP; /* scaler configuration */ @@ -2035,7 +2019,7 @@ static void update_scaler(struct pipe_ctx *pipe_ctx) pipe_ctx->plane_res.dpp, &pipe_ctx->plane_res.scl_data); } -static void update_dchubp_dpp( +void update_dchubp_dpp( struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_state *context) @@ -2052,16 +2036,22 @@ static void update_dchubp_dpp( */ if (plane_state->update_flags.bits.full_update) { bool should_divided_by_2 = context->bw.dcn.clk.dppclk_khz <= - dc->res_pool->dccg->clks.dispclk_khz / 2; + dc->res_pool->clk_mgr->clks.dispclk_khz / 2; dpp->funcs->dpp_dppclk_control( dpp, should_divided_by_2, true); - dc->res_pool->dccg->clks.dppclk_khz = should_divided_by_2 ? - dc->res_pool->dccg->clks.dispclk_khz / 2 : - dc->res_pool->dccg->clks.dispclk_khz; + if (dc->res_pool->dccg) + dc->res_pool->dccg->funcs->update_dpp_dto( + dc->res_pool->dccg, + dpp->inst, + pipe_ctx->plane_res.bw.calc.dppclk_khz); + else + dc->res_pool->clk_mgr->clks.dppclk_khz = should_divided_by_2 ? + dc->res_pool->clk_mgr->clks.dispclk_khz / 2 : + dc->res_pool->clk_mgr->clks.dispclk_khz; } /* TODO: Need input parameter to tell current DCHUB pipe tie to which OTG @@ -2182,7 +2172,7 @@ static void dcn10_blank_pixel_data( } } -static void set_hdr_multiplier(struct pipe_ctx *pipe_ctx) +void set_hdr_multiplier(struct pipe_ctx *pipe_ctx) { struct fixed31_32 multiplier = dc_fixpt_from_fraction( pipe_ctx->plane_state->sdr_white_level, 80); @@ -2257,47 +2247,7 @@ static void program_all_pipe_in_tree( } } -static void dcn10_pplib_apply_display_requirements( - struct dc *dc, - struct dc_state *context) -{ - struct dm_pp_display_configuration *pp_display_cfg = &context->pp_display_cfg; - - pp_display_cfg->min_engine_clock_khz = dc->res_pool->dccg->clks.dcfclk_khz; - pp_display_cfg->min_memory_clock_khz = dc->res_pool->dccg->clks.fclk_khz; - pp_display_cfg->min_engine_clock_deep_sleep_khz = dc->res_pool->dccg->clks.dcfclk_deep_sleep_khz; - pp_display_cfg->min_dcfc_deep_sleep_clock_khz = dc->res_pool->dccg->clks.dcfclk_deep_sleep_khz; - pp_display_cfg->min_dcfclock_khz = dc->res_pool->dccg->clks.dcfclk_khz; - pp_display_cfg->disp_clk_khz = dc->res_pool->dccg->clks.dispclk_khz; - dce110_fill_display_configs(context, pp_display_cfg); - - if (memcmp(&dc->prev_display_config, pp_display_cfg, sizeof( - struct dm_pp_display_configuration)) != 0) - dm_pp_apply_display_requirements(dc->ctx, pp_display_cfg); - - dc->prev_display_config = *pp_display_cfg; -} - -static void optimize_shared_resources(struct dc *dc) -{ - if (dc->current_state->stream_count == 0) { - /* S0i2 message */ - dcn10_pplib_apply_display_requirements(dc, dc->current_state); - } - - if (dc->debug.pplib_wm_report_mode == WM_REPORT_OVERRIDE) - dcn_bw_notify_pplib_of_wm_ranges(dc); -} - -static void ready_shared_resources(struct dc *dc, struct dc_state *context) -{ - /* S0i2 message */ - if (dc->current_state->stream_count == 0 && - context->stream_count != 0) - dcn10_pplib_apply_display_requirements(dc, context); -} - -static struct pipe_ctx *find_top_pipe_for_stream( +struct pipe_ctx *find_top_pipe_for_stream( struct dc *dc, struct dc_state *context, const struct dc_stream_state *stream) @@ -2398,10 +2348,9 @@ static void dcn10_apply_ctx_for_surface( hubbub1_wm_change_req_wa(dc->res_pool->hubbub); } -static void dcn10_set_bandwidth( +static void dcn10_prepare_bandwidth( struct dc *dc, - struct dc_state *context, - bool safe_to_lower) + struct dc_state *context) { if (dc->debug.sanity_checks) dcn10_verify_allow_pstate_change_high(dc); @@ -2410,12 +2359,39 @@ static void dcn10_set_bandwidth( if (context->stream_count == 0) context->bw.dcn.clk.phyclk_khz = 0; - dc->res_pool->dccg->funcs->update_clocks( - dc->res_pool->dccg, - &context->bw.dcn.clk, - safe_to_lower); + dc->res_pool->clk_mgr->funcs->update_clocks( + dc->res_pool->clk_mgr, + context, + false); + } - dcn10_pplib_apply_display_requirements(dc, context); + hubbub1_program_watermarks(dc->res_pool->hubbub, + &context->bw.dcn.watermarks, + dc->res_pool->ref_clock_inKhz / 1000, + true); + + if (dc->debug.pplib_wm_report_mode == WM_REPORT_OVERRIDE) + dcn_bw_notify_pplib_of_wm_ranges(dc); + + if (dc->debug.sanity_checks) + dcn10_verify_allow_pstate_change_high(dc); +} + +static void dcn10_optimize_bandwidth( + struct dc *dc, + struct dc_state *context) +{ + if (dc->debug.sanity_checks) + dcn10_verify_allow_pstate_change_high(dc); + + if (!IS_FPGA_MAXIMUS_DC(dc->ctx->dce_environment)) { + if (context->stream_count == 0) + context->bw.dcn.clk.phyclk_khz = 0; + + dc->res_pool->clk_mgr->funcs->update_clocks( + dc->res_pool->clk_mgr, + context, + true); } hubbub1_program_watermarks(dc->res_pool->hubbub, @@ -2423,6 +2399,9 @@ static void dcn10_set_bandwidth( dc->res_pool->ref_clock_inKhz / 1000, true); + if (dc->debug.pplib_wm_report_mode == WM_REPORT_OVERRIDE) + dcn_bw_notify_pplib_of_wm_ranges(dc); + if (dc->debug.sanity_checks) dcn10_verify_allow_pstate_change_high(dc); } @@ -2694,7 +2673,6 @@ static void dcn10_set_cursor_sdr_white_level(struct pipe_ctx *pipe_ctx) static const struct hw_sequencer_funcs dcn10_funcs = { .program_gamut_remap = program_gamut_remap, - .program_csc_matrix = program_csc_matrix, .init_hw = dcn10_init_hw, .apply_ctx_to_hw = dce110_apply_ctx_to_hw, .apply_ctx_for_surface = dcn10_apply_ctx_for_surface, @@ -2721,7 +2699,8 @@ static const struct hw_sequencer_funcs dcn10_funcs = { .disable_plane = dcn10_disable_plane, .blank_pixel_data = dcn10_blank_pixel_data, .pipe_control_lock = dcn10_pipe_control_lock, - .set_bandwidth = dcn10_set_bandwidth, + .prepare_bandwidth = dcn10_prepare_bandwidth, + .optimize_bandwidth = dcn10_optimize_bandwidth, .reset_hw_ctx_wrap = reset_hw_ctx_wrap, .enable_stream_timing = dcn10_enable_stream_timing, .set_drr = set_drr, @@ -2732,10 +2711,6 @@ static const struct hw_sequencer_funcs dcn10_funcs = { .log_hw_state = dcn10_log_hw_state, .get_hw_state = dcn10_get_hw_state, .wait_for_mpcc_disconnect = dcn10_wait_for_mpcc_disconnect, - .ready_shared_resources = ready_shared_resources, - .optimize_shared_resources = optimize_shared_resources, - .pplib_apply_display_requirements = - dcn10_pplib_apply_display_requirements, .edp_backlight_control = hwss_edp_backlight_control, .edp_power_control = hwss_edp_power_control, .edp_wait_for_hpd_ready = hwss_edp_wait_for_hpd_ready, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h index 84d461e0ed3e..5e5610c9e600 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.h @@ -51,4 +51,24 @@ void dcn10_get_hw_state( char *pBuf, unsigned int bufSize, unsigned int mask); +bool is_lower_pipe_tree_visible(struct pipe_ctx *pipe_ctx); + +bool is_upper_pipe_tree_visible(struct pipe_ctx *pipe_ctx); + +bool is_pipe_tree_visible(struct pipe_ctx *pipe_ctx); + +void dcn10_program_pte_vm(struct dce_hwseq *hws, struct hubp *hubp); + +void set_hdr_multiplier(struct pipe_ctx *pipe_ctx); + +void update_dchubp_dpp( + struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_state *context); + +struct pipe_ctx *find_top_pipe_for_stream( + struct dc *dc, + struct dc_state *context, + const struct dc_stream_state *stream); + #endif /* __DC_HWSS_DCN10_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.c index ba6a8686062f..477ab9222216 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.c @@ -589,7 +589,7 @@ static bool dcn10_link_encoder_validate_hdmi_output( return false; /* DCE11 HW does not support 420 */ - if (!enc10->base.features.ycbcr420_supported && + if (!enc10->base.features.hdmi_ycbcr420_supported && crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420) return false; @@ -606,8 +606,10 @@ bool dcn10_link_encoder_validate_dp_output( const struct dcn10_link_encoder *enc10, const struct dc_crtc_timing *crtc_timing) { - if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420) - return false; + if (crtc_timing->pixel_encoding == PIXEL_ENCODING_YCBCR420) { + if (!enc10->base.features.dp_ycbcr420_supported) + return false; + } return true; } diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c index 54626682bab2..7d1f66797cb3 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c @@ -87,9 +87,8 @@ static void optc1_disable_stereo(struct timing_generator *optc) REG_SET(OTG_STEREO_CONTROL, 0, OTG_STEREO_EN, 0); - REG_SET_3(OTG_3D_STRUCTURE_CONTROL, 0, + REG_SET_2(OTG_3D_STRUCTURE_CONTROL, 0, OTG_3D_STRUCTURE_EN, 0, - OTG_3D_STRUCTURE_V_UPDATE_MODE, 0, OTG_3D_STRUCTURE_STEREO_SEL_OVR, 0); } @@ -274,10 +273,12 @@ void optc1_program_timing( * program the reg for interrupt postition. */ vertical_line_start = asic_blank_end - optc->dlg_otg_param.vstartup_start + 1; - if (vertical_line_start < 0) { - ASSERT(0); + v_fp2 = 0; + if (vertical_line_start < 0) + v_fp2 = -vertical_line_start; + if (vertical_line_start < 0) vertical_line_start = 0; - } + REG_SET(OTG_VERTICAL_INTERRUPT2_POSITION, 0, OTG_VERTICAL_INTERRUPT2_LINE_START, vertical_line_start); @@ -296,9 +297,6 @@ void optc1_program_timing( if (patched_crtc_timing.flags.INTERLACE == 1) field_num = 1; } - v_fp2 = 0; - if (optc->dlg_otg_param.vstartup_start > asic_blank_end) - v_fp2 = optc->dlg_otg_param.vstartup_start > asic_blank_end; /* Interlace */ if (patched_crtc_timing.flags.INTERLACE == 1) { @@ -1155,9 +1153,8 @@ static void optc1_enable_stereo(struct timing_generator *optc, OTG_DISABLE_STEREOSYNC_OUTPUT_FOR_DP, 1); if (flags->PROGRAM_STEREO) - REG_UPDATE_3(OTG_3D_STRUCTURE_CONTROL, + REG_UPDATE_2(OTG_3D_STRUCTURE_CONTROL, OTG_3D_STRUCTURE_EN, flags->FRAME_PACKED, - OTG_3D_STRUCTURE_V_UPDATE_MODE, flags->FRAME_PACKED, OTG_3D_STRUCTURE_STEREO_SEL_OVR, flags->FRAME_PACKED); } diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c index a71453a15ae3..47dbe4bb294a 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c @@ -28,23 +28,23 @@ #include "resource.h" #include "include/irq_service_interface.h" -#include "dcn10/dcn10_resource.h" +#include "dcn10_resource.h" -#include "dcn10/dcn10_ipp.h" -#include "dcn10/dcn10_mpc.h" +#include "dcn10_ipp.h" +#include "dcn10_mpc.h" #include "irq/dcn10/irq_service_dcn10.h" -#include "dcn10/dcn10_dpp.h" +#include "dcn10_dpp.h" #include "dcn10_optc.h" -#include "dcn10/dcn10_hw_sequencer.h" +#include "dcn10_hw_sequencer.h" #include "dce110/dce110_hw_sequencer.h" -#include "dcn10/dcn10_opp.h" -#include "dcn10/dcn10_link_encoder.h" -#include "dcn10/dcn10_stream_encoder.h" -#include "dce/dce_clocks.h" +#include "dcn10_opp.h" +#include "dcn10_link_encoder.h" +#include "dcn10_stream_encoder.h" +#include "dcn10_clk_mgr.h" #include "dce/dce_clock_source.h" #include "dce/dce_audio.h" #include "dce/dce_hwseq.h" -#include "../virtual/virtual_stream_encoder.h" +#include "virtual/virtual_stream_encoder.h" #include "dce110/dce110_resource.h" #include "dce112/dce112_resource.h" #include "dcn10_hubp.h" @@ -438,6 +438,7 @@ static const struct dcn_optc_mask tg_mask = { static const struct bios_registers bios_regs = { + NBIO_SR(BIOS_SCRATCH_0), NBIO_SR(BIOS_SCRATCH_3), NBIO_SR(BIOS_SCRATCH_6) }; @@ -719,7 +720,8 @@ static struct timing_generator *dcn10_timing_generator_create( static const struct encoder_feature_support link_enc_feature = { .max_hdmi_deep_color = COLOR_DEPTH_121212, .max_hdmi_pixel_clock = 600000, - .ycbcr420_supported = true, + .hdmi_ycbcr420_supported = true, + .dp_ycbcr420_supported = false, .flags.bits.IS_HBR2_CAPABLE = true, .flags.bits.IS_HBR3_CAPABLE = true, .flags.bits.IS_TPS3_CAPABLE = true, @@ -949,8 +951,8 @@ static void destruct(struct dcn10_resource_pool *pool) if (pool->base.dmcu != NULL) dce_dmcu_destroy(&pool->base.dmcu); - if (pool->base.dccg != NULL) - dce_dccg_destroy(&pool->base.dccg); + if (pool->base.clk_mgr != NULL) + dce_clk_mgr_destroy(&pool->base.clk_mgr); kfree(pool->base.pp_smu); } @@ -1276,8 +1278,8 @@ static bool construct( } } - pool->base.dccg = dcn1_dccg_create(ctx); - if (pool->base.dccg == NULL) { + pool->base.clk_mgr = dcn1_clk_mgr_create(ctx); + if (pool->base.clk_mgr == NULL) { dm_error("DC: failed to create display clock!\n"); BREAK_TO_DEBUGGER(); goto fail; diff --git a/drivers/gpu/drm/amd/display/dc/dm_pp_smu.h b/drivers/gpu/drm/amd/display/dc/dm_pp_smu.h index f2ea8452d48f..beb08fd12b1d 100644 --- a/drivers/gpu/drm/amd/display/dc/dm_pp_smu.h +++ b/drivers/gpu/drm/amd/display/dc/dm_pp_smu.h @@ -55,10 +55,10 @@ struct pp_smu { struct pp_smu_wm_set_range { unsigned int wm_inst; - uint32_t min_fill_clk_khz; - uint32_t max_fill_clk_khz; - uint32_t min_drain_clk_khz; - uint32_t max_drain_clk_khz; + uint32_t min_fill_clk_mhz; + uint32_t max_fill_clk_mhz; + uint32_t min_drain_clk_mhz; + uint32_t max_drain_clk_mhz; }; #define MAX_WATERMARK_SETS 4 @@ -77,15 +77,15 @@ struct pp_smu_display_requirement_rv { */ unsigned int display_count; - /* PPSMC_MSG_SetHardMinFclkByFreq: khz + /* PPSMC_MSG_SetHardMinFclkByFreq: mhz * FCLK will vary with DPM, but never below requested hard min */ - unsigned int hard_min_fclk_khz; + unsigned int hard_min_fclk_mhz; - /* PPSMC_MSG_SetHardMinDcefclkByFreq: khz + /* PPSMC_MSG_SetHardMinDcefclkByFreq: mhz * fixed clock at requested freq, either from FCH bypass or DFS */ - unsigned int hard_min_dcefclk_khz; + unsigned int hard_min_dcefclk_mhz; /* PPSMC_MSG_SetMinDeepSleepDcefclk: mhz * when DF is in cstate, dcf clock is further divided down @@ -103,13 +103,19 @@ struct pp_smu_funcs_rv { void (*set_display_count)(struct pp_smu *pp, int count); /* which SMU message? are reader and writer WM separate SMU msg? */ + /* + * PPSMC_MSG_SetDriverDramAddrHigh + * PPSMC_MSG_SetDriverDramAddrLow + * PPSMC_MSG_TransferTableDram2Smu + * + * */ void (*set_wm_ranges)(struct pp_smu *pp, struct pp_smu_wm_range_sets *ranges); /* PPSMC_MSG_SetHardMinDcfclkByFreq * fixed clock at requested freq, either from FCH bypass or DFS */ - void (*set_hard_min_dcfclk_by_freq)(struct pp_smu *pp, int khz); + void (*set_hard_min_dcfclk_by_freq)(struct pp_smu *pp, int mhz); /* PPSMC_MSG_SetMinDeepSleepDcfclk * when DF is in cstate, dcf clock is further divided down @@ -120,12 +126,12 @@ struct pp_smu_funcs_rv { /* PPSMC_MSG_SetHardMinFclkByFreq * FCLK will vary with DPM, but never below requested hard min */ - void (*set_hard_min_fclk_by_freq)(struct pp_smu *pp, int khz); + void (*set_hard_min_fclk_by_freq)(struct pp_smu *pp, int mhz); /* PPSMC_MSG_SetHardMinSocclkByFreq * Needed for DWB support */ - void (*set_hard_min_socclk_by_freq)(struct pp_smu *pp, int khz); + void (*set_hard_min_socclk_by_freq)(struct pp_smu *pp, int mhz); /* PME w/a */ void (*set_pme_wa_enable)(struct pp_smu *pp); diff --git a/drivers/gpu/drm/amd/display/dc/dm_services_types.h b/drivers/gpu/drm/amd/display/dc/dm_services_types.h index 2b83f922ac02..1af8c777b3ac 100644 --- a/drivers/gpu/drm/amd/display/dc/dm_services_types.h +++ b/drivers/gpu/drm/amd/display/dc/dm_services_types.h @@ -208,22 +208,20 @@ struct dm_bl_data_point { /* Brightness level as effective value in range 0-255, * corresponding to above percentage */ - uint8_t signalLevel; + uint8_t signal_level; }; /* Total size of the structure should not exceed 256 bytes */ struct dm_acpi_atif_backlight_caps { - - uint16_t size; /* Bytes 0-1 (2 bytes) */ uint16_t flags; /* Byted 2-3 (2 bytes) */ - uint8_t errorCode; /* Byte 4 */ - uint8_t acLevelPercentage; /* Byte 5 */ - uint8_t dcLevelPercentage; /* Byte 6 */ - uint8_t minInputSignal; /* Byte 7 */ - uint8_t maxInputSignal; /* Byte 8 */ - uint8_t numOfDataPoints; /* Byte 9 */ - struct dm_bl_data_point dataPoints[99]; /* Bytes 10-207 (198 bytes)*/ + uint8_t error_code; /* Byte 4 */ + uint8_t ac_level_percentage; /* Byte 5 */ + uint8_t dc_level_percentage; /* Byte 6 */ + uint8_t min_input_signal; /* Byte 7 */ + uint8_t max_input_signal; /* Byte 8 */ + uint8_t num_data_points; /* Byte 9 */ + struct dm_bl_data_point data_points[99]; /* Bytes 10-207 (198 bytes)*/ }; enum dm_acpi_display_type { diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h index cbafce649e33..5dd04520ceca 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h @@ -113,7 +113,8 @@ struct _vcs_dpi_soc_bounding_box_st { int use_urgent_burst_bw; double max_hscl_ratio; double max_vscl_ratio; - struct _vcs_dpi_voltage_scaling_st clock_limits[7]; + unsigned int num_states; + struct _vcs_dpi_voltage_scaling_st clock_limits[8]; }; struct _vcs_dpi_ip_params_st { diff --git a/drivers/gpu/drm/amd/display/dc/inc/bw_fixed.h b/drivers/gpu/drm/amd/display/dc/inc/bw_fixed.h index 39ee8eba3c31..d1656c9d50df 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/bw_fixed.h +++ b/drivers/gpu/drm/amd/display/dc/inc/bw_fixed.h @@ -126,7 +126,7 @@ static inline struct bw_fixed bw_div(const struct bw_fixed arg1, const struct bw static inline struct bw_fixed bw_mod(const struct bw_fixed arg1, const struct bw_fixed arg2) { struct bw_fixed res; - div64_u64_rem(arg1.value, arg2.value, &res.value); + div64_u64_rem(arg1.value, arg2.value, (uint64_t *)&res.value); return res; } diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_types.h b/drivers/gpu/drm/amd/display/dc/inc/core_types.h index c1976c175b57..e3ee96afa60e 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h @@ -82,7 +82,7 @@ void core_link_disable_stream(struct pipe_ctx *pipe_ctx, int option); void core_link_set_avmute(struct pipe_ctx *pipe_ctx, bool enable); /********** DAL Core*********************/ -#include "display_clock.h" +#include "hw/clk_mgr.h" #include "transform.h" #include "dpp.h" @@ -169,6 +169,7 @@ struct resource_pool { unsigned int audio_count; struct audio_support audio_support; + struct clk_mgr *clk_mgr; struct dccg *dccg; struct irq_service *irqs; @@ -287,7 +288,7 @@ struct dc_state { struct dcn_bw_internal_vars dcn_bw_vars; #endif - struct dccg *dis_clk; + struct clk_mgr *dccg; struct kref refcount; }; diff --git a/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h b/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h index e688eb9b975c..ece954a40a8e 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h +++ b/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h @@ -31,8 +31,8 @@ #define __DCN_CALCS_H__ #include "bw_fixed.h" -#include "display_clock.h" #include "../dml/display_mode_lib.h" +#include "hw/clk_mgr.h" struct dc; struct dc_state; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/abm.h b/drivers/gpu/drm/amd/display/dc/inc/hw/abm.h index a83a48494613..abc961c0906e 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/abm.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/abm.h @@ -47,12 +47,18 @@ struct abm_funcs { bool (*set_abm_level)(struct abm *abm, unsigned int abm_level); bool (*set_abm_immediate_disable)(struct abm *abm); bool (*init_backlight)(struct abm *abm); - bool (*set_backlight_level)(struct abm *abm, - unsigned int backlight_level, + + /* backlight_pwm_u16_16 is unsigned 32 bit, + * 16 bit integer + 16 fractional, where 1.0 is max backlight value. + */ + bool (*set_backlight_level_pwm)(struct abm *abm, + unsigned int backlight_pwm_u16_16, unsigned int frame_ramp, unsigned int controller_id, bool use_smooth_brightness); - unsigned int (*get_current_backlight_8_bit)(struct abm *abm); + + unsigned int (*get_current_backlight)(struct abm *abm); + unsigned int (*get_target_backlight)(struct abm *abm); }; #endif diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/display_clock.h b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h index 689faa16c0ae..23a4b18e5fee 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/display_clock.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h @@ -23,41 +23,25 @@ * */ -#ifndef __DISPLAY_CLOCK_H__ -#define __DISPLAY_CLOCK_H__ +#ifndef __DAL_CLK_MGR_H__ +#define __DAL_CLK_MGR_H__ #include "dm_services_types.h" #include "dc.h" -/* Structure containing all state-dependent clocks - * (dependent on "enum clocks_state") */ -struct state_dependent_clocks { - int display_clk_khz; - int pixel_clk_khz; -}; - -struct dccg { +struct clk_mgr { struct dc_context *ctx; - const struct display_clock_funcs *funcs; + const struct clk_mgr_funcs *funcs; - enum dm_pp_clocks_state max_clks_state; - enum dm_pp_clocks_state cur_min_clks_state; struct dc_clocks clks; }; -struct display_clock_funcs { - void (*update_clocks)(struct dccg *dccg, - struct dc_clocks *new_clocks, +struct clk_mgr_funcs { + void (*update_clocks)(struct clk_mgr *clk_mgr, + struct dc_state *context, bool safe_to_lower); - int (*set_dispclk)(struct dccg *dccg, - int requested_clock_khz); - - int (*get_dp_ref_clk_frequency)(struct dccg *dccg); - bool (*update_dfs_bypass)(struct dccg *dccg, - struct dc *dc, - struct dc_state *context, - int requested_clock_khz); + int (*get_dp_ref_clk_frequency)(struct clk_mgr *clk_mgr); }; -#endif /* __DISPLAY_CLOCK_H__ */ +#endif /* __DAL_CLK_MGR_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h new file mode 100644 index 000000000000..95a56d012626 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h @@ -0,0 +1,44 @@ +/* + * Copyright 2018 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. + * + * Authors: AMD + * + */ + +#ifndef __DAL_DCCG_H__ +#define __DAL_DCCG_H__ + +#include "dc_types.h" + +struct dccg { + struct dc_context *ctx; + const struct dccg_funcs *funcs; + + int ref_dppclk; +}; + +struct dccg_funcs { + void (*update_dpp_dto)(struct dccg *dccg, + int dpp_inst, + int req_dppclk); +}; + +#endif //__DAL_DCCG_H__ diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h b/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h index cf7433ebf91a..da85537a4488 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h @@ -53,6 +53,12 @@ struct curve_points { uint32_t custom_float_slope; }; +struct curve_points3 { + struct curve_points red; + struct curve_points green; + struct curve_points blue; +}; + struct pwl_result_data { struct fixed31_32 red; struct fixed31_32 green; @@ -71,9 +77,17 @@ struct pwl_result_data { uint32_t delta_blue_reg; }; +/* arr_curve_points - regamma regions/segments specification + * arr_points - beginning and end point specified separately (only one on DCE) + * corner_points - beginning and end point for all 3 colors (DCN) + * rgb_resulted - final curve + */ struct pwl_params { struct gamma_curve arr_curve_points[34]; - struct curve_points arr_points[2]; + union { + struct curve_points arr_points[2]; + struct curve_points3 corner_points[2]; + }; struct pwl_result_data rgb_resulted[256 + 3]; uint32_t hw_points_num; }; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h b/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h index e28e9770e0a3..c20fdcaac53b 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h @@ -65,7 +65,8 @@ struct encoder_feature_support { enum dc_color_depth max_hdmi_deep_color; unsigned int max_hdmi_pixel_clock; - bool ycbcr420_supported; + bool hdmi_ycbcr420_supported; + bool dp_ycbcr420_supported; }; union dpcd_psr_configuration { diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h b/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h index da89c2edb07c..06df02ddff6a 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h @@ -31,7 +31,7 @@ #include "dml/display_mode_structs.h" struct dchub_init_data; -struct cstate_pstate_watermarks_st { +struct cstate_pstate_watermarks_st1 { uint32_t cstate_exit_ns; uint32_t cstate_enter_plus_exit_ns; uint32_t pstate_change_ns; @@ -40,7 +40,7 @@ struct cstate_pstate_watermarks_st { struct dcn_watermarks { uint32_t pte_meta_urgent_ns; uint32_t urgent_ns; - struct cstate_pstate_watermarks_st cstate_pstate; + struct cstate_pstate_watermarks_st1 cstate_pstate; }; struct dcn_watermark_set { diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h index 26f29d5da3d8..e9b702ce02dd 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h @@ -32,8 +32,6 @@ #include "inc/hw/link_encoder.h" #include "core_status.h" -#define EDP_BACKLIGHT_RAMP_DISABLE_LEVEL 0xFFFFFFFF - enum pipe_gating_control { PIPE_GATING_CONTROL_DISABLE = 0, PIPE_GATING_CONTROL_ENABLE, @@ -87,11 +85,6 @@ struct hw_sequencer_funcs { void (*program_gamut_remap)( struct pipe_ctx *pipe_ctx); - void (*program_csc_matrix)( - struct pipe_ctx *pipe_ctx, - enum dc_color_space colorspace, - uint16_t *matrix); - void (*program_output_csc)(struct dc *dc, struct pipe_ctx *pipe_ctx, enum dc_color_space colorspace, @@ -177,10 +170,12 @@ struct hw_sequencer_funcs { struct pipe_ctx *pipe_ctx, bool blank); - void (*set_bandwidth)( + void (*prepare_bandwidth)( struct dc *dc, - struct dc_state *context, - bool safe_to_lower); + struct dc_state *context); + void (*optimize_bandwidth)( + struct dc *dc, + struct dc_state *context); void (*set_drr)(struct pipe_ctx **pipe_ctx, int num_pipes, int vmin, int vmax); @@ -210,11 +205,6 @@ struct hw_sequencer_funcs { struct resource_pool *res_pool, struct pipe_ctx *pipe_ctx); - void (*ready_shared_resources)(struct dc *dc, struct dc_state *context); - void (*optimize_shared_resources)(struct dc *dc); - void (*pplib_apply_display_requirements)( - struct dc *dc, - struct dc_state *context); void (*edp_power_control)( struct dc_link *link, bool enable); diff --git a/drivers/gpu/drm/amd/display/dc/inc/resource.h b/drivers/gpu/drm/amd/display/dc/inc/resource.h index 33b99e3ab10d..0086a2f1d21a 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/resource.h +++ b/drivers/gpu/drm/amd/display/dc/inc/resource.h @@ -30,9 +30,6 @@ #include "dal_asic_id.h" #include "dm_pp_smu.h" -/* TODO unhardcode, 4 for CZ*/ -#define MEMORY_TYPE_MULTIPLIER 4 - enum dce_version resource_parse_asic_id( struct hw_asic_id asic_id); diff --git a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c index cdcefd087487..7480f072c375 100644 --- a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c +++ b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c @@ -306,6 +306,18 @@ static struct fixed31_32 translate_from_linear_space( a1); } +static struct fixed31_32 calculate_gamma22(struct fixed31_32 arg) +{ + struct fixed31_32 gamma = dc_fixpt_from_fraction(22, 10); + + return translate_from_linear_space(arg, + dc_fixpt_zero, + dc_fixpt_zero, + dc_fixpt_zero, + dc_fixpt_zero, + gamma); +} + static struct fixed31_32 translate_to_linear_space( struct fixed31_32 arg, struct fixed31_32 a0, @@ -709,6 +721,169 @@ static void build_regamma(struct pwl_float_data_ex *rgb_regamma, } } +static void hermite_spline_eetf(struct fixed31_32 input_x, + struct fixed31_32 max_display, + struct fixed31_32 min_display, + struct fixed31_32 max_content, + struct fixed31_32 *out_x) +{ + struct fixed31_32 min_lum_pq; + struct fixed31_32 max_lum_pq; + struct fixed31_32 max_content_pq; + struct fixed31_32 ks; + struct fixed31_32 E1; + struct fixed31_32 E2; + struct fixed31_32 E3; + struct fixed31_32 t; + struct fixed31_32 t2; + struct fixed31_32 t3; + struct fixed31_32 two; + struct fixed31_32 three; + struct fixed31_32 temp1; + struct fixed31_32 temp2; + struct fixed31_32 a = dc_fixpt_from_fraction(15, 10); + struct fixed31_32 b = dc_fixpt_from_fraction(5, 10); + struct fixed31_32 epsilon = dc_fixpt_from_fraction(1, 1000000); // dc_fixpt_epsilon is a bit too small + + if (dc_fixpt_eq(max_content, dc_fixpt_zero)) { + *out_x = dc_fixpt_zero; + return; + } + + compute_pq(input_x, &E1); + compute_pq(dc_fixpt_div(min_display, max_content), &min_lum_pq); + compute_pq(dc_fixpt_div(max_display, max_content), &max_lum_pq); + compute_pq(dc_fixpt_one, &max_content_pq); // always 1? DAL2 code is weird + a = dc_fixpt_div(dc_fixpt_add(dc_fixpt_one, b), max_content_pq); // (1+b)/maxContent + ks = dc_fixpt_sub(dc_fixpt_mul(a, max_lum_pq), b); // a * max_lum_pq - b + + if (dc_fixpt_lt(E1, ks)) + E2 = E1; + else if (dc_fixpt_le(ks, E1) && dc_fixpt_le(E1, dc_fixpt_one)) { + if (dc_fixpt_lt(epsilon, dc_fixpt_sub(dc_fixpt_one, ks))) + // t = (E1 - ks) / (1 - ks) + t = dc_fixpt_div(dc_fixpt_sub(E1, ks), + dc_fixpt_sub(dc_fixpt_one, ks)); + else + t = dc_fixpt_zero; + + two = dc_fixpt_from_int(2); + three = dc_fixpt_from_int(3); + + t2 = dc_fixpt_mul(t, t); + t3 = dc_fixpt_mul(t2, t); + temp1 = dc_fixpt_mul(two, t3); + temp2 = dc_fixpt_mul(three, t2); + + // (2t^3 - 3t^2 + 1) * ks + E2 = dc_fixpt_mul(ks, dc_fixpt_add(dc_fixpt_one, + dc_fixpt_sub(temp1, temp2))); + + // (-2t^3 + 3t^2) * max_lum_pq + E2 = dc_fixpt_add(E2, dc_fixpt_mul(max_lum_pq, + dc_fixpt_sub(temp2, temp1))); + + temp1 = dc_fixpt_mul(two, t2); + temp2 = dc_fixpt_sub(dc_fixpt_one, ks); + + // (t^3 - 2t^2 + t) * (1-ks) + E2 = dc_fixpt_add(E2, dc_fixpt_mul(temp2, + dc_fixpt_add(t, dc_fixpt_sub(t3, temp1)))); + } else + E2 = dc_fixpt_one; + + temp1 = dc_fixpt_sub(dc_fixpt_one, E2); + temp2 = dc_fixpt_mul(temp1, temp1); + temp2 = dc_fixpt_mul(temp2, temp2); + // temp2 = (1-E2)^4 + + E3 = dc_fixpt_add(E2, dc_fixpt_mul(min_lum_pq, temp2)); + compute_de_pq(E3, out_x); + + *out_x = dc_fixpt_div(*out_x, dc_fixpt_div(max_display, max_content)); +} + +static bool build_freesync_hdr(struct pwl_float_data_ex *rgb_regamma, + uint32_t hw_points_num, + const struct hw_x_point *coordinate_x, + const struct freesync_hdr_tf_params *fs_params) +{ + uint32_t i; + struct pwl_float_data_ex *rgb = rgb_regamma; + const struct hw_x_point *coord_x = coordinate_x; + struct fixed31_32 scaledX = dc_fixpt_zero; + struct fixed31_32 scaledX1 = dc_fixpt_zero; + struct fixed31_32 max_display = dc_fixpt_from_int(fs_params->max_display); + struct fixed31_32 min_display = dc_fixpt_from_fraction(fs_params->min_display, 10000); + struct fixed31_32 max_content = dc_fixpt_from_int(fs_params->max_content); + struct fixed31_32 min_content = dc_fixpt_from_fraction(fs_params->min_content, 10000); + struct fixed31_32 clip = dc_fixpt_one; + struct fixed31_32 output; + bool use_eetf = false; + bool is_clipped = false; + struct fixed31_32 sdr_white_level = dc_fixpt_from_int(fs_params->sdr_white_level); + + if (fs_params == NULL || fs_params->max_content == 0 || + fs_params->max_display == 0) + return false; + + if (fs_params->min_display > 1000) // cap at 0.1 at the bottom + min_display = dc_fixpt_from_fraction(1, 10); + if (fs_params->max_display < 100) // cap at 100 at the top + max_display = dc_fixpt_from_int(100); + + if (fs_params->min_content < fs_params->min_display) + use_eetf = true; + else + min_content = min_display; + + if (fs_params->max_content > fs_params->max_display) + use_eetf = true; + else + max_content = max_display; + + rgb += 32; // first 32 points have problems with fixed point, too small + coord_x += 32; + for (i = 32; i <= hw_points_num; i++) { + if (!is_clipped) { + if (use_eetf) { + /*max content is equal 1 */ + scaledX1 = dc_fixpt_div(coord_x->x, + dc_fixpt_div(max_content, sdr_white_level)); + hermite_spline_eetf(scaledX1, max_display, min_display, + max_content, &scaledX); + } else + scaledX = dc_fixpt_div(coord_x->x, + dc_fixpt_div(max_display, sdr_white_level)); + + if (dc_fixpt_lt(scaledX, clip)) { + if (dc_fixpt_lt(scaledX, dc_fixpt_zero)) + output = dc_fixpt_zero; + else + output = calculate_gamma22(scaledX); + + rgb->r = output; + rgb->g = output; + rgb->b = output; + } else { + is_clipped = true; + rgb->r = clip; + rgb->g = clip; + rgb->b = clip; + } + } else { + rgb->r = clip; + rgb->g = clip; + rgb->b = clip; + } + + ++coord_x; + ++rgb; + } + + return true; +} + static void build_degamma(struct pwl_float_data_ex *curve, uint32_t hw_points_num, const struct hw_x_point *coordinate_x, bool is_2_4) @@ -1356,7 +1531,8 @@ static bool map_regamma_hw_to_x_user( #define _EXTRA_POINTS 3 bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf, - const struct dc_gamma *ramp, bool mapUserRamp, bool canRomBeUsed) + const struct dc_gamma *ramp, bool mapUserRamp, bool canRomBeUsed, + const struct freesync_hdr_tf_params *fs_params) { struct dc_transfer_func_distributed_points *tf_pts = &output_tf->tf_pts; struct dividers dividers; @@ -1374,7 +1550,7 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf, /* we can use hardcoded curve for plain SRGB TF */ if (output_tf->type == TF_TYPE_PREDEFINED && canRomBeUsed == true && output_tf->tf == TRANSFER_FUNCTION_SRGB && - (!mapUserRamp && ramp->type == GAMMA_RGB_256)) + (ramp->is_identity || (!mapUserRamp && ramp->type == GAMMA_RGB_256))) return true; output_tf->type = TF_TYPE_DISTRIBUTED_POINTS; @@ -1424,6 +1600,12 @@ bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf, MAX_HW_POINTS, coordinates_x, output_tf->sdr_ref_white_level); + } else if (tf == TRANSFER_FUNCTION_GAMMA22 && + fs_params != NULL) { + build_freesync_hdr(rgb_regamma, + MAX_HW_POINTS, + coordinates_x, + fs_params); } else { tf_pts->end_exponent = 0; tf_pts->x_point_at_y1_red = 1; diff --git a/drivers/gpu/drm/amd/display/modules/color/color_gamma.h b/drivers/gpu/drm/amd/display/modules/color/color_gamma.h index 63ccb9c91224..a6e164df090a 100644 --- a/drivers/gpu/drm/amd/display/modules/color/color_gamma.h +++ b/drivers/gpu/drm/amd/display/modules/color/color_gamma.h @@ -73,12 +73,21 @@ struct regamma_lut { }; }; +struct freesync_hdr_tf_params { + unsigned int sdr_white_level; + unsigned int min_content; // luminance in 1/10000 nits + unsigned int max_content; // luminance in nits + unsigned int min_display; // luminance in 1/10000 nits + unsigned int max_display; // luminance in nits +}; + void setup_x_points_distribution(void); void precompute_pq(void); void precompute_de_pq(void); bool mod_color_calculate_regamma_params(struct dc_transfer_func *output_tf, - const struct dc_gamma *ramp, bool mapUserRamp, bool canRomBeUsed); + const struct dc_gamma *ramp, bool mapUserRamp, bool canRomBeUsed, + const struct freesync_hdr_tf_params *fs_params); bool mod_color_calculate_degamma_params(struct dc_transfer_func *output_tf, const struct dc_gamma *ramp, bool mapUserRamp); diff --git a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c index 4018c7180d00..620a171620ee 100644 --- a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c +++ b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c @@ -37,6 +37,8 @@ #define RENDER_TIMES_MAX_COUNT 10 /* Threshold to exit BTR (to avoid frequent enter-exits at the lower limit) */ #define BTR_EXIT_MARGIN 2000 +/*Threshold to exit fixed refresh rate*/ +#define FIXED_REFRESH_EXIT_MARGIN_IN_HZ 4 /* Number of consecutive frames to check before entering/exiting fixed refresh*/ #define FIXED_REFRESH_ENTER_FRAME_COUNT 5 #define FIXED_REFRESH_EXIT_FRAME_COUNT 5 @@ -257,40 +259,14 @@ static void apply_below_the_range(struct core_freesync *core_freesync, if (in_out_vrr->btr.btr_active) { in_out_vrr->btr.frame_counter = 0; in_out_vrr->btr.btr_active = false; - - /* Exit Fixed Refresh mode */ - } else if (in_out_vrr->fixed.fixed_active) { - - in_out_vrr->fixed.frame_counter++; - - if (in_out_vrr->fixed.frame_counter > - FIXED_REFRESH_EXIT_FRAME_COUNT) { - in_out_vrr->fixed.frame_counter = 0; - in_out_vrr->fixed.fixed_active = false; - } } } else if (last_render_time_in_us > max_render_time_in_us) { /* Enter Below the Range */ - if (!in_out_vrr->btr.btr_active && - in_out_vrr->btr.btr_enabled) { - in_out_vrr->btr.btr_active = true; - - /* Enter Fixed Refresh mode */ - } else if (!in_out_vrr->fixed.fixed_active && - !in_out_vrr->btr.btr_enabled) { - in_out_vrr->fixed.frame_counter++; - - if (in_out_vrr->fixed.frame_counter > - FIXED_REFRESH_ENTER_FRAME_COUNT) { - in_out_vrr->fixed.frame_counter = 0; - in_out_vrr->fixed.fixed_active = true; - } - } + in_out_vrr->btr.btr_active = true; } /* BTR set to "not active" so disengage */ if (!in_out_vrr->btr.btr_active) { - in_out_vrr->btr.btr_active = false; in_out_vrr->btr.inserted_duration_in_us = 0; in_out_vrr->btr.frames_to_insert = 0; in_out_vrr->btr.frame_counter = 0; @@ -375,7 +351,12 @@ static void apply_fixed_refresh(struct core_freesync *core_freesync, bool update = false; unsigned int max_render_time_in_us = in_out_vrr->max_duration_in_us; - if (last_render_time_in_us + BTR_EXIT_MARGIN < max_render_time_in_us) { + //Compute the exit refresh rate and exit frame duration + unsigned int exit_refresh_rate_in_milli_hz = ((1000000000/max_render_time_in_us) + + (1000*FIXED_REFRESH_EXIT_MARGIN_IN_HZ)); + unsigned int exit_frame_duration_in_us = 1000000000/exit_refresh_rate_in_milli_hz; + + if (last_render_time_in_us < exit_frame_duration_in_us) { /* Exit Fixed Refresh mode */ if (in_out_vrr->fixed.fixed_active) { in_out_vrr->fixed.frame_counter++; diff --git a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h index 64ecffd52126..58ac0b90c310 100644 --- a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h +++ b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h @@ -205,20 +205,6 @@ struct tile_config { /** * struct kfd2kgd_calls * - * @init_gtt_mem_allocation: Allocate a buffer on the gart aperture. - * The buffer can be used for mqds, hpds, kernel queue, fence and runlists - * - * @free_gtt_mem: Frees a buffer that was allocated on the gart aperture - * - * @get_local_mem_info: Retrieves information about GPU local memory - * - * @get_gpu_clock_counter: Retrieves GPU clock counter - * - * @get_max_engine_clock_in_mhz: Retrieves maximum GPU clock in MHz - * - * @alloc_pasid: Allocate a PASID - * @free_pasid: Free a PASID - * * @program_sh_mem_settings: A function that should initiate the memory * properties such as main aperture memory type (cache / non cached) and * secondary aperture base address, size and memory type. @@ -255,64 +241,16 @@ struct tile_config { * * @get_tile_config: Returns GPU-specific tiling mode information * - * @get_cu_info: Retrieves activated cu info - * - * @get_vram_usage: Returns current VRAM usage - * - * @create_process_vm: Create a VM address space for a given process and GPU - * - * @destroy_process_vm: Destroy a VM - * - * @get_process_page_dir: Get physical address of a VM page directory - * * @set_vm_context_page_table_base: Program page table base for a VMID * - * @alloc_memory_of_gpu: Allocate GPUVM memory - * - * @free_memory_of_gpu: Free GPUVM memory - * - * @map_memory_to_gpu: Map GPUVM memory into a specific VM address - * space. Allocates and updates page tables and page directories as - * needed. This function may return before all page table updates have - * completed. This allows multiple map operations (on multiple GPUs) - * to happen concurrently. Use sync_memory to synchronize with all - * pending updates. - * - * @unmap_memor_to_gpu: Unmap GPUVM memory from a specific VM address space - * - * @sync_memory: Wait for pending page table updates to complete - * - * @map_gtt_bo_to_kernel: Map a GTT BO for kernel access - * Pins the BO, maps it to kernel address space. Such BOs are never evicted. - * The kernel virtual address remains valid until the BO is freed. - * - * @restore_process_bos: Restore all BOs that belong to the - * process. This is intended for restoring memory mappings after a TTM - * eviction. - * * @invalidate_tlbs: Invalidate TLBs for a specific PASID * * @invalidate_tlbs_vmid: Invalidate TLBs for a specific VMID * - * @submit_ib: Submits an IB to the engine specified by inserting the - * IB to the corresponding ring (ring type). The IB is executed with the - * specified VMID in a user mode context. - * - * @get_vm_fault_info: Return information about a recent VM fault on - * GFXv7 and v8. If multiple VM faults occurred since the last call of - * this function, it will return information about the first of those - * faults. On GFXv9 VM fault information is fully contained in the IH - * packet and this function is not needed. - * * @read_vmid_from_vmfault_reg: On Hawaii the VMID is not set in the * IH ring entry. This function allows the KFD ISR to get the VMID * from the fault status register as early as possible. * - * @gpu_recover: let kgd reset gpu after kfd detect CPC hang - * - * @set_compute_idle: Indicates that compute is idle on a device. This - * can be used to change power profiles depending on compute activity. - * * @get_hive_id: Returns hive id of current device, 0 if xgmi is not enabled * * This structure contains function pointers to services that the kgd driver @@ -320,21 +258,6 @@ struct tile_config { * */ struct kfd2kgd_calls { - int (*init_gtt_mem_allocation)(struct kgd_dev *kgd, size_t size, - void **mem_obj, uint64_t *gpu_addr, - void **cpu_ptr, bool mqd_gfx9); - - void (*free_gtt_mem)(struct kgd_dev *kgd, void *mem_obj); - - void (*get_local_mem_info)(struct kgd_dev *kgd, - struct kfd_local_mem_info *mem_info); - uint64_t (*get_gpu_clock_counter)(struct kgd_dev *kgd); - - uint32_t (*get_max_engine_clock_in_mhz)(struct kgd_dev *kgd); - - int (*alloc_pasid)(unsigned int bits); - void (*free_pasid)(unsigned int pasid); - /* Register access functions */ void (*program_sh_mem_settings)(struct kgd_dev *kgd, uint32_t vmid, uint32_t sh_mem_config, uint32_t sh_mem_ape1_base, @@ -398,49 +321,11 @@ struct kfd2kgd_calls { uint64_t va, uint32_t vmid); int (*get_tile_config)(struct kgd_dev *kgd, struct tile_config *config); - void (*get_cu_info)(struct kgd_dev *kgd, - struct kfd_cu_info *cu_info); - uint64_t (*get_vram_usage)(struct kgd_dev *kgd); - - int (*create_process_vm)(struct kgd_dev *kgd, unsigned int pasid, void **vm, - void **process_info, struct dma_fence **ef); - int (*acquire_process_vm)(struct kgd_dev *kgd, struct file *filp, - unsigned int pasid, void **vm, void **process_info, - struct dma_fence **ef); - void (*destroy_process_vm)(struct kgd_dev *kgd, void *vm); - void (*release_process_vm)(struct kgd_dev *kgd, void *vm); - uint64_t (*get_process_page_dir)(void *vm); void (*set_vm_context_page_table_base)(struct kgd_dev *kgd, uint32_t vmid, uint64_t page_table_base); - int (*alloc_memory_of_gpu)(struct kgd_dev *kgd, uint64_t va, - uint64_t size, void *vm, - struct kgd_mem **mem, uint64_t *offset, - uint32_t flags); - int (*free_memory_of_gpu)(struct kgd_dev *kgd, struct kgd_mem *mem); - int (*map_memory_to_gpu)(struct kgd_dev *kgd, struct kgd_mem *mem, - void *vm); - int (*unmap_memory_to_gpu)(struct kgd_dev *kgd, struct kgd_mem *mem, - void *vm); - int (*sync_memory)(struct kgd_dev *kgd, struct kgd_mem *mem, bool intr); - int (*map_gtt_bo_to_kernel)(struct kgd_dev *kgd, struct kgd_mem *mem, - void **kptr, uint64_t *size); - int (*restore_process_bos)(void *process_info, struct dma_fence **ef); - int (*invalidate_tlbs)(struct kgd_dev *kgd, uint16_t pasid); int (*invalidate_tlbs_vmid)(struct kgd_dev *kgd, uint16_t vmid); - - int (*submit_ib)(struct kgd_dev *kgd, enum kgd_engine_type engine, - uint32_t vmid, uint64_t gpu_addr, - uint32_t *ib_cmd, uint32_t ib_len); - - int (*get_vm_fault_info)(struct kgd_dev *kgd, - struct kfd_vm_fault_info *info); uint32_t (*read_vmid_from_vmfault_reg)(struct kgd_dev *kgd); - - void (*gpu_recover)(struct kgd_dev *kgd); - - void (*set_compute_idle)(struct kgd_dev *kgd, bool idle); - uint64_t (*get_hive_id)(struct kgd_dev *kgd); }; diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_powertune.c b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_powertune.c index 5e19f5977eb1..d138ddae563d 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_powertune.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_powertune.c @@ -967,7 +967,7 @@ int smu7_enable_didt_config(struct pp_hwmgr *hwmgr) PP_CAP(PHM_PlatformCaps_TDRamping) || PP_CAP(PHM_PlatformCaps_TCPRamping)) { - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); mutex_lock(&adev->grbm_idx_mutex); value = 0; value2 = cgs_read_register(hwmgr->device, mmGRBM_GFX_INDEX); @@ -1014,13 +1014,13 @@ int smu7_enable_didt_config(struct pp_hwmgr *hwmgr) "Failed to enable DPM DIDT.", goto error); } mutex_unlock(&adev->grbm_idx_mutex); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } return 0; error: mutex_unlock(&adev->grbm_idx_mutex); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return result; } @@ -1034,7 +1034,7 @@ int smu7_disable_didt_config(struct pp_hwmgr *hwmgr) PP_CAP(PHM_PlatformCaps_TDRamping) || PP_CAP(PHM_PlatformCaps_TCPRamping)) { - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); result = smu7_enable_didt(hwmgr, false); PP_ASSERT_WITH_CODE((result == 0), @@ -1046,12 +1046,12 @@ int smu7_disable_didt_config(struct pp_hwmgr *hwmgr) PP_ASSERT_WITH_CODE((0 == result), "Failed to disable DPM DIDT.", goto error); } - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); } return 0; error: - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return result; } diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c index 2d88abf97e7b..6f26cb241ecc 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega10_powertune.c @@ -937,7 +937,7 @@ static int vega10_enable_cac_driving_se_didt_config(struct pp_hwmgr *hwmgr) num_se = adev->gfx.config.max_shader_engines; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); mutex_lock(&adev->grbm_idx_mutex); for (count = 0; count < num_se; count++) { @@ -962,7 +962,7 @@ static int vega10_enable_cac_driving_se_didt_config(struct pp_hwmgr *hwmgr) vega10_didt_set_mask(hwmgr, true); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return 0; } @@ -971,11 +971,11 @@ static int vega10_disable_cac_driving_se_didt_config(struct pp_hwmgr *hwmgr) { struct amdgpu_device *adev = hwmgr->adev; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); vega10_didt_set_mask(hwmgr, false); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return 0; } @@ -988,7 +988,7 @@ static int vega10_enable_psm_gc_didt_config(struct pp_hwmgr *hwmgr) num_se = adev->gfx.config.max_shader_engines; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); mutex_lock(&adev->grbm_idx_mutex); for (count = 0; count < num_se; count++) { @@ -1007,7 +1007,7 @@ static int vega10_enable_psm_gc_didt_config(struct pp_hwmgr *hwmgr) vega10_didt_set_mask(hwmgr, true); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); vega10_program_gc_didt_config_registers(hwmgr, GCDiDtDroopCtrlConfig_vega10); if (PP_CAP(PHM_PlatformCaps_GCEDC)) @@ -1024,11 +1024,11 @@ static int vega10_disable_psm_gc_didt_config(struct pp_hwmgr *hwmgr) struct amdgpu_device *adev = hwmgr->adev; uint32_t data; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); vega10_didt_set_mask(hwmgr, false); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); if (PP_CAP(PHM_PlatformCaps_GCEDC)) { data = 0x00000000; @@ -1049,7 +1049,7 @@ static int vega10_enable_se_edc_config(struct pp_hwmgr *hwmgr) num_se = adev->gfx.config.max_shader_engines; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); mutex_lock(&adev->grbm_idx_mutex); for (count = 0; count < num_se; count++) { @@ -1070,7 +1070,7 @@ static int vega10_enable_se_edc_config(struct pp_hwmgr *hwmgr) vega10_didt_set_mask(hwmgr, true); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return 0; } @@ -1079,11 +1079,11 @@ static int vega10_disable_se_edc_config(struct pp_hwmgr *hwmgr) { struct amdgpu_device *adev = hwmgr->adev; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); vega10_didt_set_mask(hwmgr, false); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return 0; } @@ -1097,7 +1097,7 @@ static int vega10_enable_psm_gc_edc_config(struct pp_hwmgr *hwmgr) num_se = adev->gfx.config.max_shader_engines; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); vega10_program_gc_didt_config_registers(hwmgr, AvfsPSMResetConfig_vega10); @@ -1118,7 +1118,7 @@ static int vega10_enable_psm_gc_edc_config(struct pp_hwmgr *hwmgr) vega10_didt_set_mask(hwmgr, true); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); vega10_program_gc_didt_config_registers(hwmgr, PSMGCEDCDroopCtrlConfig_vega10); @@ -1138,11 +1138,11 @@ static int vega10_disable_psm_gc_edc_config(struct pp_hwmgr *hwmgr) struct amdgpu_device *adev = hwmgr->adev; uint32_t data; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); vega10_didt_set_mask(hwmgr, false); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); if (PP_CAP(PHM_PlatformCaps_GCEDC)) { data = 0x00000000; @@ -1160,7 +1160,7 @@ static int vega10_enable_se_edc_force_stall_config(struct pp_hwmgr *hwmgr) struct amdgpu_device *adev = hwmgr->adev; int result; - adev->gfx.rlc.funcs->enter_safe_mode(adev); + amdgpu_gfx_rlc_enter_safe_mode(adev); mutex_lock(&adev->grbm_idx_mutex); WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, 0xE0000000); @@ -1173,7 +1173,7 @@ static int vega10_enable_se_edc_force_stall_config(struct pp_hwmgr *hwmgr) vega10_didt_set_mask(hwmgr, false); - adev->gfx.rlc.funcs->exit_safe_mode(adev); + amdgpu_gfx_rlc_exit_safe_mode(adev); return 0; } diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c index 99861f32b1f9..f2daf00cc911 100644 --- a/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +++ b/drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c @@ -2755,7 +2755,7 @@ static int vega20_print_clock_levels(struct pp_hwmgr *hwmgr, for (i = 0; i < clocks.num_levels; i++) size += sprintf(buf + size, "%d: %uMhz %s\n", i, clocks.data[i].clocks_in_khz / 1000, - (clocks.data[i].clocks_in_khz == now) ? "*" : ""); + (clocks.data[i].clocks_in_khz == now * 10) ? "*" : ""); break; case PP_MCLK: @@ -2772,7 +2772,7 @@ static int vega20_print_clock_levels(struct pp_hwmgr *hwmgr, for (i = 0; i < clocks.num_levels; i++) size += sprintf(buf + size, "%d: %uMhz %s\n", i, clocks.data[i].clocks_in_khz / 1000, - (clocks.data[i].clocks_in_khz == now) ? "*" : ""); + (clocks.data[i].clocks_in_khz == now * 10) ? "*" : ""); break; case PP_PCIE: @@ -3454,109 +3454,64 @@ static int vega20_get_thermal_temperature_range(struct pp_hwmgr *hwmgr, static const struct pp_hwmgr_func vega20_hwmgr_funcs = { /* init/fini related */ - .backend_init = - vega20_hwmgr_backend_init, - .backend_fini = - vega20_hwmgr_backend_fini, - .asic_setup = - vega20_setup_asic_task, - .power_off_asic = - vega20_power_off_asic, - .dynamic_state_management_enable = - vega20_enable_dpm_tasks, - .dynamic_state_management_disable = - vega20_disable_dpm_tasks, + .backend_init = vega20_hwmgr_backend_init, + .backend_fini = vega20_hwmgr_backend_fini, + .asic_setup = vega20_setup_asic_task, + .power_off_asic = vega20_power_off_asic, + .dynamic_state_management_enable = vega20_enable_dpm_tasks, + .dynamic_state_management_disable = vega20_disable_dpm_tasks, /* power state related */ - .apply_clocks_adjust_rules = - vega20_apply_clocks_adjust_rules, - .pre_display_config_changed = - vega20_pre_display_configuration_changed_task, - .display_config_changed = - vega20_display_configuration_changed_task, + .apply_clocks_adjust_rules = vega20_apply_clocks_adjust_rules, + .pre_display_config_changed = vega20_pre_display_configuration_changed_task, + .display_config_changed = vega20_display_configuration_changed_task, .check_smc_update_required_for_display_configuration = vega20_check_smc_update_required_for_display_configuration, .notify_smc_display_config_after_ps_adjustment = vega20_notify_smc_display_config_after_ps_adjustment, /* export to DAL */ - .get_sclk = - vega20_dpm_get_sclk, - .get_mclk = - vega20_dpm_get_mclk, - .get_dal_power_level = - vega20_get_dal_power_level, - .get_clock_by_type_with_latency = - vega20_get_clock_by_type_with_latency, - .get_clock_by_type_with_voltage = - vega20_get_clock_by_type_with_voltage, - .set_watermarks_for_clocks_ranges = - vega20_set_watermarks_for_clocks_ranges, - .display_clock_voltage_request = - vega20_display_clock_voltage_request, - .get_performance_level = - vega20_get_performance_level, + .get_sclk = vega20_dpm_get_sclk, + .get_mclk = vega20_dpm_get_mclk, + .get_dal_power_level = vega20_get_dal_power_level, + .get_clock_by_type_with_latency = vega20_get_clock_by_type_with_latency, + .get_clock_by_type_with_voltage = vega20_get_clock_by_type_with_voltage, + .set_watermarks_for_clocks_ranges = vega20_set_watermarks_for_clocks_ranges, + .display_clock_voltage_request = vega20_display_clock_voltage_request, + .get_performance_level = vega20_get_performance_level, /* UMD pstate, profile related */ - .force_dpm_level = - vega20_dpm_force_dpm_level, - .get_power_profile_mode = - vega20_get_power_profile_mode, - .set_power_profile_mode = - vega20_set_power_profile_mode, + .force_dpm_level = vega20_dpm_force_dpm_level, + .get_power_profile_mode = vega20_get_power_profile_mode, + .set_power_profile_mode = vega20_set_power_profile_mode, /* od related */ - .set_power_limit = - vega20_set_power_limit, - .get_sclk_od = - vega20_get_sclk_od, - .set_sclk_od = - vega20_set_sclk_od, - .get_mclk_od = - vega20_get_mclk_od, - .set_mclk_od = - vega20_set_mclk_od, - .odn_edit_dpm_table = - vega20_odn_edit_dpm_table, + .set_power_limit = vega20_set_power_limit, + .get_sclk_od = vega20_get_sclk_od, + .set_sclk_od = vega20_set_sclk_od, + .get_mclk_od = vega20_get_mclk_od, + .set_mclk_od = vega20_set_mclk_od, + .odn_edit_dpm_table = vega20_odn_edit_dpm_table, /* for sysfs to retrive/set gfxclk/memclk */ - .force_clock_level = - vega20_force_clock_level, - .print_clock_levels = - vega20_print_clock_levels, - .read_sensor = - vega20_read_sensor, + .force_clock_level = vega20_force_clock_level, + .print_clock_levels = vega20_print_clock_levels, + .read_sensor = vega20_read_sensor, /* powergate related */ - .powergate_uvd = - vega20_power_gate_uvd, - .powergate_vce = - vega20_power_gate_vce, + .powergate_uvd = vega20_power_gate_uvd, + .powergate_vce = vega20_power_gate_vce, /* thermal related */ - .start_thermal_controller = - vega20_start_thermal_controller, - .stop_thermal_controller = - vega20_thermal_stop_thermal_controller, - .get_thermal_temperature_range = - vega20_get_thermal_temperature_range, - .register_irq_handlers = - smu9_register_irq_handlers, - .disable_smc_firmware_ctf = - vega20_thermal_disable_alert, + .start_thermal_controller = vega20_start_thermal_controller, + .stop_thermal_controller = vega20_thermal_stop_thermal_controller, + .get_thermal_temperature_range = vega20_get_thermal_temperature_range, + .register_irq_handlers = smu9_register_irq_handlers, + .disable_smc_firmware_ctf = vega20_thermal_disable_alert, /* fan control related */ - .get_fan_speed_percent = - vega20_fan_ctrl_get_fan_speed_percent, - .set_fan_speed_percent = - vega20_fan_ctrl_set_fan_speed_percent, - .get_fan_speed_info = - vega20_fan_ctrl_get_fan_speed_info, - .get_fan_speed_rpm = - vega20_fan_ctrl_get_fan_speed_rpm, - .set_fan_speed_rpm = - vega20_fan_ctrl_set_fan_speed_rpm, - .get_fan_control_mode = - vega20_get_fan_control_mode, - .set_fan_control_mode = - vega20_set_fan_control_mode, + .get_fan_speed_percent = vega20_fan_ctrl_get_fan_speed_percent, + .set_fan_speed_percent = vega20_fan_ctrl_set_fan_speed_percent, + .get_fan_speed_info = vega20_fan_ctrl_get_fan_speed_info, + .get_fan_speed_rpm = vega20_fan_ctrl_get_fan_speed_rpm, + .set_fan_speed_rpm = vega20_fan_ctrl_set_fan_speed_rpm, + .get_fan_control_mode = vega20_get_fan_control_mode, + .set_fan_control_mode = vega20_set_fan_control_mode, /* smu memory related */ - .notify_cac_buffer_info = - vega20_notify_cac_buffer_info, - .enable_mgpu_fan_boost = - vega20_enable_mgpu_fan_boost, + .notify_cac_buffer_info = vega20_notify_cac_buffer_info, + .enable_mgpu_fan_boost = vega20_enable_mgpu_fan_boost, }; int vega20_hwmgr_init(struct pp_hwmgr *hwmgr) diff --git a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h index e5a60aa44b5d..07d180ce4d18 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h +++ b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h @@ -28,7 +28,6 @@ #include "hardwaremanager.h" #include "hwmgr_ppt.h" #include "ppatomctrl.h" -#include "hwmgr_ppt.h" #include "power_state.h" #include "smu_helper.h" diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu7_common.h b/drivers/gpu/drm/amd/powerplay/inc/smu7_common.h index 65eb630bfea3..94bf7b649c20 100644 --- a/drivers/gpu/drm/amd/powerplay/inc/smu7_common.h +++ b/drivers/gpu/drm/amd/powerplay/inc/smu7_common.h @@ -40,10 +40,6 @@ #include "bif/bif_5_0_d.h" #include "bif/bif_5_0_sh_mask.h" - -#include "bif/bif_5_0_d.h" -#include "bif/bif_5_0_sh_mask.h" - #include "dce/dce_10_0_d.h" #include "dce/dce_10_0_sh_mask.h" diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/polaris10_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/polaris10_smumgr.c index 872d3824337b..2b2c26616902 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/polaris10_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/polaris10_smumgr.c @@ -44,7 +44,6 @@ #include "smu7_hwmgr.h" #include "hardwaremanager.h" -#include "ppatomctrl.h" #include "atombios.h" #include "pppcielanes.h" diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c index d0eb8ab50148..d111dd4e03d7 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smu10_smumgr.c @@ -29,7 +29,6 @@ #include "rv_ppsmc.h" #include "smu10_driver_if.h" #include "smu10.h" -#include "ppatomctrl.h" #include "pp_debug.h" diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c index 09b844ec3eab..e2787e14a500 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smu8_smumgr.c @@ -24,6 +24,7 @@ #include <linux/delay.h> #include <linux/gfp.h> #include <linux/kernel.h> +#include <linux/ktime.h> #include <linux/slab.h> #include <linux/types.h> @@ -61,9 +62,13 @@ static uint32_t smu8_get_argument(struct pp_hwmgr *hwmgr) mmSMU_MP1_SRBM2P_ARG_0); } -static int smu8_send_msg_to_smc_async(struct pp_hwmgr *hwmgr, uint16_t msg) +/* Send a message to the SMC, and wait for its response.*/ +static int smu8_send_msg_to_smc_with_parameter(struct pp_hwmgr *hwmgr, + uint16_t msg, uint32_t parameter) { int result = 0; + ktime_t t_start; + s64 elapsed_us; if (hwmgr == NULL || hwmgr->device == NULL) return -EINVAL; @@ -74,28 +79,31 @@ static int smu8_send_msg_to_smc_async(struct pp_hwmgr *hwmgr, uint16_t msg) /* Read the last message to SMU, to report actual cause */ uint32_t val = cgs_read_register(hwmgr->device, mmSMU_MP1_SRBM2P_MSG_0); - pr_err("smu8_send_msg_to_smc_async (0x%04x) failed\n", msg); - pr_err("SMU still servicing msg (0x%04x)\n", val); + pr_err("%s(0x%04x) aborted; SMU still servicing msg (0x%04x)\n", + __func__, msg, val); return result; } + t_start = ktime_get(); + + cgs_write_register(hwmgr->device, mmSMU_MP1_SRBM2P_ARG_0, parameter); cgs_write_register(hwmgr->device, mmSMU_MP1_SRBM2P_RESP_0, 0); cgs_write_register(hwmgr->device, mmSMU_MP1_SRBM2P_MSG_0, msg); - return 0; + result = PHM_WAIT_FIELD_UNEQUAL(hwmgr, + SMU_MP1_SRBM2P_RESP_0, CONTENT, 0); + + elapsed_us = ktime_us_delta(ktime_get(), t_start); + + WARN(result, "%s(0x%04x, %#x) timed out after %lld us\n", + __func__, msg, parameter, elapsed_us); + + return result; } -/* Send a message to the SMC, and wait for its response.*/ static int smu8_send_msg_to_smc(struct pp_hwmgr *hwmgr, uint16_t msg) { - int result = 0; - - result = smu8_send_msg_to_smc_async(hwmgr, msg); - if (result != 0) - return result; - - return PHM_WAIT_FIELD_UNEQUAL(hwmgr, - SMU_MP1_SRBM2P_RESP_0, CONTENT, 0); + return smu8_send_msg_to_smc_with_parameter(hwmgr, msg, 0); } static int smu8_set_smc_sram_address(struct pp_hwmgr *hwmgr, @@ -135,17 +143,6 @@ static int smu8_write_smc_sram_dword(struct pp_hwmgr *hwmgr, return result; } -static int smu8_send_msg_to_smc_with_parameter(struct pp_hwmgr *hwmgr, - uint16_t msg, uint32_t parameter) -{ - if (hwmgr == NULL || hwmgr->device == NULL) - return -EINVAL; - - cgs_write_register(hwmgr->device, mmSMU_MP1_SRBM2P_ARG_0, parameter); - - return smu8_send_msg_to_smc(hwmgr, msg); -} - static int smu8_check_fw_load_finish(struct pp_hwmgr *hwmgr, uint32_t firmware) { @@ -737,6 +734,10 @@ static int smu8_start_smu(struct pp_hwmgr *hwmgr) cgs_write_register(hwmgr->device, mmMP0PUB_IND_INDEX, index); hwmgr->smu_version = cgs_read_register(hwmgr->device, mmMP0PUB_IND_DATA); + pr_info("smu version %02d.%02d.%02d\n", + ((hwmgr->smu_version >> 16) & 0xFF), + ((hwmgr->smu_version >> 8) & 0xFF), + (hwmgr->smu_version & 0xFF)); adev->pm.fw_version = hwmgr->smu_version >> 8; return smu8_request_smu_load_fw(hwmgr); diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/smumgr.c index 99d5e4f98f49..a6edd5df33b0 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/smumgr.c @@ -37,10 +37,13 @@ MODULE_FIRMWARE("amdgpu/fiji_smc.bin"); MODULE_FIRMWARE("amdgpu/polaris10_smc.bin"); MODULE_FIRMWARE("amdgpu/polaris10_smc_sk.bin"); MODULE_FIRMWARE("amdgpu/polaris10_k_smc.bin"); +MODULE_FIRMWARE("amdgpu/polaris10_k2_smc.bin"); MODULE_FIRMWARE("amdgpu/polaris11_smc.bin"); MODULE_FIRMWARE("amdgpu/polaris11_smc_sk.bin"); MODULE_FIRMWARE("amdgpu/polaris11_k_smc.bin"); +MODULE_FIRMWARE("amdgpu/polaris11_k2_smc.bin"); MODULE_FIRMWARE("amdgpu/polaris12_smc.bin"); +MODULE_FIRMWARE("amdgpu/polaris12_k_smc.bin"); MODULE_FIRMWARE("amdgpu/vegam_smc.bin"); MODULE_FIRMWARE("amdgpu/vega10_smc.bin"); MODULE_FIRMWARE("amdgpu/vega10_acg_smc.bin"); diff --git a/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c b/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c index 9f71512b2510..1e69300f6175 100644 --- a/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c +++ b/drivers/gpu/drm/amd/powerplay/smumgr/vegam_smumgr.c @@ -40,7 +40,6 @@ #include "smu7_hwmgr.h" #include "hardwaremanager.h" -#include "ppatomctrl.h" #include "atombios.h" #include "pppcielanes.h" diff --git a/drivers/gpu/drm/ast/ast_drv.h b/drivers/gpu/drm/ast/ast_drv.h index e6c4cd3dc50e..bfc65040dfcb 100644 --- a/drivers/gpu/drm/ast/ast_drv.h +++ b/drivers/gpu/drm/ast/ast_drv.h @@ -104,8 +104,6 @@ struct ast_private { int fb_mtrr; struct { - struct drm_global_reference mem_global_ref; - struct ttm_bo_global_ref bo_global_ref; struct ttm_bo_device bdev; } ttm; diff --git a/drivers/gpu/drm/ast/ast_ttm.c b/drivers/gpu/drm/ast/ast_ttm.c index fe354ebf374d..c168d62fe8f9 100644 --- a/drivers/gpu/drm/ast/ast_ttm.c +++ b/drivers/gpu/drm/ast/ast_ttm.c @@ -36,63 +36,6 @@ ast_bdev(struct ttm_bo_device *bd) return container_of(bd, struct ast_private, ttm.bdev); } -static int -ast_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void -ast_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -static int ast_ttm_global_init(struct ast_private *ast) -{ - struct drm_global_reference *global_ref; - int r; - - global_ref = &ast->ttm.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &ast_ttm_mem_global_init; - global_ref->release = &ast_ttm_mem_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM memory accounting " - "subsystem.\n"); - return r; - } - - ast->ttm.bo_global_ref.mem_glob = - ast->ttm.mem_global_ref.object; - global_ref = &ast->ttm.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM BO subsystem.\n"); - drm_global_item_unref(&ast->ttm.mem_global_ref); - return r; - } - return 0; -} - -static void -ast_ttm_global_release(struct ast_private *ast) -{ - if (ast->ttm.mem_global_ref.release == NULL) - return; - - drm_global_item_unref(&ast->ttm.bo_global_ref.ref); - drm_global_item_unref(&ast->ttm.mem_global_ref); - ast->ttm.mem_global_ref.release = NULL; -} - - static void ast_bo_ttm_destroy(struct ttm_buffer_object *tbo) { struct ast_bo *bo; @@ -232,12 +175,7 @@ int ast_mm_init(struct ast_private *ast) struct drm_device *dev = ast->dev; struct ttm_bo_device *bdev = &ast->ttm.bdev; - ret = ast_ttm_global_init(ast); - if (ret) - return ret; - ret = ttm_bo_device_init(&ast->ttm.bdev, - ast->ttm.bo_global_ref.ref.object, &ast_bo_driver, dev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, @@ -268,8 +206,6 @@ void ast_mm_fini(struct ast_private *ast) ttm_bo_device_release(&ast->ttm.bdev); - ast_ttm_global_release(ast); - arch_phys_wc_del(ast->fb_mtrr); arch_io_free_memtype_wc(pci_resource_start(dev->pdev, 0), pci_resource_len(dev->pdev, 0)); diff --git a/drivers/gpu/drm/bochs/bochs.h b/drivers/gpu/drm/bochs/bochs.h index 577a8b917cb9..fb38c8b857b5 100644 --- a/drivers/gpu/drm/bochs/bochs.h +++ b/drivers/gpu/drm/bochs/bochs.h @@ -77,8 +77,6 @@ struct bochs_device { /* ttm */ struct { - struct drm_global_reference mem_global_ref; - struct ttm_bo_global_ref bo_global_ref; struct ttm_bo_device bdev; bool initialized; } ttm; diff --git a/drivers/gpu/drm/bochs/bochs_mm.c b/drivers/gpu/drm/bochs/bochs_mm.c index e6ccf7fa92d4..0980411e41bf 100644 --- a/drivers/gpu/drm/bochs/bochs_mm.c +++ b/drivers/gpu/drm/bochs/bochs_mm.c @@ -16,61 +16,6 @@ static inline struct bochs_device *bochs_bdev(struct ttm_bo_device *bd) return container_of(bd, struct bochs_device, ttm.bdev); } -static int bochs_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void bochs_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -static int bochs_ttm_global_init(struct bochs_device *bochs) -{ - struct drm_global_reference *global_ref; - int r; - - global_ref = &bochs->ttm.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &bochs_ttm_mem_global_init; - global_ref->release = &bochs_ttm_mem_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM memory accounting " - "subsystem.\n"); - return r; - } - - bochs->ttm.bo_global_ref.mem_glob = - bochs->ttm.mem_global_ref.object; - global_ref = &bochs->ttm.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM BO subsystem.\n"); - drm_global_item_unref(&bochs->ttm.mem_global_ref); - return r; - } - - return 0; -} - -static void bochs_ttm_global_release(struct bochs_device *bochs) -{ - if (bochs->ttm.mem_global_ref.release == NULL) - return; - - drm_global_item_unref(&bochs->ttm.bo_global_ref.ref); - drm_global_item_unref(&bochs->ttm.mem_global_ref); - bochs->ttm.mem_global_ref.release = NULL; -} - - static void bochs_bo_ttm_destroy(struct ttm_buffer_object *tbo) { struct bochs_bo *bo; @@ -208,12 +153,7 @@ int bochs_mm_init(struct bochs_device *bochs) struct ttm_bo_device *bdev = &bochs->ttm.bdev; int ret; - ret = bochs_ttm_global_init(bochs); - if (ret) - return ret; - ret = ttm_bo_device_init(&bochs->ttm.bdev, - bochs->ttm.bo_global_ref.ref.object, &bochs_bo_driver, bochs->dev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, @@ -240,7 +180,6 @@ void bochs_mm_fini(struct bochs_device *bochs) return; ttm_bo_device_release(&bochs->ttm.bdev); - bochs_ttm_global_release(bochs); bochs->ttm.initialized = false; } diff --git a/drivers/gpu/drm/cirrus/cirrus_drv.h b/drivers/gpu/drm/cirrus/cirrus_drv.h index a29f87e98d9d..f2b2e0d169fa 100644 --- a/drivers/gpu/drm/cirrus/cirrus_drv.h +++ b/drivers/gpu/drm/cirrus/cirrus_drv.h @@ -136,8 +136,6 @@ struct cirrus_device { int fb_mtrr; struct { - struct drm_global_reference mem_global_ref; - struct ttm_bo_global_ref bo_global_ref; struct ttm_bo_device bdev; } ttm; bool mm_inited; diff --git a/drivers/gpu/drm/cirrus/cirrus_ttm.c b/drivers/gpu/drm/cirrus/cirrus_ttm.c index f21953243790..e075810b4bd4 100644 --- a/drivers/gpu/drm/cirrus/cirrus_ttm.c +++ b/drivers/gpu/drm/cirrus/cirrus_ttm.c @@ -36,63 +36,6 @@ cirrus_bdev(struct ttm_bo_device *bd) return container_of(bd, struct cirrus_device, ttm.bdev); } -static int -cirrus_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void -cirrus_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -static int cirrus_ttm_global_init(struct cirrus_device *cirrus) -{ - struct drm_global_reference *global_ref; - int r; - - global_ref = &cirrus->ttm.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &cirrus_ttm_mem_global_init; - global_ref->release = &cirrus_ttm_mem_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM memory accounting " - "subsystem.\n"); - return r; - } - - cirrus->ttm.bo_global_ref.mem_glob = - cirrus->ttm.mem_global_ref.object; - global_ref = &cirrus->ttm.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM BO subsystem.\n"); - drm_global_item_unref(&cirrus->ttm.mem_global_ref); - return r; - } - return 0; -} - -static void -cirrus_ttm_global_release(struct cirrus_device *cirrus) -{ - if (cirrus->ttm.mem_global_ref.release == NULL) - return; - - drm_global_item_unref(&cirrus->ttm.bo_global_ref.ref); - drm_global_item_unref(&cirrus->ttm.mem_global_ref); - cirrus->ttm.mem_global_ref.release = NULL; -} - - static void cirrus_bo_ttm_destroy(struct ttm_buffer_object *tbo) { struct cirrus_bo *bo; @@ -232,12 +175,7 @@ int cirrus_mm_init(struct cirrus_device *cirrus) struct drm_device *dev = cirrus->dev; struct ttm_bo_device *bdev = &cirrus->ttm.bdev; - ret = cirrus_ttm_global_init(cirrus); - if (ret) - return ret; - ret = ttm_bo_device_init(&cirrus->ttm.bdev, - cirrus->ttm.bo_global_ref.ref.object, &cirrus_bo_driver, dev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, @@ -273,8 +211,6 @@ void cirrus_mm_fini(struct cirrus_device *cirrus) ttm_bo_device_release(&cirrus->ttm.bdev); - cirrus_ttm_global_release(cirrus); - arch_phys_wc_del(cirrus->fb_mtrr); cirrus->fb_mtrr = 0; arch_io_free_memtype_wc(pci_resource_start(dev->pdev, 0), diff --git a/drivers/gpu/drm/drm_drv.c b/drivers/gpu/drm/drm_drv.c index aa1cef794f9a..e2ffecd5e453 100644 --- a/drivers/gpu/drm/drm_drv.c +++ b/drivers/gpu/drm/drm_drv.c @@ -973,14 +973,12 @@ static void drm_core_exit(void) drm_sysfs_destroy(); idr_destroy(&drm_minors_idr); drm_connector_ida_destroy(); - drm_global_release(); } static int __init drm_core_init(void) { int ret; - drm_global_init(); drm_connector_ida_init(); idr_init(&drm_minors_idr); diff --git a/drivers/gpu/drm/drm_global.c b/drivers/gpu/drm/drm_global.c deleted file mode 100644 index 5799e2782dd1..000000000000 --- a/drivers/gpu/drm/drm_global.c +++ /dev/null @@ -1,137 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0 OR MIT -/************************************************************************** - * - * Copyright 2008-2009 VMware, Inc., Palo Alto, CA., USA - * - * 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, sub license, 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 (including the - * next paragraph) 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 NON-INFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS 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. - * - **************************************************************************/ -/* - * Authors: Thomas Hellstrom <thellstrom-at-vmware-dot-com> - */ - -#include <linux/mutex.h> -#include <linux/slab.h> -#include <linux/module.h> -#include <drm/drm_global.h> - -struct drm_global_item { - struct mutex mutex; - void *object; - int refcount; -}; - -static struct drm_global_item glob[DRM_GLOBAL_NUM]; - -void drm_global_init(void) -{ - int i; - - for (i = 0; i < DRM_GLOBAL_NUM; ++i) { - struct drm_global_item *item = &glob[i]; - mutex_init(&item->mutex); - item->object = NULL; - item->refcount = 0; - } -} - -void drm_global_release(void) -{ - int i; - for (i = 0; i < DRM_GLOBAL_NUM; ++i) { - struct drm_global_item *item = &glob[i]; - BUG_ON(item->object != NULL); - BUG_ON(item->refcount != 0); - } -} - -/** - * drm_global_item_ref - Initialize and acquire reference to memory - * object - * @ref: Object for initialization - * - * This initializes a memory object, allocating memory and calling the - * .init() hook. Further calls will increase the reference count for - * that item. - * - * Returns: - * Zero on success, non-zero otherwise. - */ -int drm_global_item_ref(struct drm_global_reference *ref) -{ - int ret = 0; - struct drm_global_item *item = &glob[ref->global_type]; - - mutex_lock(&item->mutex); - if (item->refcount == 0) { - ref->object = kzalloc(ref->size, GFP_KERNEL); - if (unlikely(ref->object == NULL)) { - ret = -ENOMEM; - goto error_unlock; - } - ret = ref->init(ref); - if (unlikely(ret != 0)) - goto error_free; - - item->object = ref->object; - } else { - ref->object = item->object; - } - - ++item->refcount; - mutex_unlock(&item->mutex); - return 0; - -error_free: - kfree(ref->object); - ref->object = NULL; -error_unlock: - mutex_unlock(&item->mutex); - return ret; -} -EXPORT_SYMBOL(drm_global_item_ref); - -/** - * drm_global_item_unref - Drop reference to memory - * object - * @ref: Object being removed - * - * Drop a reference to the memory object and eventually call the - * release() hook. The allocated object should be dropped in the - * release() hook or before calling this function - * - */ - -void drm_global_item_unref(struct drm_global_reference *ref) -{ - struct drm_global_item *item = &glob[ref->global_type]; - - mutex_lock(&item->mutex); - BUG_ON(item->refcount == 0); - BUG_ON(ref->object != item->object); - if (--item->refcount == 0) { - ref->release(ref); - item->object = NULL; - } - mutex_unlock(&item->mutex); -} -EXPORT_SYMBOL(drm_global_item_unref); - diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c index 9b476368aa31..49a6763693f1 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c @@ -105,8 +105,6 @@ static void etnaviv_sched_timedout_job(struct drm_sched_job *sched_job) change = dma_addr - gpu->hangcheck_dma_addr; if (change < 0 || change > 16) { gpu->hangcheck_dma_addr = dma_addr; - schedule_delayed_work(&sched_job->sched->work_tdr, - sched_job->sched->timeout); return; } @@ -127,6 +125,8 @@ static void etnaviv_sched_free_job(struct drm_sched_job *sched_job) { struct etnaviv_gem_submit *submit = to_etnaviv_submit(sched_job); + drm_sched_job_cleanup(sched_job); + etnaviv_submit_put(submit); } @@ -159,6 +159,7 @@ int etnaviv_sched_push_job(struct drm_sched_entity *sched_entity, submit->out_fence, 0, INT_MAX, GFP_KERNEL); if (submit->out_fence_id < 0) { + drm_sched_job_cleanup(&submit->sched_job); ret = -ENOMEM; goto out_unlock; } diff --git a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.h b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.h index 45c25a488f42..3c168ae77b0c 100644 --- a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.h +++ b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.h @@ -49,8 +49,6 @@ struct hibmc_drm_private { bool mode_config_initialized; /* ttm */ - struct drm_global_reference mem_global_ref; - struct ttm_bo_global_ref bo_global_ref; struct ttm_bo_device bdev; bool initialized; diff --git a/drivers/gpu/drm/hisilicon/hibmc/hibmc_ttm.c b/drivers/gpu/drm/hisilicon/hibmc/hibmc_ttm.c index 2e3e0bdb8932..dd383267884c 100644 --- a/drivers/gpu/drm/hisilicon/hibmc/hibmc_ttm.c +++ b/drivers/gpu/drm/hisilicon/hibmc/hibmc_ttm.c @@ -29,55 +29,6 @@ hibmc_bdev(struct ttm_bo_device *bd) return container_of(bd, struct hibmc_drm_private, bdev); } -static int -hibmc_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void -hibmc_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -static int hibmc_ttm_global_init(struct hibmc_drm_private *hibmc) -{ - int ret; - - hibmc->mem_global_ref.global_type = DRM_GLOBAL_TTM_MEM; - hibmc->mem_global_ref.size = sizeof(struct ttm_mem_global); - hibmc->mem_global_ref.init = &hibmc_ttm_mem_global_init; - hibmc->mem_global_ref.release = &hibmc_ttm_mem_global_release; - ret = drm_global_item_ref(&hibmc->mem_global_ref); - if (ret) { - DRM_ERROR("could not get ref on ttm global: %d\n", ret); - return ret; - } - - hibmc->bo_global_ref.mem_glob = - hibmc->mem_global_ref.object; - hibmc->bo_global_ref.ref.global_type = DRM_GLOBAL_TTM_BO; - hibmc->bo_global_ref.ref.size = sizeof(struct ttm_bo_global); - hibmc->bo_global_ref.ref.init = &ttm_bo_global_init; - hibmc->bo_global_ref.ref.release = &ttm_bo_global_release; - ret = drm_global_item_ref(&hibmc->bo_global_ref.ref); - if (ret) { - DRM_ERROR("failed setting up TTM BO subsystem: %d\n", ret); - drm_global_item_unref(&hibmc->mem_global_ref); - return ret; - } - return 0; -} - -static void -hibmc_ttm_global_release(struct hibmc_drm_private *hibmc) -{ - drm_global_item_unref(&hibmc->bo_global_ref.ref); - drm_global_item_unref(&hibmc->mem_global_ref); - hibmc->mem_global_ref.release = NULL; -} - static void hibmc_bo_ttm_destroy(struct ttm_buffer_object *tbo) { struct hibmc_bo *bo = container_of(tbo, struct hibmc_bo, bo); @@ -237,18 +188,12 @@ int hibmc_mm_init(struct hibmc_drm_private *hibmc) struct drm_device *dev = hibmc->dev; struct ttm_bo_device *bdev = &hibmc->bdev; - ret = hibmc_ttm_global_init(hibmc); - if (ret) - return ret; - ret = ttm_bo_device_init(&hibmc->bdev, - hibmc->bo_global_ref.ref.object, &hibmc_bo_driver, dev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, true); if (ret) { - hibmc_ttm_global_release(hibmc); DRM_ERROR("error initializing bo driver: %d\n", ret); return ret; } @@ -256,7 +201,6 @@ int hibmc_mm_init(struct hibmc_drm_private *hibmc) ret = ttm_bo_init_mm(bdev, TTM_PL_VRAM, hibmc->fb_size >> PAGE_SHIFT); if (ret) { - hibmc_ttm_global_release(hibmc); DRM_ERROR("failed ttm VRAM init: %d\n", ret); return ret; } @@ -271,7 +215,6 @@ void hibmc_mm_fini(struct hibmc_drm_private *hibmc) return; ttm_bo_device_release(&hibmc->bdev); - hibmc_ttm_global_release(hibmc); hibmc->mm_inited = false; } diff --git a/drivers/gpu/drm/mgag200/mgag200_drv.h b/drivers/gpu/drm/mgag200/mgag200_drv.h index 04f1dfba12e5..0aaedc554879 100644 --- a/drivers/gpu/drm/mgag200/mgag200_drv.h +++ b/drivers/gpu/drm/mgag200/mgag200_drv.h @@ -212,8 +212,6 @@ struct mga_device { int fb_mtrr; struct { - struct drm_global_reference mem_global_ref; - struct ttm_bo_global_ref bo_global_ref; struct ttm_bo_device bdev; } ttm; diff --git a/drivers/gpu/drm/mgag200/mgag200_ttm.c b/drivers/gpu/drm/mgag200/mgag200_ttm.c index 05570f0de4d7..d96a9b32455e 100644 --- a/drivers/gpu/drm/mgag200/mgag200_ttm.c +++ b/drivers/gpu/drm/mgag200/mgag200_ttm.c @@ -36,63 +36,6 @@ mgag200_bdev(struct ttm_bo_device *bd) return container_of(bd, struct mga_device, ttm.bdev); } -static int -mgag200_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void -mgag200_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -static int mgag200_ttm_global_init(struct mga_device *ast) -{ - struct drm_global_reference *global_ref; - int r; - - global_ref = &ast->ttm.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &mgag200_ttm_mem_global_init; - global_ref->release = &mgag200_ttm_mem_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM memory accounting " - "subsystem.\n"); - return r; - } - - ast->ttm.bo_global_ref.mem_glob = - ast->ttm.mem_global_ref.object; - global_ref = &ast->ttm.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM BO subsystem.\n"); - drm_global_item_unref(&ast->ttm.mem_global_ref); - return r; - } - return 0; -} - -static void -mgag200_ttm_global_release(struct mga_device *ast) -{ - if (ast->ttm.mem_global_ref.release == NULL) - return; - - drm_global_item_unref(&ast->ttm.bo_global_ref.ref); - drm_global_item_unref(&ast->ttm.mem_global_ref); - ast->ttm.mem_global_ref.release = NULL; -} - - static void mgag200_bo_ttm_destroy(struct ttm_buffer_object *tbo) { struct mgag200_bo *bo; @@ -232,12 +175,7 @@ int mgag200_mm_init(struct mga_device *mdev) struct drm_device *dev = mdev->dev; struct ttm_bo_device *bdev = &mdev->ttm.bdev; - ret = mgag200_ttm_global_init(mdev); - if (ret) - return ret; - ret = ttm_bo_device_init(&mdev->ttm.bdev, - mdev->ttm.bo_global_ref.ref.object, &mgag200_bo_driver, dev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, @@ -268,8 +206,6 @@ void mgag200_mm_fini(struct mga_device *mdev) ttm_bo_device_release(&mdev->ttm.bdev); - mgag200_ttm_global_release(mdev); - arch_io_free_memtype_wc(pci_resource_start(dev->pdev, 0), pci_resource_len(dev->pdev, 0)); arch_phys_wc_del(mdev->fb_mtrr); diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h index 0b2191fa96f7..d20b9ba4b1c1 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.h +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h @@ -146,8 +146,6 @@ struct nouveau_drm { /* TTM interface support */ struct { - struct drm_global_reference mem_global_ref; - struct ttm_bo_global_ref bo_global_ref; struct ttm_bo_device bdev; atomic_t validate_sequence; int (*move)(struct nouveau_channel *, diff --git a/drivers/gpu/drm/nouveau/nouveau_ttm.c b/drivers/gpu/drm/nouveau/nouveau_ttm.c index 8edb9f2a4269..1543c2f8d3d3 100644 --- a/drivers/gpu/drm/nouveau/nouveau_ttm.c +++ b/drivers/gpu/drm/nouveau/nouveau_ttm.c @@ -175,66 +175,6 @@ nouveau_ttm_mmap(struct file *filp, struct vm_area_struct *vma) } static int -nouveau_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void -nouveau_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -int -nouveau_ttm_global_init(struct nouveau_drm *drm) -{ - struct drm_global_reference *global_ref; - int ret; - - global_ref = &drm->ttm.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &nouveau_ttm_mem_global_init; - global_ref->release = &nouveau_ttm_mem_global_release; - - ret = drm_global_item_ref(global_ref); - if (unlikely(ret != 0)) { - DRM_ERROR("Failed setting up TTM memory accounting\n"); - drm->ttm.mem_global_ref.release = NULL; - return ret; - } - - drm->ttm.bo_global_ref.mem_glob = global_ref->object; - global_ref = &drm->ttm.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - - ret = drm_global_item_ref(global_ref); - if (unlikely(ret != 0)) { - DRM_ERROR("Failed setting up TTM BO subsystem\n"); - drm_global_item_unref(&drm->ttm.mem_global_ref); - drm->ttm.mem_global_ref.release = NULL; - return ret; - } - - return 0; -} - -void -nouveau_ttm_global_release(struct nouveau_drm *drm) -{ - if (drm->ttm.mem_global_ref.release == NULL) - return; - - drm_global_item_unref(&drm->ttm.bo_global_ref.ref); - drm_global_item_unref(&drm->ttm.mem_global_ref); - drm->ttm.mem_global_ref.release = NULL; -} - -static int nouveau_ttm_init_host(struct nouveau_drm *drm, u8 kind) { struct nvif_mmu *mmu = &drm->client.mmu; @@ -296,12 +236,7 @@ nouveau_ttm_init(struct nouveau_drm *drm) drm->agp.cma = pci->agp.cma; } - ret = nouveau_ttm_global_init(drm); - if (ret) - return ret; - ret = ttm_bo_device_init(&drm->ttm.bdev, - drm->ttm.bo_global_ref.ref.object, &nouveau_bo_driver, dev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, @@ -356,8 +291,6 @@ nouveau_ttm_fini(struct nouveau_drm *drm) ttm_bo_device_release(&drm->ttm.bdev); - nouveau_ttm_global_release(drm); - arch_phys_wc_del(drm->ttm.mtrr); drm->ttm.mtrr = 0; arch_io_free_memtype_wc(device->func->resource_addr(device, 1), diff --git a/drivers/gpu/drm/qxl/qxl_drv.h b/drivers/gpu/drm/qxl/qxl_drv.h index 14d3fa855708..13a0254b59a1 100644 --- a/drivers/gpu/drm/qxl/qxl_drv.h +++ b/drivers/gpu/drm/qxl/qxl_drv.h @@ -126,9 +126,6 @@ struct qxl_output { #define drm_encoder_to_qxl_output(x) container_of(x, struct qxl_output, enc) struct qxl_mman { - struct ttm_bo_global_ref bo_global_ref; - struct drm_global_reference mem_global_ref; - unsigned int mem_global_referenced:1; struct ttm_bo_device bdev; }; diff --git a/drivers/gpu/drm/qxl/qxl_ttm.c b/drivers/gpu/drm/qxl/qxl_ttm.c index 559a10113837..886f61e94f24 100644 --- a/drivers/gpu/drm/qxl/qxl_ttm.c +++ b/drivers/gpu/drm/qxl/qxl_ttm.c @@ -46,62 +46,6 @@ static struct qxl_device *qxl_get_qdev(struct ttm_bo_device *bdev) return qdev; } -static int qxl_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void qxl_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -static int qxl_ttm_global_init(struct qxl_device *qdev) -{ - struct drm_global_reference *global_ref; - int r; - - qdev->mman.mem_global_referenced = false; - global_ref = &qdev->mman.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &qxl_ttm_mem_global_init; - global_ref->release = &qxl_ttm_mem_global_release; - - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM memory accounting " - "subsystem.\n"); - return r; - } - - qdev->mman.bo_global_ref.mem_glob = - qdev->mman.mem_global_ref.object; - global_ref = &qdev->mman.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM BO subsystem.\n"); - drm_global_item_unref(&qdev->mman.mem_global_ref); - return r; - } - - qdev->mman.mem_global_referenced = true; - return 0; -} - -static void qxl_ttm_global_fini(struct qxl_device *qdev) -{ - if (qdev->mman.mem_global_referenced) { - drm_global_item_unref(&qdev->mman.bo_global_ref.ref); - drm_global_item_unref(&qdev->mman.mem_global_ref); - qdev->mman.mem_global_referenced = false; - } -} - static struct vm_operations_struct qxl_ttm_vm_ops; static const struct vm_operations_struct *ttm_vm_ops; @@ -372,12 +316,8 @@ int qxl_ttm_init(struct qxl_device *qdev) int r; int num_io_pages; /* != rom->num_io_pages, we include surface0 */ - r = qxl_ttm_global_init(qdev); - if (r) - return r; /* No others user of address space so set it to 0 */ r = ttm_bo_device_init(&qdev->mman.bdev, - qdev->mman.bo_global_ref.ref.object, &qxl_bo_driver, qdev->ddev.anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, 0); @@ -413,7 +353,6 @@ void qxl_ttm_fini(struct qxl_device *qdev) ttm_bo_clean_mm(&qdev->mman.bdev, TTM_PL_VRAM); ttm_bo_clean_mm(&qdev->mman.bdev, TTM_PL_PRIV); ttm_bo_device_release(&qdev->mman.bdev); - qxl_ttm_global_fini(qdev); DRM_INFO("qxl: ttm finalized\n"); } diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c index 21161aa8acbf..652126fd6dd4 100644 --- a/drivers/gpu/drm/radeon/r300.c +++ b/drivers/gpu/drm/radeon/r300.c @@ -814,7 +814,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, ((idx_value >> 21) & 0xF)); return -EINVAL; } - /* Pass through. */ + /* Fall through. */ case 6: track->cb[i].cpp = 4; break; @@ -965,7 +965,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, return -EINVAL; } /* The same rules apply as for DXT3/5. */ - /* Pass through. */ + /* Fall through. */ case R300_TX_FORMAT_DXT3: case R300_TX_FORMAT_DXT5: track->textures[i].cpp = 1; diff --git a/drivers/gpu/drm/radeon/r420.c b/drivers/gpu/drm/radeon/r420.c index 45e1d4e60759..2318d9e3ed96 100644 --- a/drivers/gpu/drm/radeon/r420.c +++ b/drivers/gpu/drm/radeon/r420.c @@ -109,6 +109,7 @@ void r420_pipes_init(struct radeon_device *rdev) default: /* force to 1 pipe */ num_pipes = 1; + /* fall through */ case 1: tmp = (0 << 1); break; diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index 1a6f6edb3515..32808e50be12 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -448,10 +448,7 @@ struct radeon_surface_reg { * TTM. */ struct radeon_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) diff --git a/drivers/gpu/drm/radeon/radeon_legacy_tv.c b/drivers/gpu/drm/radeon/radeon_legacy_tv.c index 4278272e3191..3dae2c4dec71 100644 --- a/drivers/gpu/drm/radeon/radeon_legacy_tv.c +++ b/drivers/gpu/drm/radeon/radeon_legacy_tv.c @@ -421,24 +421,14 @@ static void radeon_legacy_write_tv_restarts(struct radeon_encoder *radeon_encode static bool radeon_legacy_tv_init_restarts(struct drm_encoder *encoder) { - struct drm_device *dev = encoder->dev; - struct radeon_device *rdev = dev->dev_private; struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct radeon_encoder_tv_dac *tv_dac = radeon_encoder->enc_priv; - struct radeon_crtc *radeon_crtc; int restart; unsigned int h_total, v_total, f_total; int v_offset, h_offset; u16 p1, p2, h_inc; bool h_changed; const struct radeon_tv_mode_constants *const_ptr; - struct radeon_pll *pll; - - radeon_crtc = to_radeon_crtc(radeon_encoder->base.crtc); - if (radeon_crtc->crtc_id == 1) - pll = &rdev->clock.p2pll; - else - pll = &rdev->clock.p1pll; const_ptr = radeon_legacy_tv_get_std_mode(radeon_encoder, NULL); if (!const_ptr) diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 92f6d4002eea..833e909706a9 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -314,11 +314,9 @@ struct radeon_bo *radeon_bo_ref(struct radeon_bo *bo) void radeon_bo_unref(struct radeon_bo **bo) { struct ttm_buffer_object *tbo; - struct radeon_device *rdev; if ((*bo) == NULL) return; - rdev = (*bo)->rdev; tbo = &((*bo)->tbo); ttm_bo_put(tbo); *bo = NULL; diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index cbb67e9ffb3a..9920a6fc11bf 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -60,65 +60,6 @@ static struct radeon_device *radeon_get_rdev(struct ttm_bo_device *bdev) return rdev; } - -/* - * Global memory. - */ -static int radeon_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void radeon_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -static int radeon_ttm_global_init(struct radeon_device *rdev) -{ - struct drm_global_reference *global_ref; - int r; - - rdev->mman.mem_global_referenced = false; - global_ref = &rdev->mman.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &radeon_ttm_mem_global_init; - global_ref->release = &radeon_ttm_mem_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM memory accounting " - "subsystem.\n"); - return r; - } - - rdev->mman.bo_global_ref.mem_glob = - rdev->mman.mem_global_ref.object; - global_ref = &rdev->mman.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM BO subsystem.\n"); - drm_global_item_unref(&rdev->mman.mem_global_ref); - return r; - } - - rdev->mman.mem_global_referenced = true; - return 0; -} - -static void radeon_ttm_global_fini(struct radeon_device *rdev) -{ - if (rdev->mman.mem_global_referenced) { - drm_global_item_unref(&rdev->mman.bo_global_ref.ref); - drm_global_item_unref(&rdev->mman.mem_global_ref); - rdev->mman.mem_global_referenced = false; - } -} - static int radeon_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags) { return 0; @@ -847,13 +788,8 @@ int radeon_ttm_init(struct radeon_device *rdev) { int r; - r = radeon_ttm_global_init(rdev); - if (r) { - return r; - } /* No others user of address space so set it to 0 */ r = ttm_bo_device_init(&rdev->mman.bdev, - rdev->mman.bo_global_ref.ref.object, &radeon_bo_driver, rdev->ddev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, @@ -925,7 +861,6 @@ void radeon_ttm_fini(struct radeon_device *rdev) ttm_bo_clean_mm(&rdev->mman.bdev, TTM_PL_TT); ttm_bo_device_release(&rdev->mman.bdev); radeon_gart_fini(rdev); - radeon_ttm_global_fini(rdev); rdev->mman.initialized = false; DRM_INFO("radeon: ttm finalized\n"); } diff --git a/drivers/gpu/drm/scheduler/sched_entity.c b/drivers/gpu/drm/scheduler/sched_entity.c index 3e22a54a99c2..4463d3826ecb 100644 --- a/drivers/gpu/drm/scheduler/sched_entity.c +++ b/drivers/gpu/drm/scheduler/sched_entity.c @@ -130,7 +130,14 @@ drm_sched_entity_get_free_sched(struct drm_sched_entity *entity) int i; for (i = 0; i < entity->num_rq_list; ++i) { - num_jobs = atomic_read(&entity->rq_list[i]->sched->num_jobs); + struct drm_gpu_scheduler *sched = entity->rq_list[i]->sched; + + if (!entity->rq_list[i]->sched->ready) { + DRM_WARN("sched%s is not ready, skipping", sched->name); + continue; + } + + num_jobs = atomic_read(&sched->num_jobs); if (num_jobs < min_jobs) { min_jobs = num_jobs; rq = entity->rq_list[i]; @@ -204,7 +211,6 @@ static void drm_sched_entity_kill_jobs_cb(struct dma_fence *f, drm_sched_fence_finished(job->s_fence); WARN_ON(job->s_fence->parent); - dma_fence_put(&job->s_fence->finished); job->sched->ops->free_job(job); } diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 44fe587aaef9..18ebbb05762e 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -196,6 +196,19 @@ static void drm_sched_start_timeout(struct drm_gpu_scheduler *sched) schedule_delayed_work(&sched->work_tdr, sched->timeout); } +/** + * drm_sched_fault - immediately start timeout handler + * + * @sched: scheduler where the timeout handling should be started. + * + * Start timeout handling immediately when the driver detects a hardware fault. + */ +void drm_sched_fault(struct drm_gpu_scheduler *sched) +{ + mod_delayed_work(system_wq, &sched->work_tdr, 0); +} +EXPORT_SYMBOL(drm_sched_fault); + /* job_finish is called after hw fence signaled */ static void drm_sched_job_finish(struct work_struct *work) @@ -220,7 +233,6 @@ static void drm_sched_job_finish(struct work_struct *work) drm_sched_start_timeout(sched); spin_unlock(&sched->job_list_lock); - dma_fence_put(&s_job->s_fence->finished); sched->ops->free_job(s_job); } @@ -283,6 +295,7 @@ static void drm_sched_job_timedout(struct work_struct *work) already_signaled: ; } + drm_sched_start_timeout(sched); spin_unlock(&sched->job_list_lock); } @@ -406,6 +419,9 @@ int drm_sched_job_init(struct drm_sched_job *job, struct drm_gpu_scheduler *sched; drm_sched_entity_select_rq(entity); + if (!entity->rq) + return -ENOENT; + sched = entity->rq->sched; job->sched = sched; @@ -424,6 +440,18 @@ int drm_sched_job_init(struct drm_sched_job *job, EXPORT_SYMBOL(drm_sched_job_init); /** + * drm_sched_job_cleanup - clean up scheduler job resources + * + * @job: scheduler job to clean up + */ +void drm_sched_job_cleanup(struct drm_sched_job *job) +{ + dma_fence_put(&job->s_fence->finished); + job->s_fence = NULL; +} +EXPORT_SYMBOL(drm_sched_job_cleanup); + +/** * drm_sched_ready - is the scheduler ready * * @sched: scheduler instance @@ -619,6 +647,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched, return PTR_ERR(sched->thread); } + sched->ready = true; return 0; } EXPORT_SYMBOL(drm_sched_init); @@ -634,5 +663,7 @@ void drm_sched_fini(struct drm_gpu_scheduler *sched) { if (sched->thread) kthread_stop(sched->thread); + + sched->ready = false; } EXPORT_SYMBOL(drm_sched_fini); diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 83b4657ffb10..d87935bf8e30 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -45,6 +45,14 @@ static void ttm_bo_global_kobj_release(struct kobject *kobj); +/** + * ttm_global_mutex - protecting the global BO state + */ +DEFINE_MUTEX(ttm_global_mutex); +struct ttm_bo_global ttm_bo_glob = { + .use_count = 0 +}; + static struct attribute ttm_bo_count = { .name = "bo_count", .mode = S_IRUGO @@ -1519,35 +1527,45 @@ static void ttm_bo_global_kobj_release(struct kobject *kobj) container_of(kobj, struct ttm_bo_global, kobj); __free_page(glob->dummy_read_page); - kfree(glob); } -void ttm_bo_global_release(struct drm_global_reference *ref) +static void ttm_bo_global_release(void) { - struct ttm_bo_global *glob = ref->object; + struct ttm_bo_global *glob = &ttm_bo_glob; + + mutex_lock(&ttm_global_mutex); + if (--glob->use_count > 0) + goto out; kobject_del(&glob->kobj); kobject_put(&glob->kobj); + ttm_mem_global_release(&ttm_mem_glob); +out: + mutex_unlock(&ttm_global_mutex); } -EXPORT_SYMBOL(ttm_bo_global_release); -int ttm_bo_global_init(struct drm_global_reference *ref) +static int ttm_bo_global_init(void) { - struct ttm_bo_global_ref *bo_ref = - container_of(ref, struct ttm_bo_global_ref, ref); - struct ttm_bo_global *glob = ref->object; - int ret; + struct ttm_bo_global *glob = &ttm_bo_glob; + int ret = 0; unsigned i; - mutex_init(&glob->device_list_mutex); + mutex_lock(&ttm_global_mutex); + if (++glob->use_count > 1) + goto out; + + ret = ttm_mem_global_init(&ttm_mem_glob); + if (ret) + goto out; + spin_lock_init(&glob->lru_lock); - glob->mem_glob = bo_ref->mem_glob; + glob->mem_glob = &ttm_mem_glob; glob->mem_glob->bo_glob = glob; glob->dummy_read_page = alloc_page(__GFP_ZERO | GFP_DMA32); if (unlikely(glob->dummy_read_page == NULL)) { ret = -ENOMEM; - goto out_no_drp; + goto out; } for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) @@ -1559,13 +1577,10 @@ int ttm_bo_global_init(struct drm_global_reference *ref) &glob->kobj, &ttm_bo_glob_kobj_type, ttm_get_kobj(), "buffer_objects"); if (unlikely(ret != 0)) kobject_put(&glob->kobj); - return ret; -out_no_drp: - kfree(glob); +out: + mutex_unlock(&ttm_global_mutex); return ret; } -EXPORT_SYMBOL(ttm_bo_global_init); - int ttm_bo_device_release(struct ttm_bo_device *bdev) { @@ -1587,9 +1602,9 @@ int ttm_bo_device_release(struct ttm_bo_device *bdev) } } - mutex_lock(&glob->device_list_mutex); + mutex_lock(&ttm_global_mutex); list_del(&bdev->device_list); - mutex_unlock(&glob->device_list_mutex); + mutex_unlock(&ttm_global_mutex); cancel_delayed_work_sync(&bdev->wq); @@ -1604,18 +1619,25 @@ int ttm_bo_device_release(struct ttm_bo_device *bdev) drm_vma_offset_manager_destroy(&bdev->vma_manager); + if (!ret) + ttm_bo_global_release(); + return ret; } EXPORT_SYMBOL(ttm_bo_device_release); int ttm_bo_device_init(struct ttm_bo_device *bdev, - struct ttm_bo_global *glob, struct ttm_bo_driver *driver, struct address_space *mapping, uint64_t file_page_offset, bool need_dma32) { - int ret = -EINVAL; + struct ttm_bo_global *glob = &ttm_bo_glob; + int ret; + + ret = ttm_bo_global_init(); + if (ret) + return ret; bdev->driver = driver; @@ -1636,12 +1658,13 @@ int ttm_bo_device_init(struct ttm_bo_device *bdev, bdev->dev_mapping = mapping; bdev->glob = glob; bdev->need_dma32 = need_dma32; - mutex_lock(&glob->device_list_mutex); + mutex_lock(&ttm_global_mutex); list_add_tail(&bdev->device_list, &glob->device_list); - mutex_unlock(&glob->device_list_mutex); + mutex_unlock(&ttm_global_mutex); return 0; out_no_sys: + ttm_bo_global_release(); return ret; } EXPORT_SYMBOL(ttm_bo_device_init); diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c b/drivers/gpu/drm/ttm/ttm_execbuf_util.c index e493edb0d3e7..efa005a1c1b7 100644 --- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c +++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c @@ -187,14 +187,12 @@ void ttm_eu_fence_buffer_objects(struct ww_acquire_ctx *ticket, struct ttm_buffer_object *bo; struct ttm_bo_global *glob; struct ttm_bo_device *bdev; - struct ttm_bo_driver *driver; if (list_empty(list)) return; bo = list_first_entry(list, struct ttm_validate_buffer, head)->bo; bdev = bo->bdev; - driver = bdev->driver; glob = bo->bdev->glob; spin_lock(&glob->lru_lock); diff --git a/drivers/gpu/drm/ttm/ttm_memory.c b/drivers/gpu/drm/ttm/ttm_memory.c index 450387c92b63..f1567c353b54 100644 --- a/drivers/gpu/drm/ttm/ttm_memory.c +++ b/drivers/gpu/drm/ttm/ttm_memory.c @@ -41,6 +41,9 @@ #define TTM_MEMORY_ALLOC_RETRIES 4 +struct ttm_mem_global ttm_mem_glob; +EXPORT_SYMBOL(ttm_mem_glob); + struct ttm_mem_zone { struct kobject kobj; struct ttm_mem_global *glob; @@ -216,14 +219,6 @@ static ssize_t ttm_mem_global_store(struct kobject *kobj, return size; } -static void ttm_mem_global_kobj_release(struct kobject *kobj) -{ - struct ttm_mem_global *glob = - container_of(kobj, struct ttm_mem_global, kobj); - - kfree(glob); -} - static struct attribute *ttm_mem_global_attrs[] = { &ttm_mem_global_lower_mem_limit, NULL @@ -235,7 +230,6 @@ static const struct sysfs_ops ttm_mem_global_ops = { }; static struct kobj_type ttm_mem_glob_kobj_type = { - .release = &ttm_mem_global_kobj_release, .sysfs_ops = &ttm_mem_global_ops, .default_attrs = ttm_mem_global_attrs, }; @@ -464,7 +458,6 @@ out_no_zone: ttm_mem_global_release(glob); return ret; } -EXPORT_SYMBOL(ttm_mem_global_init); void ttm_mem_global_release(struct ttm_mem_global *glob) { @@ -486,7 +479,6 @@ void ttm_mem_global_release(struct ttm_mem_global *glob) kobject_del(&glob->kobj); kobject_put(&glob->kobj); } -EXPORT_SYMBOL(ttm_mem_global_release); static void ttm_check_swapping(struct ttm_mem_global *glob) { diff --git a/drivers/gpu/drm/v3d/v3d_sched.c b/drivers/gpu/drm/v3d/v3d_sched.c index 9243dea6e6ad..445b2ef03303 100644 --- a/drivers/gpu/drm/v3d/v3d_sched.c +++ b/drivers/gpu/drm/v3d/v3d_sched.c @@ -35,6 +35,8 @@ v3d_job_free(struct drm_sched_job *sched_job) { struct v3d_job *job = to_v3d_job(sched_job); + drm_sched_job_cleanup(sched_job); + v3d_exec_put(job->exec); } @@ -167,9 +169,6 @@ v3d_job_timedout(struct drm_sched_job *sched_job) if (job->timedout_ctca != ctca || job->timedout_ctra != ctra) { job->timedout_ctca = ctca; job->timedout_ctra = ctra; - - schedule_delayed_work(&job->base.sched->work_tdr, - job->base.sched->timeout); return; } diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h index 6474e83cbf3d..2a8aaea72af3 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.h +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h @@ -143,9 +143,6 @@ struct virtio_gpu_fbdev { }; struct virtio_gpu_mman { - struct ttm_bo_global_ref bo_global_ref; - struct drm_global_reference mem_global_ref; - bool mem_global_referenced; struct ttm_bo_device bdev; }; diff --git a/drivers/gpu/drm/virtio/virtgpu_ttm.c b/drivers/gpu/drm/virtio/virtgpu_ttm.c index cd63dffa6d40..4bfbf25fabff 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ttm.c +++ b/drivers/gpu/drm/virtio/virtgpu_ttm.c @@ -50,62 +50,6 @@ virtio_gpu_device *virtio_gpu_get_vgdev(struct ttm_bo_device *bdev) return vgdev; } -static int virtio_gpu_ttm_mem_global_init(struct drm_global_reference *ref) -{ - return ttm_mem_global_init(ref->object); -} - -static void virtio_gpu_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -static int virtio_gpu_ttm_global_init(struct virtio_gpu_device *vgdev) -{ - struct drm_global_reference *global_ref; - int r; - - vgdev->mman.mem_global_referenced = false; - global_ref = &vgdev->mman.mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &virtio_gpu_ttm_mem_global_init; - global_ref->release = &virtio_gpu_ttm_mem_global_release; - - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM memory accounting " - "subsystem.\n"); - return r; - } - - vgdev->mman.bo_global_ref.mem_glob = - vgdev->mman.mem_global_ref.object; - global_ref = &vgdev->mman.bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - r = drm_global_item_ref(global_ref); - if (r != 0) { - DRM_ERROR("Failed setting up TTM BO subsystem.\n"); - drm_global_item_unref(&vgdev->mman.mem_global_ref); - return r; - } - - vgdev->mman.mem_global_referenced = true; - return 0; -} - -static void virtio_gpu_ttm_global_fini(struct virtio_gpu_device *vgdev) -{ - if (vgdev->mman.mem_global_referenced) { - drm_global_item_unref(&vgdev->mman.bo_global_ref.ref); - drm_global_item_unref(&vgdev->mman.mem_global_ref); - vgdev->mman.mem_global_referenced = false; - } -} - int virtio_gpu_mmap(struct file *filp, struct vm_area_struct *vma) { struct drm_file *file_priv; @@ -382,12 +326,8 @@ int virtio_gpu_ttm_init(struct virtio_gpu_device *vgdev) { int r; - r = virtio_gpu_ttm_global_init(vgdev); - if (r) - return r; /* No others user of address space so set it to 0 */ r = ttm_bo_device_init(&vgdev->mman.bdev, - vgdev->mman.bo_global_ref.ref.object, &virtio_gpu_bo_driver, vgdev->ddev->anon_inode->i_mapping, DRM_FILE_PAGE_OFFSET, 0); @@ -406,13 +346,11 @@ int virtio_gpu_ttm_init(struct virtio_gpu_device *vgdev) err_mm_init: ttm_bo_device_release(&vgdev->mman.bdev); err_dev_init: - virtio_gpu_ttm_global_fini(vgdev); return r; } void virtio_gpu_ttm_fini(struct virtio_gpu_device *vgdev) { ttm_bo_device_release(&vgdev->mman.bdev); - virtio_gpu_ttm_global_fini(vgdev); DRM_INFO("virtio_gpu: ttm finalized\n"); } diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index 61a84b958d67..b9c078860a7c 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -801,11 +801,6 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset) DRM_INFO("MMIO at 0x%08x size is %u kiB\n", dev_priv->mmio_start, dev_priv->mmio_size / 1024); - ret = vmw_ttm_global_init(dev_priv); - if (unlikely(ret != 0)) - goto out_err0; - - vmw_master_init(&dev_priv->fbdev_master); ttm_lock_set_kill(&dev_priv->fbdev_master.lock, false, SIGTERM); dev_priv->active_master = &dev_priv->fbdev_master; @@ -816,7 +811,7 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset) if (unlikely(dev_priv->mmio_virt == NULL)) { ret = -ENOMEM; DRM_ERROR("Failed mapping MMIO.\n"); - goto out_err3; + goto out_err0; } /* Need mmio memory to check for fifo pitchlock cap. */ @@ -828,8 +823,8 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset) goto out_err4; } - dev_priv->tdev = ttm_object_device_init - (dev_priv->mem_global_ref.object, 12, &vmw_prime_dmabuf_ops); + dev_priv->tdev = ttm_object_device_init(&ttm_mem_glob, 12, + &vmw_prime_dmabuf_ops); if (unlikely(dev_priv->tdev == NULL)) { DRM_ERROR("Unable to initialize TTM object management.\n"); @@ -870,7 +865,6 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset) } ret = ttm_bo_device_init(&dev_priv->bdev, - dev_priv->bo_global_ref.ref.object, &vmw_bo_driver, dev->anon_inode->i_mapping, VMWGFX_FILE_PAGE_OFFSET, @@ -992,8 +986,6 @@ out_no_device: ttm_object_device_release(&dev_priv->tdev); out_err4: memunmap(dev_priv->mmio_virt); -out_err3: - vmw_ttm_global_release(dev_priv); out_err0: for (i = vmw_res_context; i < vmw_res_max; ++i) idr_destroy(&dev_priv->res_idr[i]); @@ -1045,7 +1037,6 @@ static void vmw_driver_unload(struct drm_device *dev) memunmap(dev_priv->mmio_virt); if (dev_priv->ctx.staged_bindings) vmw_binding_state_free(dev_priv->ctx.staged_bindings); - vmw_ttm_global_release(dev_priv); for (i = vmw_res_context; i < vmw_res_max; ++i) idr_destroy(&dev_priv->res_idr[i]); diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h index 59f614225bcd..28df788da44e 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h @@ -417,8 +417,6 @@ enum { struct vmw_private { struct ttm_bo_device bdev; - struct ttm_bo_global_ref bo_global_ref; - struct drm_global_reference mem_global_ref; struct vmw_fifo_state fifo; @@ -842,8 +840,6 @@ extern int vmw_fifo_flush(struct vmw_private *dev_priv, * TTM glue - vmwgfx_ttm_glue.c */ -extern int vmw_ttm_global_init(struct vmw_private *dev_priv); -extern void vmw_ttm_global_release(struct vmw_private *dev_priv); extern int vmw_mmap(struct file *filp, struct vm_area_struct *vma); /** @@ -1363,7 +1359,7 @@ vmw_bo_reference(struct vmw_buffer_object *buf) static inline struct ttm_mem_global *vmw_mem_glob(struct vmw_private *dev_priv) { - return (struct ttm_mem_global *) dev_priv->mem_global_ref.object; + return &ttm_mem_glob; } static inline void vmw_fifo_resource_inc(struct vmw_private *dev_priv) diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_glue.c b/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_glue.c index 7b1e5a5cbd2c..154eb09aa91e 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_glue.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_glue.c @@ -42,57 +42,3 @@ int vmw_mmap(struct file *filp, struct vm_area_struct *vma) dev_priv = vmw_priv(file_priv->minor->dev); return ttm_bo_mmap(filp, vma, &dev_priv->bdev); } - -static int vmw_ttm_mem_global_init(struct drm_global_reference *ref) -{ - DRM_INFO("global init.\n"); - return ttm_mem_global_init(ref->object); -} - -static void vmw_ttm_mem_global_release(struct drm_global_reference *ref) -{ - ttm_mem_global_release(ref->object); -} - -int vmw_ttm_global_init(struct vmw_private *dev_priv) -{ - struct drm_global_reference *global_ref; - int ret; - - global_ref = &dev_priv->mem_global_ref; - global_ref->global_type = DRM_GLOBAL_TTM_MEM; - global_ref->size = sizeof(struct ttm_mem_global); - global_ref->init = &vmw_ttm_mem_global_init; - global_ref->release = &vmw_ttm_mem_global_release; - - ret = drm_global_item_ref(global_ref); - if (unlikely(ret != 0)) { - DRM_ERROR("Failed setting up TTM memory accounting.\n"); - return ret; - } - - dev_priv->bo_global_ref.mem_glob = - dev_priv->mem_global_ref.object; - global_ref = &dev_priv->bo_global_ref.ref; - global_ref->global_type = DRM_GLOBAL_TTM_BO; - global_ref->size = sizeof(struct ttm_bo_global); - global_ref->init = &ttm_bo_global_init; - global_ref->release = &ttm_bo_global_release; - ret = drm_global_item_ref(global_ref); - - if (unlikely(ret != 0)) { - DRM_ERROR("Failed setting up TTM buffer objects.\n"); - goto out_no_bo; - } - - return 0; -out_no_bo: - drm_global_item_unref(&dev_priv->mem_global_ref); - return ret; -} - -void vmw_ttm_global_release(struct vmw_private *dev_priv) -{ - drm_global_item_unref(&dev_priv->bo_global_ref.ref); - drm_global_item_unref(&dev_priv->mem_global_ref); -} |