diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 56 |
1 files changed, 50 insertions, 6 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 047ec1930d12..046949c4b695 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -94,8 +94,11 @@ struct amdgpu_mem_stats; #define AMDGPU_VM_NORETRY_FLAGS_TF (AMDGPU_PTE_VALID | AMDGPU_PTE_SYSTEM | \ AMDGPU_PTE_PRT) /* For GFX9 */ -#define AMDGPU_PTE_MTYPE_VG10(a) ((uint64_t)(a) << 57) -#define AMDGPU_PTE_MTYPE_VG10_MASK AMDGPU_PTE_MTYPE_VG10(3ULL) +#define AMDGPU_PTE_MTYPE_VG10_SHIFT(mtype) ((uint64_t)(mtype) << 57) +#define AMDGPU_PTE_MTYPE_VG10_MASK AMDGPU_PTE_MTYPE_VG10_SHIFT(3ULL) +#define AMDGPU_PTE_MTYPE_VG10(flags, mtype) \ + (((uint64_t)(flags) & (~AMDGPU_PTE_MTYPE_VG10_MASK)) | \ + AMDGPU_PTE_MTYPE_VG10_SHIFT(mtype)) #define AMDGPU_MTYPE_NC 0 #define AMDGPU_MTYPE_CC 2 @@ -108,8 +111,34 @@ struct amdgpu_mem_stats; | AMDGPU_PTE_MTYPE_VG10(AMDGPU_MTYPE_CC)) /* gfx10 */ -#define AMDGPU_PTE_MTYPE_NV10(a) ((uint64_t)(a) << 48) -#define AMDGPU_PTE_MTYPE_NV10_MASK AMDGPU_PTE_MTYPE_NV10(7ULL) +#define AMDGPU_PTE_MTYPE_NV10_SHIFT(mtype) ((uint64_t)(mtype) << 48) +#define AMDGPU_PTE_MTYPE_NV10_MASK AMDGPU_PTE_MTYPE_NV10_SHIFT(7ULL) +#define AMDGPU_PTE_MTYPE_NV10(flags, mtype) \ + (((uint64_t)(flags) & (~AMDGPU_PTE_MTYPE_NV10_MASK)) | \ + AMDGPU_PTE_MTYPE_NV10_SHIFT(mtype)) + +/* gfx12 */ +#define AMDGPU_PTE_PRT_GFX12 (1ULL << 56) +#define AMDGPU_PTE_PRT_FLAG(adev) \ + ((amdgpu_ip_version((adev), GC_HWIP, 0) >= IP_VERSION(12, 0, 0)) ? AMDGPU_PTE_PRT_GFX12 : AMDGPU_PTE_PRT) + +#define AMDGPU_PTE_MTYPE_GFX12_SHIFT(mtype) ((uint64_t)(mtype) << 54) +#define AMDGPU_PTE_MTYPE_GFX12_MASK AMDGPU_PTE_MTYPE_GFX12_SHIFT(3ULL) +#define AMDGPU_PTE_MTYPE_GFX12(flags, mtype) \ + (((uint64_t)(flags) & (~AMDGPU_PTE_MTYPE_GFX12_MASK)) | \ + AMDGPU_PTE_MTYPE_GFX12_SHIFT(mtype)) + +#define AMDGPU_PTE_DCC (1ULL << 58) +#define AMDGPU_PTE_IS_PTE (1ULL << 63) + +/* PDE Block Fragment Size for gfx v12 */ +#define AMDGPU_PDE_BFS_GFX12(a) ((uint64_t)((a) & 0x1fULL) << 58) +#define AMDGPU_PDE_BFS_FLAG(adev, a) \ + ((amdgpu_ip_version((adev), GC_HWIP, 0) >= IP_VERSION(12, 0, 0)) ? AMDGPU_PDE_BFS_GFX12(a) : AMDGPU_PDE_BFS(a)) +/* PDE is handled as PTE for gfx v12 */ +#define AMDGPU_PDE_PTE_GFX12 (1ULL << 63) +#define AMDGPU_PDE_PTE_FLAG(adev) \ + ((amdgpu_ip_version((adev), GC_HWIP, 0) >= IP_VERSION(12, 0, 0)) ? AMDGPU_PDE_PTE_GFX12 : AMDGPU_PDE_PTE) /* How to program VM fault handling */ #define AMDGPU_VM_FAULT_STOP_NEVER 0 @@ -257,15 +286,20 @@ struct amdgpu_vm_update_params { unsigned int num_dw_left; /** - * @table_freed: return true if page table is freed when updating + * @needs_flush: true whenever we need to invalidate the TLB */ - bool table_freed; + bool needs_flush; /** * @allow_override: true for memory that is not uncached: allows MTYPE * to be overridden for NUMA local memory. */ bool allow_override; + + /** + * @tlb_flush_waitlist: temporary storage for BOs until tlb_flush + */ + struct list_head tlb_flush_waitlist; }; struct amdgpu_vm_update_funcs { @@ -342,6 +376,7 @@ struct amdgpu_vm { atomic64_t tlb_seq; struct dma_fence *last_tlb_flush; atomic64_t kfd_last_flushed_seq; + uint64_t tlb_fence_context; /* How many times we had to re-generate the page tables */ uint64_t generation; @@ -422,6 +457,8 @@ struct amdgpu_vm_manager { * look up VM of a page fault */ struct xarray pasids; + /* Global registration of recent page fault information */ + struct amdgpu_vm_fault_info fault_info; }; struct amdgpu_bo_va_mapping; @@ -544,6 +581,8 @@ int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params, uint64_t start, uint64_t end, uint64_t dst, uint64_t flags); void amdgpu_vm_pt_free_work(struct work_struct *work); +void amdgpu_vm_pt_free_list(struct amdgpu_device *adev, + struct amdgpu_vm_update_params *params); #if defined(CONFIG_DEBUG_FS) void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m); @@ -551,6 +590,8 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m); int amdgpu_vm_pt_map_tables(struct amdgpu_device *adev, struct amdgpu_vm *vm); +bool amdgpu_vm_is_bo_always_valid(struct amdgpu_vm *vm, struct amdgpu_bo *bo); + /** * amdgpu_vm_tlb_seq - return tlb flush sequence number * @vm: the amdgpu_vm structure to query @@ -609,5 +650,8 @@ void amdgpu_vm_update_fault_cache(struct amdgpu_device *adev, uint64_t addr, uint32_t status, unsigned int vmhub); +void amdgpu_vm_tlb_fence_create(struct amdgpu_device *adev, + struct amdgpu_vm *vm, + struct dma_fence **fence); #endif |