Commit 6716a823 authored by Christian König's avatar Christian König Committed by Alex Deucher
Browse files

drm/amdgpu: rework how PTE flags are generated v3 



Previously we tried to keep the HW specific PTE flags in each mapping,
but for CRIU that isn't sufficient any more since the original value is
needed for the checkpoint procedure.

So rework the whole handling, nuke the early mapping function, keep the
UAPI flags in each mapping instead of the HW flags and translate them to
the HW flags while filling in the PTEs.

Only tested on Navi 23 for now, so probably needs quite a bit of more
work.

v2: fix KFD and SVN handling
v3: one more SVN fix pointed out by Felix
v4: squash in gfx12 fix from David

Signed-off-by: default avatarChristian König <christian.koenig@amd.com>
Reviewed-by: default avatarFelix Kuehling <felix.kuehling@amd.com>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
parent 9f1f7cd4

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c

+4 −3
Original line number Diff line number Diff line
@@ -494,7 +494,8 @@ static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync) 
	return amdgpu_sync_fence(sync, vm->last_update, GFP_KERNEL);

}



static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem)

static uint64_t get_pte_flags(struct amdgpu_device *adev, struct amdgpu_vm *vm,

			      struct kgd_mem *mem)

{

	uint32_t mapping_flags = AMDGPU_VM_PAGE_READABLE |

				 AMDGPU_VM_MTYPE_DEFAULT;
@@ -504,7 +505,7 @@ static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem) 
	if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE)

		mapping_flags |= AMDGPU_VM_PAGE_EXECUTABLE;



	return amdgpu_gem_va_map_flags(adev, mapping_flags);

	return mapping_flags;

}



/**
@@ -961,7 +962,7 @@ static int kfd_mem_attach(struct amdgpu_device *adev, struct kgd_mem *mem, 
			goto unwind;

		}

		attachment[i]->va = va;

		attachment[i]->pte_flags = get_pte_flags(adev, mem);

		attachment[i]->pte_flags = get_pte_flags(adev, vm, mem);

		attachment[i]->adev = adev;

		list_add(&attachment[i]->list, &mem->attachments);

drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c

+2 −35
Original line number Diff line number Diff line
@@ -790,36 +790,6 @@ amdgpu_gem_va_update_vm(struct amdgpu_device *adev, 
	return fence;

}



/**

 * amdgpu_gem_va_map_flags - map GEM UAPI flags into hardware flags

 *

 * @adev: amdgpu_device pointer

 * @flags: GEM UAPI flags

 *

 * Returns the GEM UAPI flags mapped into hardware for the ASIC.

 */

uint64_t amdgpu_gem_va_map_flags(struct amdgpu_device *adev, uint32_t flags)

{

	uint64_t pte_flag = 0;



	if (flags & AMDGPU_VM_PAGE_EXECUTABLE)

		pte_flag |= AMDGPU_PTE_EXECUTABLE;

	if (flags & AMDGPU_VM_PAGE_READABLE)

		pte_flag |= AMDGPU_PTE_READABLE;

	if (flags & AMDGPU_VM_PAGE_WRITEABLE)

		pte_flag |= AMDGPU_PTE_WRITEABLE;

	if (flags & AMDGPU_VM_PAGE_PRT)

		pte_flag |= AMDGPU_PTE_PRT_FLAG(adev);

	if (flags & AMDGPU_VM_PAGE_NOALLOC)

		pte_flag |= AMDGPU_PTE_NOALLOC;



	if (adev->gmc.gmc_funcs->map_mtype)

		pte_flag |= amdgpu_gmc_map_mtype(adev,

						 flags & AMDGPU_VM_MTYPE_MASK);



	return pte_flag;

}



int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,

			  struct drm_file *filp)

{
@@ -840,7 +810,6 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data, 
	struct dma_fence_chain *timeline_chain = NULL;

	struct dma_fence *fence;

	struct drm_exec exec;

	uint64_t va_flags;

	uint64_t vm_size;

	int r = 0;
@@ -944,10 +913,9 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data, 


	switch (args->operation) {

	case AMDGPU_VA_OP_MAP:

		va_flags = amdgpu_gem_va_map_flags(adev, args->flags);

		r = amdgpu_vm_bo_map(adev, bo_va, args->va_address,

				     args->offset_in_bo, args->map_size,

				     va_flags);

				     args->flags);

		break;

	case AMDGPU_VA_OP_UNMAP:

		r = amdgpu_vm_bo_unmap(adev, bo_va, args->va_address);
@@ -959,10 +927,9 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data, 
						args->map_size);

		break;

	case AMDGPU_VA_OP_REPLACE:

		va_flags = amdgpu_gem_va_map_flags(adev, args->flags);

		r = amdgpu_vm_bo_replace_map(adev, bo_va, args->va_address,

					     args->offset_in_bo, args->map_size,

					     va_flags);

					     args->flags);

		break;

	default:

		break;

drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h

+0 −1
Original line number Diff line number Diff line
@@ -63,7 +63,6 @@ int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data, 
			  struct drm_file *filp);

int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,

			      struct drm_file *filp);

uint64_t amdgpu_gem_va_map_flags(struct amdgpu_device *adev, uint32_t flags);

int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,

			  struct drm_file *filp);

int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,

drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h

+8 −7
Original line number Diff line number Diff line
@@ -154,15 +154,15 @@ struct amdgpu_gmc_funcs { 
				   unsigned pasid);

	/* enable/disable PRT support */

	void (*set_prt)(struct amdgpu_device *adev, bool enable);

	/* map mtype to hardware flags */

	uint64_t (*map_mtype)(struct amdgpu_device *adev, uint32_t flags);

	/* get the pde for a given mc addr */

	void (*get_vm_pde)(struct amdgpu_device *adev, int level,

			   u64 *dst, u64 *flags);

	/* get the pte flags to use for a BO VA mapping */

	/* get the pte flags to use for PTEs */

	void (*get_vm_pte)(struct amdgpu_device *adev,

			   struct amdgpu_bo_va_mapping *mapping,

			   uint64_t *flags);

			   struct amdgpu_vm *vm,

			   struct amdgpu_bo *bo,

			   uint32_t vm_flags,

			   uint64_t *pte_flags);

	/* override per-page pte flags */

	void (*override_vm_pte_flags)(struct amdgpu_device *dev,

				      struct amdgpu_vm *vm,
@@ -356,9 +356,10 @@ struct amdgpu_gmc { 


#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_map_mtype(adev, flags) (adev)->gmc.gmc_funcs->map_mtype((adev),(flags))

#define amdgpu_gmc_get_vm_pde(adev, level, dst, flags) (adev)->gmc.gmc_funcs->get_vm_pde((adev), (level), (dst), (flags))

#define amdgpu_gmc_get_vm_pte(adev, mapping, flags) (adev)->gmc.gmc_funcs->get_vm_pte((adev), (mapping), (flags))

#define amdgpu_gmc_get_vm_pte(adev, vm, bo, vm_flags, pte_flags) \

	((adev)->gmc.gmc_funcs->get_vm_pte((adev), (vm), (bo), (vm_flags), \

					   (pte_flags)))

#define amdgpu_gmc_override_vm_pte_flags(adev, vm, addr, pte_flags)	\

	(adev)->gmc.gmc_funcs->override_vm_pte_flags			\

		((adev), (vm), (addr), (pte_flags))

drivers/gpu/drm/amd/amdgpu/amdgpu_object.h

+1 −1
Original line number Diff line number Diff line
@@ -69,7 +69,7 @@ struct amdgpu_bo_va_mapping { 
	uint64_t			last;

	uint64_t			__subtree_last;

	uint64_t			offset;

	uint64_t			flags;

	uint32_t			flags;

};



/* User space allocated BO in a VM */