[PATCH v3] drm/amdgpu: reset vm state machine after gpu reset(vram lost)

[Christian König]

Am 24.07.24 um 05:00 schrieb ZhenGuo Yin:
> [Why]
> Page table of compute VM in the VRAM will lost after gpu reset.
> VRAM won't be restored since compute VM has no shadows.
>
> [How]
> Use higher 32-bit of vm->generation to record a vram_lost_counter.
> Reset the VM state machine when vm->genertaion is not equal to
> the new generation token.
>
> v2: Check vm->generation instead of calling drm_sched_entity_error
> in amdgpu_vm_validate.
> v3: Use new generation token instead of vram_lost_counter for check.
>
> Signed-off-by: ZhenGuo Yin <zhenguo.yin at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 +++++----
>   1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index 3abfa66d72a2..6c6170f0f318 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -434,7 +434,7 @@ uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm)
>   	if (!vm)
>   		return result;
>   
> -	result += vm->generation;
> +	result += (vm->generation & 0xFFFFFFFFULL);

Please use the lower_32_bits() macro here.

With that fixed the patch is Reviewed-by: Christian König 
<christian.koenig at amd.com>

Thanks and sorry that I didn't initially got what the actual problem 
here is,
Christian.

>   	/* Add one if the page tables will be re-generated on next CS */
>   	if (drm_sched_entity_error(&vm->delayed))
>   		++result;
> @@ -463,13 +463,14 @@ int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
>   		       int (*validate)(void *p, struct amdgpu_bo *bo),
>   		       void *param)
>   {
> +	uint64_t new_vm_generation = amdgpu_vm_generation(adev, vm);
>   	struct amdgpu_vm_bo_base *bo_base;
>   	struct amdgpu_bo *shadow;
>   	struct amdgpu_bo *bo;
>   	int r;
>   
> -	if (drm_sched_entity_error(&vm->delayed)) {
> -		++vm->generation;
> +	if (vm->generation != new_vm_generation) {
> +		vm->generation = new_vm_generation;
>   		amdgpu_vm_bo_reset_state_machine(vm);
>   		amdgpu_vm_fini_entities(vm);
>   		r = amdgpu_vm_init_entities(adev, vm);
> @@ -2439,7 +2440,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
>   	vm->last_update = dma_fence_get_stub();
>   	vm->last_unlocked = dma_fence_get_stub();
>   	vm->last_tlb_flush = dma_fence_get_stub();
> -	vm->generation = 0;
> +	vm->generation = amdgpu_vm_generation(adev, NULL);
>   
>   	mutex_init(&vm->eviction_lock);
>   	vm->evicting = false;