[Intel-xe] [PATCH] drm/xe: skip rebind_list if vma destroyed

[Rodrigo Vivi]

On Tue, Aug 08, 2023 at 10:12:09AM +0100, Matthew Auld wrote:
> If we are closing a vm, mark each vma as XE_VMA_DESTROYED and skip
> touching the rebind_list if this is seen on the eviction path. That way
> we can safely drop the vm dma-resv lock on the close path without
> needing to worry about racing with the eviction path trying to add stuff
> to the rebind_list which can corrupt our contended list, since the
> destroy and rebind links are the same list entry underneath.
> 
> References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/514
> Signed-off-by: Matthew Auld <matthew.auld at intel.com>
> Cc: Matthew Brost <matthew.brost at intel.com>

it looks right to me. Although it would still be good to take
Matthew's eyes here,

Reviewed-by: Rodrigo Vivi <rodrigo.vivi at intel.com>

> ---
>  drivers/gpu/drm/xe/xe_bo.c | 10 ++++++----
>  drivers/gpu/drm/xe/xe_vm.c |  1 +
>  2 files changed, 7 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c
> index 284c86107a5f..6a7dc14feab0 100644
> --- a/drivers/gpu/drm/xe/xe_bo.c
> +++ b/drivers/gpu/drm/xe/xe_bo.c
> @@ -504,15 +504,17 @@ static int xe_bo_trigger_rebind(struct xe_device *xe, struct xe_bo *bo,
>  				vm_resv_locked = true;
>  			else if (ctx->resv != &vm->resv) {
>  				spin_lock(&vm->notifier.list_lock);
> -				list_move_tail(&vma->notifier.rebind_link,
> -					       &vm->notifier.rebind_list);
> +				if (!(vma->gpuva.flags & XE_VMA_DESTROYED))
> +					list_move_tail(&vma->notifier.rebind_link,
> +						       &vm->notifier.rebind_list);
>  				spin_unlock(&vm->notifier.list_lock);
>  				continue;
>  			}
>  
>  			xe_vm_assert_held(vm);
> -			if (list_empty(&vma->combined_links.rebind) &&
> -			    vma->tile_present)
> +			if (vma->tile_present &&
> +			    !(vma->gpuva.flags & XE_VMA_DESTROYED) &&
> +			    list_empty(&vma->combined_links.rebind))
>  				list_add_tail(&vma->combined_links.rebind,
>  					      &vm->rebind_list);
>  
> diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
> index 936492915ecb..aef032ba8e27 100644
> --- a/drivers/gpu/drm/xe/xe_vm.c
> +++ b/drivers/gpu/drm/xe/xe_vm.c
> @@ -1450,6 +1450,7 @@ void xe_vm_close_and_put(struct xe_vm *vm)
>  		}
>  
>  		list_move_tail(&vma->combined_links.destroy, &contested);
> +		vma->gpuva.flags |= XE_VMA_DESTROYED;
>  	}
>  
>  	/*
> -- 
> 2.41.0
>