[PATCH v2 13/32] drm/xe/svm: Incase of atomic access ensure get_pages happens from vram

[Matthew Brost]

On Mon, Apr 07, 2025 at 03:47:00PM +0530, Himal Prasad Ghimiray wrote:
> Ranges can be invalidated in between vram allocation and get_pages,
> ensure the dma mapping is happening from vram only incase of atomic
> access. Retry 3 times before calling out fault in case of concurrent
> cpu/gpu access.
> 

Again I pulled this patch into a series which will minimally enable
atomics per UMD request. See the version of the patch [1] I landed on -
that is basically my review feedback. I took ownership but left SoB by
you as it based on this patch. We will need another reviewer though as
we are both contributors but feel to comment there.

Matt

[1] https://patchwork.freedesktop.org/patch/649010/?series=147846&rev=2

> Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray at intel.com>
> ---
>  drivers/gpu/drm/xe/xe_svm.c | 43 ++++++++++++++++++++++++-------------
>  1 file changed, 28 insertions(+), 15 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_svm.c b/drivers/gpu/drm/xe/xe_svm.c
> index f4ae3feaf9d3..7ec7ecd7eb1f 100644
> --- a/drivers/gpu/drm/xe/xe_svm.c
> +++ b/drivers/gpu/drm/xe/xe_svm.c
> @@ -778,11 +778,13 @@ int xe_svm_handle_pagefault(struct xe_vm *vm, struct xe_vma *vma,
>  			IS_ENABLED(CONFIG_DRM_XE_DEVMEM_MIRROR),
>  		.check_pages_threshold = IS_DGFX(vm->xe) &&
>  			IS_ENABLED(CONFIG_DRM_XE_DEVMEM_MIRROR) ? SZ_64K : 0,
> +		.vram_only = 0,
>  	};
>  	struct xe_svm_range *range;
>  	struct drm_exec exec;
>  	struct dma_fence *fence;
>  	struct xe_tile *tile = gt_to_tile(gt);
> +	int retry_count = 3;
>  	ktime_t end = 0;
>  	int err;
>  
> @@ -792,6 +794,7 @@ int xe_svm_handle_pagefault(struct xe_vm *vm, struct xe_vma *vma,
>  	xe_gt_stats_incr(gt, XE_GT_STATS_ID_SVM_PAGEFAULT_COUNT, 1);
>  
>  retry:
> +	retry_count--;
>  	/* Always process UNMAPs first so view SVM ranges is current */
>  	err = xe_svm_garbage_collector(vm);
>  	if (err)
> @@ -807,30 +810,40 @@ int xe_svm_handle_pagefault(struct xe_vm *vm, struct xe_vma *vma,
>  
>  	range_debug(range, "PAGE FAULT");
>  
> -	/* XXX: Add migration policy, for now migrate range once */
> -	if (!range->skip_migrate &&
> -	    xe_svm_range_needs_migrate_to_vram(range, vma, IS_DGFX(vm->xe))) {
> -		range->skip_migrate = true;
> -
> +	if (xe_svm_range_needs_migrate_to_vram(range, vma, IS_DGFX(vm->xe))) {
>  		err = xe_svm_alloc_vram(vm, tile, range, &ctx);
>  		if (err) {
> -			drm_dbg(&vm->xe->drm,
> -				"VRAM allocation failed, falling back to "
> -				"retrying fault, asid=%u, errno=%pe\n",
> -				vm->usm.asid, ERR_PTR(err));
> -			goto retry;
> +			if (retry_count) {
> +				drm_dbg(&vm->xe->drm,
> +					"VRAM allocation failed, falling back to retrying fault, asid=%u, errno=%pe\n",
> +					vm->usm.asid, ERR_PTR(err));
> +				goto retry;
> +			} else {
> +				drm_err(&vm->xe->drm,
> +					"VRAM allocation failed, retry count exceeded, asid=%u, errno=%pe\n",
> +					vm->usm.asid, ERR_PTR(err));
> +				return err;
> +			}
>  		}
> +
>  	}
>  
> +	if (atomic)
> +		ctx.vram_only = 1;
> +
>  	range_debug(range, "GET PAGES");
>  	err = xe_svm_range_get_pages(vm, range, &ctx);
>  	/* Corner where CPU mappings have changed */
>  	if (err == -EOPNOTSUPP || err == -EFAULT || err == -EPERM) {
> -		drm_dbg(&vm->xe->drm,
> -			"Get pages failed, falling back to retrying, asid=%u, gpusvm=%p, errno=%pe\n",
> -			vm->usm.asid, &vm->svm.gpusvm, ERR_PTR(err));
> -		range_debug(range, "PAGE FAULT - RETRY PAGES");
> -		goto retry;
> +		if (retry_count) {
> +			drm_dbg(&vm->xe->drm, "Get pages failed, falling back to retrying, asid=%u, gpusvm=%p, errno=%pe\n",
> +				vm->usm.asid, &vm->svm.gpusvm, ERR_PTR(err));
> +			range_debug(range, "PAGE FAULT - RETRY PAGES");
> +			goto retry;
> +		} else {
> +			drm_err(&vm->xe->drm, "Get pages failed,, retry count exceeded, asid=%u,, errno=%pe\n",
> +				vm->usm.asid, ERR_PTR(err));
> +		}
>  	}
>  	if (err) {
>  		range_debug(range, "PAGE FAULT - FAIL PAGE COLLECT");
> -- 
> 2.34.1
>