Re: [PATCH v6 1/3] drm/xe: Bump the LRU on a successful page fault

["Thomas Hellström" <thomas.hellstrom@linux.intel.com>]

On Tue, 2026-10-27 at 15:49 -0700, Matthew Brost wrote:
> Servicing a page fault validates the VMA's backing store and binds
> it,
> making the BO part of the VM's active working set, but unlike exec it
> never moves anything on the TTM LRU. In fault mode, where a VM may
> run
> for a long time without a single exec, the BOs the GPU actually
> touches
> therefore drift towards the LRU head and are picked by eviction and
> by
> the shrinker ahead of genuinely cold BOs.
> 
> Bump the BO's LRU on a successful page fault.
> 
> Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel
> GPUs")
> Signed-off-by: Matthew Brost <matthew.brost@intel.com>

Found the newer version:
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>


> ---
>  drivers/gpu/drm/xe/xe_pagefault.c | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/drivers/gpu/drm/xe/xe_pagefault.c
> b/drivers/gpu/drm/xe/xe_pagefault.c
> index 5247321c805e..6828f81df50e 100644
> --- a/drivers/gpu/drm/xe/xe_pagefault.c
> +++ b/drivers/gpu/drm/xe/xe_pagefault.c
> @@ -201,6 +201,8 @@ static int xe_pagefault_handle_vma(struct xe_gt
> *gt, struct xe_vma *vma,
>  		xe_validation_ctx_init(&ctx, &vm->xe->val, &exec,
>  				       (struct xe_val_flags) {});
>  		drm_exec_until_all_locked(&exec) {
> +			struct xe_bo *bo = xe_vma_bo(vma);
> +
>  			err = xe_pagefault_begin(&exec, vma, tile-
> >mem.vram,
>  						 needs_vram == 1);
>  			drm_exec_retry_on_contention(&exec);
> @@ -221,6 +223,11 @@ static int xe_pagefault_handle_vma(struct xe_gt
> *gt, struct xe_vma *vma,
>  				xe_pagefault_set_error(pf,
> XE_PAGEFAULT_ERROR_VMA_REBIND);
>  				break;
>  			}
> +
> +			if (bo) {
> +				scoped_guard(spinlock, &vm->xe-
> >ttm.lru_lock)
> +					ttm_bo_move_to_lru_tail(&bo-
> >ttm);
> +			}
>  		}
>  		xe_validation_ctx_fini(&ctx);
>  	} while (err == -EAGAIN);