Re: [PATCH 1/2] drm/xe/svm: Ensure data will be migrated to system if indicated by madvise.

[Matthew Brost <matthew.brost@intel.com>]

On Wed, Oct 08, 2025 at 01:10:49PM +0200, Thomas Hellström wrote:
> If the location madvise() is set to
> DRM_XE_PREFERRED_LOC_DEFAULT_SYSTEM, the drm_pagemap in the
> SVM gpu fault handler will be set to NULL. However there is nothing
> that explicitly migrates the data to system if it is already present
> in device memory.
> 
> In that case, set the device memory owner to NULL to ensure
> data gets properly migrated to system on page-fault.
> 
> Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
> ---
>  drivers/gpu/drm/xe/xe_svm.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/gpu/drm/xe/xe_svm.c b/drivers/gpu/drm/xe/xe_svm.c
> index 7e2db71ff34e..fd906eb03d71 100644
> --- a/drivers/gpu/drm/xe/xe_svm.c
> +++ b/drivers/gpu/drm/xe/xe_svm.c
> @@ -1034,6 +1034,9 @@ static int __xe_svm_handle_pagefault(struct xe_vm *vm, struct xe_vma *vma,
>  	if (err)
>  		return err;
>  
> +	dpagemap = xe_vma_resolve_pagemap(vma, tile);
> +	if (!dpagemap && !ctx.devmem_only)
> +		ctx.device_private_page_owner = NULL;

I think xe_svm_range_is_valid below can still make the page fault
handler bail out if the memory is in valid in VRAM. That a really odd
case but maybe it needs to be handled?

What certainly needs to be handled in xe_svm_range_is_valid is if system
is the preferred placement, the range has mappings, and is in system. If
we get page fault storms on system memory address every fault will issue
a rebind.

Matt

>  	range = xe_svm_range_find_or_insert(vm, fault_addr, vma, &ctx);
>  
>  	if (IS_ERR(range))
> -- 
> 2.51.0
>