[PATCH v4 03/20] drm/xe/vm: Add attributes struct as member of vma

[Matthew Brost]

On Fri, Jun 13, 2025 at 06:25:41PM +0530, Himal Prasad Ghimiray wrote:
> The attribute of xe_vma will determine the migration policy and the
> encoding of the page table entries (PTEs) for that vma.
> This attribute helps manage how memory pages are moved and how their
> addresses are translated. It will be used by madvise to set the
> behavior of the vma.
> 
> v2 (Matthew Brost)
> - Add docs
> 
> Cc: Matthew Brost <matthew.brost at intel.com>
> Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray at intel.com>
> ---
>  drivers/gpu/drm/xe/xe_vm_types.h | 28 ++++++++++++++++++++++++++++
>  1 file changed, 28 insertions(+)
> 
> diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h
> index bed6088e1bb3..c7663058a7e3 100644
> --- a/drivers/gpu/drm/xe/xe_vm_types.h
> +++ b/drivers/gpu/drm/xe/xe_vm_types.h
> @@ -77,6 +77,27 @@ struct xe_userptr {
>  #endif
>  };
>  
> +/**
> + * struct xe_vma_mem_attr - memory attributes associated with vma
> + */
> +struct xe_vma_mem_attr {
> +	/** @preferred_loc: perferred memory_location*/
> +	struct {
> +		/** @preferred_loc.migration_policy: Policy for migration of pages */
> +		u32 migration_policy;
> +
> +		/**
> +		 * @preferred_loc.devmem_fd: used for determining pagemap_fd requested by user
> +		 * DRM_XE_PREFERRED_LOC_DEFAULT_SYSTEM and DRM_XE_PREFERRED_LOC_DEFAULT_DEVICE mean
> +		 * system memory or closest device memory respectively.

Nit, I generally prefer 80 character line wraps. I know checkpatch
allows 100 but in general I think Xe generally wraps at 80 unless it is
an odd case with nested code and an 80 character wrap makes the code
unreadable.

80 generally allows 2 windows split vertically in vi on a typical laptop
which makes navigating the code a bit easier.

> +		 */
> +		u32 devmem_fd;
> +	} preferred_loc;
> +
> +	/** @atomic_access: The atomic access type for the vma */

I'd reference the uAPI here. e.g., DRM_XE_VMA_ATOMIC_*

With the minor nits fixed:
Reviewed-by: Matthew Brost <matthew.brost at intel.com>

Matt

> +	u32 atomic_access;
> +};
> +
>  struct xe_vma {
>  	/** @gpuva: Base GPUVA object */
>  	struct drm_gpuva gpuva;
> @@ -135,6 +156,13 @@ struct xe_vma {
>  	 * Needs to be signalled before UNMAP can be processed.
>  	 */
>  	struct xe_user_fence *ufence;
> +
> +	/**
> +	 * @attr: The attributes of vma which determines the migration policy
> +	 * and encoding of the PTEs for this vma.
> +	 */
> +	struct xe_vma_mem_attr attr;
> +
>  };
>  
>  /**
> -- 
> 2.34.1
>