Re: [PATCH] drm/xe/migrate: Support copying between sysmem and stolen

[Matthew Auld <matthew.auld@intel.com>]

On 21/07/2026 20:25, Maarten Lankhorst wrote:
> Stolen memory is used on integrated devices to hold the
> firmware framebuffer. In the next commit we want to copy
> the framebuffer to normal system memory.
> 
> This requires treating stolen memory mostly as VRAM, but
> without access to the identity mapping.
> 
> On discrete the identity mapping can be used, but we never
> need to copy there, so just prohibit identity map on stolen.
> 
> Signed-off-by: Maarten Lankhorst <dev@lankhorst.se>
> ---
>   drivers/gpu/drm/xe/xe_migrate.c | 18 +++++++++++++++---
>   1 file changed, 15 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_migrate.c b/drivers/gpu/drm/xe/xe_migrate.c
> index 4366c41bd3254..505d105472fca 100644
> --- a/drivers/gpu/drm/xe/xe_migrate.c
> +++ b/drivers/gpu/drm/xe/xe_migrate.c
> @@ -574,6 +574,10 @@ static u64 xe_migrate_res_sizes(struct xe_migrate *m, struct xe_res_cursor *cur)
>   
>   static bool xe_migrate_allow_identity(u64 size, const struct xe_res_cursor *cur)
>   {
> +	/* Stolen only used by integrated, no VRAM or identity map there */
> +	if (cur->mem_type == XE_PL_STOLEN)
> +		return false;
> +
>   	/* If the chunk is not fragmented, allow identity map. */
>   	return cur->size >= size;
>   }
> @@ -879,6 +883,14 @@ static u32 xe_migrate_ccs_copy(struct xe_migrate *m,
>   	return flush_flags;
>   }
>   
> +static bool is_devmem(u32 mem_type)
> +{
> +	if (mem_type_is_vram(mem_type) || mem_type == XE_PL_STOLEN)
> +		return true;
> +
> +	return false;
> +}

Do we need to worry about XE_PPGTT_PTE_DM in xelp_pte_encode_addr() ? I 
don't think that bit is used for stolen until recent platforms? We seem 
to guard against this in xelp_pte_encode_bo() with xe_bo_is_stolen_devmem().

> +
>   static struct dma_fence *__xe_migrate_copy(struct xe_migrate *m,
>   					   struct xe_bo *src_bo,
>   					   struct xe_bo *dst_bo,
> @@ -899,8 +911,8 @@ static struct dma_fence *__xe_migrate_copy(struct xe_migrate *m,
>   	int err;
>   	bool src_is_pltt = src->mem_type == XE_PL_TT;
>   	bool dst_is_pltt = dst->mem_type == XE_PL_TT;
> -	bool src_is_vram = mem_type_is_vram(src->mem_type);
> -	bool dst_is_vram = mem_type_is_vram(dst->mem_type);
> +	bool src_is_vram = is_devmem(src->mem_type);
> +	bool dst_is_vram = is_devmem(dst->mem_type);
>   	bool type_device = src_bo->ttm.type == ttm_bo_type_device;
>   	bool needs_ccs_emit = type_device && xe_migrate_needs_ccs_emit(xe);
>   	bool copy_ccs = xe_device_has_flat_ccs(xe) &&
> @@ -1597,7 +1609,7 @@ struct dma_fence *xe_migrate_clear(struct xe_migrate *m,
>   				   struct ttm_resource *dst,
>   				   u32 clear_flags)
>   {
> -	bool clear_vram = mem_type_is_vram(dst->mem_type);
> +	bool clear_vram = is_devmem(dst->mem_type);

Do we actually trigger this path for stolen? Or is this just for 
symmetry with copy?

Also, I guess overall idea here is to call copy() directly, and not mess 
with this at the bo_move()/validate() level?

>   	bool clear_bo_data = XE_MIGRATE_CLEAR_FLAG_BO_DATA & clear_flags;
>   	bool clear_ccs = XE_MIGRATE_CLEAR_FLAG_CCS_DATA & clear_flags;
>   	struct xe_gt *gt = m->tile->primary_gt;