Re: [RFC 2/7] drm/ttm: Add NUMA node aware restore helpers

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] drm/ttm: unused helper and EXPORT_SYMBOL() on ttm_pool_restore_and_alloc()
--

commit 8d0a6d62696c67cc37673b025155aa37dad8e6a7
Author: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>

drm/ttm: Add NUMA node aware restore helpers

This commit introduces ttm_tt_restore_node() and
ttm_pool_restore_and_alloc_node() so callers can preserve a specific NUMA
node placement when restoring backed-up TTM pages.

> diff --git a/drivers/gpu/drm/ttm/ttm_pool.c b/drivers/gpu/drm/ttm/ttm_pool.c
> index 54d2f6b5c229c..2107dc6c69b36 100644
> --- a/drivers/gpu/drm/ttm/ttm_pool.c
> +++ b/drivers/gpu/drm/ttm/ttm_pool.c
> @@ -962,6 +962,30 @@ EXPORT_SYMBOL(ttm_pool_alloc_node);
>   */
>  int ttm_pool_restore_and_alloc(struct ttm_pool *pool, struct ttm_tt *tt,
>  			       const struct ttm_operation_ctx *ctx)
> +{
> +	return ttm_pool_restore_and_alloc_node(pool, tt, ctx, NUMA_NO_NODE);
> +}
> +EXPORT_SYMBOL(ttm_pool_restore_and_alloc);

[Severity: Low]
Is ttm_pool_restore_and_alloc() still needed now that ttm_tt_restore() in
drivers/gpu/drm/ttm/ttm_tt.c delegates to ttm_tt_restore_node(), which
calls ttm_pool_restore_and_alloc_node() directly?

Previously, ttm_pool_restore_and_alloc() was an unexported TTM-internal
helper whose only caller was ttm_tt_restore(), and external drivers cannot
call it directly because only ttm_tt_restore() and ttm_tt_restore_node()
clear TTM_TT_FLAG_BACKED_UP on tt->page_flags. With ttm_tt_restore()
switched over to ttm_tt_restore_node(), ttm_pool_restore_and_alloc() has no
remaining callers in the tree, yet it is kept and newly exported with
EXPORT_SYMBOL().

Should ttm_pool_restore_and_alloc() and its EXPORT_SYMBOL() be dropped (or
kept unexported if there is a planned internal user)?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261102101803.455921-9-himal.prasad.ghimiray@intel.com?part=2