[PATCH v3 4/6] drm/pagemap: Allocate folios when possible

[Matthew Brost]

On Wed, Jul 30, 2025 at 04:22:49PM +0200, Francois Dugast wrote:
> If the order is greater than zero, allocate a folio when populating the
> RAM PFNs instead of allocating individual pages one after the other. For
> example if 2MB folios are used instead of 4KB pages, this reduces the
> number of calls to the allocation API by 512.
> 
> v2:
> - Use page order instead of extra argument (Matthew Brost)
> - Allocate with folio_alloc() (Matthew Brost)
> - Loop for mpages and free_pages based on order (Matthew Brost)
> 
> v3:
> - Fix loops in drm_pagemap_migrate_populate_ram_pfn() (Matthew Brost)
> 
> Signed-off-by: Francois Dugast <francois.dugast at intel.com>
> Cc: Matthew Brost <matthew.brost at intel.com>
> ---
>  drivers/gpu/drm/drm_pagemap.c | 50 ++++++++++++++++++++++++++---------
>  1 file changed, 38 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_pagemap.c b/drivers/gpu/drm/drm_pagemap.c
> index 0312edb8d4a8..e43676e15fe2 100644
> --- a/drivers/gpu/drm/drm_pagemap.c
> +++ b/drivers/gpu/drm/drm_pagemap.c
> @@ -460,54 +460,80 @@ static int drm_pagemap_migrate_populate_ram_pfn(struct vm_area_struct *vas,
>  {
>  	unsigned long i;
>  
> -	for (i = 0; i < npages; ++i, addr += PAGE_SIZE) {
> +	for (i = 0; i < npages;) {
>  		struct page *page, *src_page;
> +		struct folio *folio;
> +		unsigned int order = 0;
>  
>  		if (!(src_mpfn[i] & MIGRATE_PFN_MIGRATE))
> -			continue;
> +			goto next;
>  
>  		src_page = migrate_pfn_to_page(src_mpfn[i]);
>  		if (!src_page)
> -			continue;
> +			goto next;
>  
>  		if (fault_page) {
>  			if (src_page->zone_device_data !=
>  			    fault_page->zone_device_data)
> -				continue;
> +				goto next;
>  		}
>  
> +		order = folio_order(page_folio(src_page));
> +
> +		/* TODO: Support fallback to single pages if THP allocation fails */
>  		if (vas)
> -			page = alloc_page_vma(GFP_HIGHUSER, vas, addr);
> +			folio = vma_alloc_folio(GFP_HIGHUSER, order, vas, addr);
>  		else
> -			page = alloc_page(GFP_HIGHUSER);
> +			folio = folio_alloc(GFP_HIGHUSER, order);
>  
> -		if (!page)
> +		if (!folio)
>  			goto free_pages;
>  
> +		page = folio_page(folio, 0);
>  		mpfn[i] = migrate_pfn(page_to_pfn(page));
> +
> +next:
> +		if (page)
> +			addr += page_size(page);
> +		else
> +			addr += PAGE_SIZE;
> +
> +		i += NR_PAGES(order);
>  	}
>  
> -	for (i = 0; i < npages; ++i) {
> +	for (i = 0; i < npages;) {
>  		struct page *page = migrate_pfn_to_page(mpfn[i]);
> +		unsigned int order = 0;
>  
>  		if (!page)
> -			continue;
> +			goto next_lock;
>  
>  		WARN_ON_ONCE(!trylock_page(page));

I missed this last time—I'd suggest using this to make it clear we're
now operating on folios, even though the functionality is the same:

s/trylock_page(page)/folio_trylock(page_folio(page))

Reviewed-by: Matthew Brost matthew.brost at intel.com

> -		++*mpages;
> +
> +		order = folio_order(page_folio(page));
> +		*mpages += NR_PAGES(order);
> +
> +next_lock:
> +		i += NR_PAGES(order);
>  	}
>  
>  	return 0;
>  
>  free_pages:
> -	for (i = 0; i < npages; ++i) {
> +	for (i = 0; i < npages;) {
>  		struct page *page = migrate_pfn_to_page(mpfn[i]);
> +		unsigned int order = 0;
>  
>  		if (!page)
> -			continue;
> +			goto next_put;
>  
>  		put_page(page);
>  		mpfn[i] = 0;
> +
> +		order = folio_order(page_folio(page));
> +
> +next_put:
> +		i += NR_PAGES(order);
>  	}
>  	return -ENOMEM;
>  }
> -- 
> 2.43.0
>