Re: [PATCH v3 06/18] mm/huge_memory: consolidate irq and locking for folio split

[David Hildenbrand (Arm)]

> +
> +	/*
> +	 * Check if the folio is present in page cache.
> +	 * We assume all tail are present too, if folio is there.
> +	 */
> +	if (xas_load(xas) != folio) {
> +		ret = -EAGAIN;
> +		goto fail;
> +	}

How can this actually happen? We locked the folio and folio->mapping is set. How
could it possibly get truncated in the meantime?

> +
> +	if (!folio_ref_freeze(folio, folio_cache_ref_count(folio) + 1)) {
> +		ret = -EAGAIN;
> +		goto fail;
> +	}
>  
>  	if (folio_test_pmd_mappable(folio) &&
>  	    new_order < HPAGE_PMD_ORDER) {
> @@ -4106,6 +4123,8 @@ static int __folio_freeze_split_unmapped_file(struct folio *folio, unsigned int
>  	if (do_lru)
>  		lruvec_unlock(lruvec);
>  
> +fail:
> +	xas_unlock_irq(xas);
>  	return ret;
>  }
>  
> @@ -4245,19 +4264,7 @@ static int __folio_split(struct folio *folio, unsigned int new_order,
>  
>  	unmap_folio(folio);
>  
> -	/* block interrupt reentry in xa_lock and spinlock */
> -	local_irq_disable();
> -	if (mapping) {
> -		/*
> -		 * Check if the folio is present in page cache.
> -		 * We assume all tail are present too, if folio is there.
> -		 */
> -		xas_lock(&xas);
> -		xas_reset(&xas);
> -		if (xas_load(&xas) != folio) {
> -			ret = -EAGAIN;
> -			goto fail;
> -		}
> +	if (!is_anon) {

Ah, there it is :)

Nice!


-- 
Cheers,

David