Re: [Stable 4.4 - NEEDS REVIEW - 2/3] mm: filemap: don't plant shadow entries without radix tree node

[Johannes Weiner]

Hi Michal,

On Mon, Oct 24, 2016 at 05:26:04PM +0200, Michal Hocko wrote:
> @@ -155,44 +155,27 @@ static void page_cache_tree_delete(struct address_space *mapping,
>  				   struct page *page, void *shadow)
>  {
>  	struct radix_tree_node *node;
> -	unsigned long index;
> -	unsigned int offset;
> -	unsigned int tag;
>  	void **slot;
>  
>  	VM_BUG_ON(!PageLocked(page));
>  
>  	__radix_tree_lookup(&mapping->page_tree, page->index, &node, &slot);
>  
> -	if (shadow) {
> -		mapping->nrshadows++;
> -		/*
> -		 * Make sure the nrshadows update is committed before
> -		 * the nrpages update so that final truncate racing
> -		 * with reclaim does not see both counters 0 at the
> -		 * same time and miss a shadow entry.
> -		 */
> -		smp_wmb();
> -	}
> -	mapping->nrpages--;
> +	radix_tree_clear_tags(&mapping->page_tree, node, slot);
>  
>  	if (!node) {
> -		/* Clear direct pointer tags in root node */
> -		mapping->page_tree.gfp_mask &= __GFP_BITS_MASK;
> -		radix_tree_replace_slot(slot, shadow);
> -		return;
> -	}

There is no need to include the refactoring of the tag clearing in the
stable backport. I already sent a simpler backport of this patch for
4.4 to Greg, attached here for reference: