Re: Splitting a THP beyond EOF
From: Matthew Wilcox <willy@infradead.org>
Date: 2020-10-20 14:26:56
Also in:
linux-fsdevel, linux-xfs
On Tue, Oct 20, 2020 at 02:43:57AM +0100, Matthew Wilcox wrote:
I think the easiest way to fix this is to decline to allocate readahead pages beyond EOF. That is, if we have a file which is, say, 61 pages long, read the last 5 pages into an order-2 THP and an order-0 THP instead of allocating an order-3 THP and zeroing the last three pages.
Oh yeah, really easy.
+++ b/mm/readahead.c@@ -481,6 +481,11 @@ void page_cache_ra_order(struct readahead_control *ractl, if (order == 1) order = 0; } + /* Don't allocate pages past EOF */ + while (index + (1UL << order) - 1 > limit) { + if (--order == 1) + order = 0; + } err = ra_alloc_page(ractl, index, mark, order, gfp); if (err) break;
I've added that to an earlier patch and I've pushed out commit cd3fa4bc6516 as the head of http://git.infradead.org/users/willy/pagecache.git/shortlog