mm/swapfile: mark racy access on si->highest_bit

In scan_swap_map_slots(), si->highest_bit can by changed by
swap_range_alloc() concurrently.  All reads on si->highest_bit except one
is either protected by lock or read using READ_ONCE.  So mark the one racy
read on si->highest_bit as benign using READ_ONCE.

This patch is aimed at reducing the number of benign races reported by
KCSAN in order to focus future debugging effort on harmful races.

Link: https://lkml.kernel.org/r/tencent_912BC3E8B0291DA4A0028AB424076375DA07@qq.com
Signed-off-by: linke li <lilinke99@qq.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 96606580ee091d5b5635249b09f3d88511e9cd92..f6ca215fb92f3b3f3bc284925c02578252af8fb8 100644 (file)
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -902,7 +902,7 @@ static int scan_swap_map_slots(struct swap_info_struct *si,
                last_in_cluster = offset + SWAPFILE_CLUSTER - 1;
 
                /* Locate the first empty (unaligned) cluster */
-               for (; last_in_cluster <= si->highest_bit; offset++) {
+               for (; last_in_cluster <= READ_ONCE(si->highest_bit); offset++) {
                        if (si->swap_map[offset])
                                last_in_cluster = offset + SWAPFILE_CLUSTER;
                        else if (offset == last_in_cluster) {