fs/drop_caches.c: avoid softlockups in drop_pagecache_sb()
authorJan Kara <jack@suse.cz>
Fri, 1 Feb 2019 22:21:23 +0000 (14:21 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 1 Feb 2019 23:46:24 +0000 (15:46 -0800)
When superblock has lots of inodes without any pagecache (like is the
case for /proc), drop_pagecache_sb() will iterate through all of them
without dropping sb->s_inode_list_lock which can lead to softlockups
(one of our customers hit this).

Fix the problem by going to the slow path and doing cond_resched() in
case the process needs rescheduling.

Link: http://lkml.kernel.org/r/20190114085343.15011-1-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/drop_caches.c

index 82377017130f0137a4208e0082bb61946cc0b817..d31b6c72b47646fd8ae2ebd3746de307f8017fa0 100644 (file)
@@ -21,8 +21,13 @@ static void drop_pagecache_sb(struct super_block *sb, void *unused)
        spin_lock(&sb->s_inode_list_lock);
        list_for_each_entry(inode, &sb->s_inodes, i_sb_list) {
                spin_lock(&inode->i_lock);
+               /*
+                * We must skip inodes in unusual state. We may also skip
+                * inodes without pages but we deliberately won't in case
+                * we need to reschedule to avoid softlockups.
+                */
                if ((inode->i_state & (I_FREEING|I_WILL_FREE|I_NEW)) ||
-                   (inode->i_mapping->nrpages == 0)) {
+                   (inode->i_mapping->nrpages == 0 && !need_resched())) {
                        spin_unlock(&inode->i_lock);
                        continue;
                }
@@ -30,6 +35,7 @@ static void drop_pagecache_sb(struct super_block *sb, void *unused)
                spin_unlock(&inode->i_lock);
                spin_unlock(&sb->s_inode_list_lock);
 
+               cond_resched();
                invalidate_mapping_pages(inode->i_mapping, 0, -1);
                iput(toput_inode);
                toput_inode = inode;