Re: [syzbot] Re: [syzbot] [squashfs?] possible deadlock in fsnotify_destroy_mark

[syzbot <syzbot+c679f13773f295d2da53@syzkaller.appspotmail.com>]

For archival purposes, forwarding an incoming command email to
linux-kernel@vger.kernel.org.

***

Subject: Re: [syzbot] [squashfs?] possible deadlock in fsnotify_destroy_mark
Author: lizhi.xu@windriver.com

Use memalloc_nofs_save/memalloc_nofs_restore to make sure we don't end
up with the fs reclaim dependency.

#syz test

diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index c7e451d5bd51..70b77b6186a6 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -643,8 +643,13 @@ static int inotify_update_watch(struct fsnotify_group *group, struct inode *inod
 	/* try to update and existing watch with the new arg */
 	ret = inotify_update_existing_watch(group, inode, arg);
 	/* no mark present, try to add a new one */
-	if (ret == -ENOENT)
+	if (ret == -ENOENT) {
+		unsigned int nofs_flag;
+
+		nofs_flag = memalloc_nofs_save();
 		ret = inotify_new_watch(group, inode, arg);
+		memalloc_nofs_restore(nofs_flag);
+	}
 	fsnotify_group_unlock(group);
 
 	return ret;