quota: Properly invalidate caches even for filesystems with blocksize < pagesize

Sometimes invalidate_bdev() can fail to invalidate a part of block
device cache because of dirty data. If the filesystem has blocksize
smaller than page size, this can happen even for pages containing
quota files and thus kernel would operate on stale data. Fix the
issue by syncing the filesystem before invalidating the cache.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>

1 files changed, 7 insertions, 5 deletions

diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c
index 4d2041fddefc..10d021dd37c1 100644
--- a/fs/quota/dquot.c
+++ b/fs/quota/dquot.c
@@ -2033,11 +2033,13 @@ static int vfs_load_quota_inode(struct inode *inode, int type, int format_id,
         }
 
         if (!(dqopt->flags & DQUOT_QUOTA_SYS_FILE)) {
-                /* As we bypass the pagecache we must now flush the inode so
-                 * that we see all the changes from userspace... */
-                write_inode_now(inode, 1);
-                /* And now flush the block cache so that kernel sees the
-                 * changes */
+                /* As we bypass the pagecache we must now flush all the
+                 * dirty data and invalidate caches so that kernel sees
+                 * changes from userspace. It is not enough to just flush
+                 * the quota file since if blocksize < pagesize, invalidation
+                 * of the cache could fail because of other unrelated dirty
+                 * data */
+                sync_filesystem(sb);
                 invalidate_bdev(sb->s_bdev);
         }
         mutex_lock(&dqopt->dqonoff_mutex);