ext4: fix bdev write error check failed when mount fs with ro
authorZhang Xiaoxu <zhangxiaoxu5@huawei.com>
Mon, 28 Sep 2020 02:05:56 +0000 (22:05 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Sun, 18 Oct 2020 14:36:59 +0000 (10:36 -0400)
Consider a situation when a filesystem was uncleanly shutdown and the
orphan list is not empty and a read-only mount is attempted. The orphan
list cleanup during mount will fail with:

ext4_check_bdev_write_error:193: comm mount: Error while async write back metadata

This happens because sbi->s_bdev_wb_err is not initialized when mounting
the filesystem in read only mode and so ext4_check_bdev_write_error()
falsely triggers.

Initialize sbi->s_bdev_wb_err unconditionally to avoid this problem.

Fixes: bc71726c7257 ("ext4: abort the filesystem if failed to async write metadata buffer")
Cc: stable@kernel.org
Signed-off-by: Zhang Xiaoxu <zhangxiaoxu5@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20200928020556.710971-1-zhangxiaoxu5@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
fs/ext4/super.c

index d1ea1c7fc8c44b83d5eef6144d8d1d08ff745c14..61af903e7b32e47df88c15a31ef0408258057f13 100644 (file)
@@ -4825,9 +4825,8 @@ no_journal:
         * used to detect the metadata async write error.
         */
        spin_lock_init(&sbi->s_bdev_wb_lock);
-       if (!sb_rdonly(sb))
-               errseq_check_and_advance(&sb->s_bdev->bd_inode->i_mapping->wb_err,
-                                        &sbi->s_bdev_wb_err);
+       errseq_check_and_advance(&sb->s_bdev->bd_inode->i_mapping->wb_err,
+                                &sbi->s_bdev_wb_err);
        sb->s_bdev->bd_super = sb;
        EXT4_SB(sb)->s_mount_state |= EXT4_ORPHAN_FS;
        ext4_orphan_cleanup(sb, es);
@@ -5719,14 +5718,6 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
                                goto restore_opts;
                        }
 
-                       /*
-                        * Update the original bdev mapping's wb_err value
-                        * which could be used to detect the metadata async
-                        * write error.
-                        */
-                       errseq_check_and_advance(&sb->s_bdev->bd_inode->i_mapping->wb_err,
-                                                &sbi->s_bdev_wb_err);
-
                        /*
                         * Mounting a RDONLY partition read-write, so reread
                         * and store the current valid flag.  (It may have