ext4: fix bug_on in start_this_handle during umount filesystem

[ Upstream commit b98535d ]

We got issue as follows:
------------[ cut here ]------------
kernel BUG at fs/jbd2/transaction.c:389!
invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI
CPU: 9 PID: 131 Comm: kworker/9:1 Not tainted 5.17.0-862.14.0.6.x86_64-00001-g23f87daf7d74-dirty torvalds#197
Workqueue: events flush_stashed_error_work
RIP: 0010:start_this_handle+0x41c/0x1160
RSP: 0018:ffff888106b47c20 EFLAGS: 00010202
RAX: ffffed10251b8400 RBX: ffff888128dc204c RCX: ffffffffb52972ac
RDX: 0000000000000200 RSI: 0000000000000004 RDI: ffff888128dc2050
RBP: 0000000000000039 R08: 0000000000000001 R09: ffffed10251b840a
R10: ffff888128dc204f R11: ffffed10251b8409 R12: ffff888116d78000
R13: 0000000000000000 R14: dffffc0000000000 R15: ffff888128dc2000
FS:  0000000000000000(0000) GS:ffff88839d680000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000001620068 CR3: 0000000376c0e000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <TASK>
 jbd2__journal_start+0x38a/0x790
 jbd2_journal_start+0x19/0x20
 flush_stashed_error_work+0x110/0x2b3
 process_one_work+0x688/0x1080
 worker_thread+0x8b/0xc50
 kthread+0x26f/0x310
 ret_from_fork+0x22/0x30
 </TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---

Above issue may happen as follows:
      umount            read procfs            error_work
ext4_put_super
  flush_work(&sbi->s_error_work);

                      ext4_mb_seq_groups_show
	                ext4_mb_load_buddy_gfp
			  ext4_mb_init_group
			    ext4_mb_init_cache
	                      ext4_read_block_bitmap_nowait
			        ext4_validate_block_bitmap
				  ext4_error
			            ext4_handle_error
			              schedule_work(&EXT4_SB(sb)->s_error_work);

  ext4_unregister_sysfs(sb);
  jbd2_journal_destroy(sbi->s_journal);
    journal_kill_thread
      journal->j_flags |= JBD2_UNMOUNT;

                                          flush_stashed_error_work
				            jbd2_journal_start
					      start_this_handle
					        BUG_ON(journal->j_flags & JBD2_UNMOUNT);

To solve this issue, we call 'ext4_unregister_sysfs() before flushing
s_error_work in ext4_put_super().

Signed-off-by: Ye Bin <[email protected]>
Reviewed-by: Jan Kara <[email protected]>
Reviewed-by: Ritesh Harjani <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Signed-off-by: Sasha Levin <[email protected]>

Author: Ye Bin

fs/ext4/super.c

@@ -1176,18 +1176,23 @@ static void ext4_put_super(struct super_block *sb)
 	int aborted = 0;
 	int i, err;
 
-	ext4_unregister_li_request(sb);
-	ext4_quota_off_umount(sb);
-
-	destroy_workqueue(sbi->rsv_conversion_wq);
-
 	/*
 	 * Unregister sysfs before destroying jbd2 journal.
 	 * Since we could still access attr_journal_task attribute via sysfs
 	 * path which could have sbi->s_journal->j_task as NULL
+	 * Unregister sysfs before flush sbi->s_error_work.
+	 * Since user may read /proc/fs/ext4/xx/mb_groups during umount, If
+	 * read metadata verify failed then will queue error work.
+	 * flush_stashed_error_work will call start_this_handle may trigger
+	 * BUG_ON.
 	 */
 	ext4_unregister_sysfs(sb);
 
+	ext4_unregister_li_request(sb);
+	ext4_quota_off_umount(sb);
+
+	destroy_workqueue(sbi->rsv_conversion_wq);
+
 	if (sbi->s_journal) {
 		aborted = is_journal_aborted(sbi->s_journal);
 		err = jbd2_journal_destroy(sbi->s_journal);