Re: [PATCH for 6.19-rc1] fs: preserve file type in make_bad_inode() unless invalid

[Tetsuo Handa]

On 2026/01/07 18:36, Jan Kara wrote:
> Since Jens has picked up
> https://lore.kernel.org/all/20251217190040.490204-2-rpthibeault@xxxxxxxxx/
> yesterday I suspect the original reproducer for OCFS2 will not cause issue
> anymore even without 58b6fcd2ab34 because as far as I had a look the
> original problem was caused by the loop device getting messed up under a
> mounted OCFS2 filesystem. It would be good to verify my analysis is correct
> but I think just reverting 58b6fcd2ab34 might be the best option at this
> point.

Well, that patch wants

  	err = mutex_lock_killable(&lo->lo_mutex);
- 	if (err)
+ 	if (err) {
+		if (!(mode & BLK_OPEN_EXCL))
+			bd_abort_claiming(bdev, loop_set_status);
  		return err;
+	}

change...