Consequences of the ext4 bug

Apparently, problems caused by last week's Ext4 bug only occur when combining several critical mount and umount options; this renders the bug harmless for most Linux users – so far, it has only affected one user. Nevertheless, ext4 lead developer Theodore "Ted" Ts'o plans to draw the necessary conclusions from the incident.

On Google+, Ts'o writes that two of the critical (journal_checksum and journal_async_commit) mount options were only intended for developers and will no longer be accessible to normal users in the next version of ext4. The developer notes that the ext4 team had planned to create "warnings for the more experimental bits of ext4" for some time, and that the implementation of these warnings will now be given a higher priority.

Although the conditions required to expose the ext4 bug are now quite well understood, the developers are continuing their research into the details of how combining lazy umount with the special mount options can trigger filesystem errors. Ts'o says that he has written a new patch for ext4, but that he wants to wait and see whether the patch can resolve the affected user's problem before releasing it.

(fab)