Messages in this thread | | | Date | Tue, 13 Mar 2012 18:00:01 -0400 | From | Ted Ts'o <> | Subject | Re: Fwd: ext4 extents.c bug |
| |
On Tue, Mar 13, 2012 at 09:14:18PM +0100, Frej Eriksson wrote: > > Thank you for your the information! If i get you correct does it mean > that you need to have an updated version of both the kernel and > e2fsprogs in order to have the issue fixed? Or is it enough to run > e2fsprogs with the patch applied?
If you run e2fsprogs with the patch applied, it will fix the file system corruption that caused your kernel to BUG. If the file system corruption happens again (for whatever reason; kernel bug, hardware problem, etc.) then the kernel will BUG instead of printing an "EXT4-fs error" message and marking the file system as corrupted.
If you have your file system set to "tune2fs -e panic" there will be no difference (since after the error message the kernel will be forced to reboot so e2fsck can fix the problem). But if you have your file system set to "tune2fs -e continue" or "tune2fs -c read-only", where you prefer that the file system be mounted read-only or just simply continue after noting a file system corruption problem, and you want that behavior, then you'd have to apply the kernel patch as well. (You can see what the corrent behaviour on file systems errors by looking at the output of "dumpe2fs -h /dev/sdXX | grep ^Errors".)
Regards,
- Ted
| |