guest fs corruption during heavy I/O in guest
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
qemu (Ubuntu) |
Fix Released
|
Undecided
|
Unassigned |
Bug Description
Binary package hint: qemu
I've been trying to install Gentoo in a qemu VM running under Ubuntu Feisty. Whenever I get to the point of actually copying a large number of files to the guest root partition, I get filesystem corruption.
I've tried booting from the Gentoo 2006.01 CD, which has an old 2.6 kernel. I can partition and format /dev/hda in the guest (a 5G sparse raw image-file on an ext2 filesystem with 4k clusters), but when I untar the stage3 archive I get disk errors somewhere, and the partition is remounted read-only.
The first time I saw this, I was using reiserfs as the guest filesystem. I got a kernel oops with reiserfs. I tried using ext2fs instead, and still got corruption and a remount read-only (but no oops).
I've also tried booting from a DSL ISO image that uses a 2.4.31 kernel, creating partitions, formatting them, and untarring the same stage3 archive. Once again, I get corruption and a remount read-only. I see messages like the following in the kernel log:
attempt to access beyond end of device
03:03: rw=1, want=302013672, limit=4731142
If I mount the partition using loopback on the host and untar the archive there, I can boot from a rescue CD and chroot to it. However, when I do something else involving a lot of write I/O (like emerge --sync), I get corruption again.
Changed in qemu (Ubuntu): | |
status: | Incomplete → Fix Released |
Screenshot of the kernel oops.