ZFS Can anyone confirm this simple ZFS bug?

D

DecimalD

Would anyone be interested in testing a panic condition on a FreeBSD host? I can easily trigger a ZFS-related panic on builds from the latest 10.3 going all the way back to last summer. I was hoping someone could independently reproduce the issue to verify that it isn't something specific to my setup.

For reference, I filed a bug on this last week: PR 207464

All it takes is any zpool with one filesystem mounted at / (called 'data' in this example). Then, run the three operations below twice in succession:
Code: 
zfs snapshot data@panic
ls -d /data/.zfs/snapshot/panic
zfs destroy data@panic
 
Yes; it does happen with a brand-new pool. The only situation I've seen where there is no panic is after a bunch of snapshots have already been successfully created/destroyed. So in that regard, it may require a freshly booted system.
 
I can't replicate this on stable/10 r296319, amd64 version of FreeBSD under VirtualBox without any modifications to the source, GENERIC kernel.

You should double check that those custom patches you're using are actually valid for a recent stable/10.
 
The bug doesn't have anything to do with possible kernel modifications in the bug report. I've been able to reproduce this consistently with a clean FreeBSD install directly off of the official 10.2 memstick image; with both UFS and ZFS boot.

Either the bug has been fixed in just the last few days, or testing it in VirtualBox does not trigger it. I suppose this may also be specific to the Intel E5 CPU architecture, which is all I have to test with.
 
You must log in or register to reply here.
Back
Top