Xeon E3 1230 kernel panic with VT-d disabled in Bios

[mattconnolly]

I have a Xeon E3 1230 system which is having the same problems as Issue #1 and/or #4. I'm using OpenIndiana oi-151 straight from pkg.

In an effort to get it running, I disabled VT-d in the BIOS (Intel S1200BTL motherboard) and tried again. This time there was a kernel panic when booting OpenIndiana which saved a dump to disk. I took a photo of the stack trace on the screen, which can be seen here: http://s895.photobucket.com/albums/ac156/matt_connolly/?action=view&current=crashstacktrace.jpg

I'm just writing this one up in case it's a different issue from #1 and #4. Apologies for the duplicate it if turns out to be the same.

And, of course, I'm happy to help out building and testing a patched version.

[rmustacc]

Hi Matt,

A few things to consider. It sounds like when you do have VT-d enabled in the BIOS you hit #1. Can you work around it by placing the line 'set apix_enable=0' in /etc/system. Once you have that, please see if this still happens in both cases.

If it does still happen when VT-d is disabled (probably a chance), there are two things to try. The first would be to grab the latest driver from here and build it yourself. We don't maintain the OpenIndiana pacakges, so I'm not sure what version of the binary they have.

The other thing that would be quite helpful, is if you can get get us the dump from the panic that occurred. Right now I don't have any of the Xeon E3 Sandy Bridge chips to test on, though hopefully I'll have some of the by 2 Xeons in the next few weeks. In lieu of that, having the actual core dump will aid us with debugging.

[mattconnolly]

Ok, so I've added the 'set apix_enable=0' into /etc/system, and the system now successfully boots and can launch qemu-kvm with both VT-d enabled and disabled. Also, hyperthreading is enabled in the bios.

So this looks like it is just a duplicate of #1.

Would you still like the kernel panic dump? It's zvol is nearly 8GB in size...

[rmustacc]

In this case, no. Because it didn't occur with that set, I'm not as interested in the dump. Thanks for verifying that. I'm going to close this for now and we'll use issue #1 to track this.