Re: pmspcv panic on boot on this box

On Fri, Jul 31, 2015 at 01:33:32AM -0500, Larry Rosenman wrote:
> On 2015-07-30 23:21, Glen Barber wrote:
> >On Fri, Jul 31, 2015 at 04:18:15AM +0000, Glen Barber wrote:
> >>On Thu, Jul 30, 2015 at 06:57:05PM -0500, Larry Rosenman wrote:
> >>> On 2015-07-30 17:17, Glen Barber wrote:
> >>> >On Thu, Jul 30, 2015 at 03:20:46PM -0500, Larry Rosenman wrote:
> >>> >>Kernel compiling -- give mr a bit and I'll boot it and make sure it
> >>> >>comes
> >>> >>up.
> >>> >>
> >>> >
> >>> >Larry, have you had any luck with this patch applied?  If it resolves
> >>> >your issue, I want to make sure it is included in the 10.2-RC2 build.
> >>> >
> >>> >Thanks.
> >>> >
> >>> [...]
> >>>
> >>> There are 3 pictures from the CURRENT panic.
> >>>
> >>> it STILL fails. :(
> >>>
> >>
> >>Larry, I am sorry this is causing headaches for you, and I certainly
> >>appreciate the prompt (and detailed) report, and assistance debugging
> >>this.
> >>
> >>Would you be able to try one more test case?
> >>
> >>I'm interested in the behavior without the 'nodevice pmsdrv' addition to
> >>your kernel configuration (effectively, removing the device from the
> >>GENERIC kernel), and *without* the patch provided from Benno?
> >>
> >
> >To be more specific on what I'm interested in, 'nodevice pmsdrv' and
> >'device pmsdrv' should *both* be removed from the kernel configuration,
> >and the pms(4) should only be available as a .ko module.
> >
> >>In particular, I'm interested in if ahd(4) attaches or if loader(8)
> >>auto-loads pms(4) after the PCI IDs are detected.
> >>
> >>As this issue also affects the upcoming 10.2-RELEASE, your willingness
> >>to help debug this is greatly appreciated, as you've tripped over
> >>something that would have caused a great deal of pain after 10.2 was
> >>release.
> >>
> >>I owe you several beers.
> >>
> >
> >Glen
> My kernel is based on GENERIC (I.E. include GENERIC).
> With the nodedevice pmspcv, after boot, if I kldload pmspcv, NO PANIC.
> 

Okay.  I'm interested in the behavior if the pmspcv is removed entirely
(no 'nodevice' or 'device' entry).

> Unfortunately, I've got a funeral to go to in Dallas (~150Mi away from home)
> today, Friday, and will
> not be back until late Saturday (US/CDT, GMT-5).
> 

My condolences.  :(

I'm going to be occupied testing the 10.2-RC2 (while taking into account
the pms(4) issue), so no worries regarding getting back to me.  There
was another report as well, so I'll parallelize as needed.

> I can probably help someone with this on Sunday, but would appreciate a
> real-tome conversation, as the box in question is at
> a COLO facility, and does NOT have IPMI, so I need to be physically there.
> 
> Where in the world are you located, Glen?
> 

I'm in Pennsylvania.  I'll email you my phone number off-list when you
return if that is helpful for debugging, and we can try to streamline
this.

Deal with the non-computer stuff first though, please.

Glen