Re: 2.6.19-git20 cciss: cmd f7b00000 timedout

[Jens Axboe <jens.axboe@oracle.com>]

On Thu, Dec 14 2006, Miller, Mike (OS Dev) wrote:
>  
> 
> > -----Original Message-----
> > From: Frazier, Daniel Kent 
> > Sent: Thursday, December 14, 2006 3:12 PM
> > To: Miller, Mike (OS Dev)
> > Cc: Jens Axboe; Steve Roemen; LKML; ISS StorageDev
> > Subject: Re: 2.6.19-git20 cciss: cmd f7b00000 timedout
> > 
> > On Thu, Dec 14, 2006 at 01:44:34PM -0600, Miller, Mike (OS Dev) wrote:
> > >  
> > > 
> > > > -----Original Message-----
> > > > From: Jens Axboe [mailto:jens.axboe@oracle.com]
> > > > Sent: Thursday, December 14, 2006 12:51 PM
> > > > To: Steve Roemen
> > > > Cc: LKML; ISS StorageDev; Miller, Mike (OS Dev)
> > > > Subject: Re: 2.6.19-git20 cciss: cmd f7b00000 timedout
> > > > 
> > > > On Thu, Dec 14 2006, Steve Roemen wrote:
> > > > > -----BEGIN PGP SIGNED MESSAGE-----
> > > > > Hash: SHA1
> > > > > 
> > > > > All,
> > > > >     I tried out the 2.6.19-git20 kernel on one of my 
> > machines (HP 
> > > > > DL380 G3) that has the on board 5i controller (disabled),
> > > > > 2 smart array 642 controllers.
> > > > > 
> > > > > I get the error (cciss: cmd f7b00000 timedout) with Buffer
> > > > I/O error
> > > > > on device cciss/c (all cards, and disks) logical block 0, 1, 2, 
> > > > > etc
> > > > 
> > > > I saw this on another box, but it works on the ones that I have. 
> > > > Does
> > > > 2.6.19 work? Any chance you can try and narrow down when it broke?
> > > 
> > > Jens/Steve:
> > > We also encountered a time out issue on the 642. This one 
> > is connected 
> > > to an MSA500. Do either of you have MSA500? What controller 
> > fw are you 
> > > running? Check /proc/driver/cciss/ccissN.
> > 
> > fyi, we've been seeing this in Debian too (which is why Mike 
> > added me to the CC list), and I've narrowed it down to the 
> > 2TB patch that went into 2.6.19:
> >   http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=402787
> 
> Hmmmm. Dann, did you see this on 32-bit Debian? I have this running in
> the lab on x86_64 and ia64. Someone else was _supposed_ to test ia32 for
> me. Dammit.
> 
> Jens/Steve:
> Are your os'es 32-bit?

Yep, the machine I saw this on is 32-bit. The one I tested on myself is
x86-64, and that works. It's not driving an MSA500 either, but it's a
data point for you.

-- 
Jens Axboe