Re: ATA driver failures with pvops dom0, (was: aic79xx failures with pvops dom0 2.6.32.25)

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Mon, Nov 22, 2010 at 04:46:46PM -0500, micah anderson wrote:
> On Mon, 22 Nov 2010 11:28:29 -0500, Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> wrote:
> > > [   16.572048] scsi0 : Adaptec AIC79XX PCI-X SCSI HBA DRIVER, Rev 3.0
> > > [   16.572051]         <Adaptec AIC7902 Ultra320 SCSI adapter>
> > > [   16.572053]         aic7902: Ultra320 Wide Channel A, SCSI Id=7, PCI-X 101-133MHz, 512 SCBs
> > > [   16.572598] aic79xx 0000:03:02.1: found PCI INT B -> IRQ 5
> > 
> > That is a rather odd IRQ number that is shared amongst all of the devices. Is there
> > a "OS Compatibility" BIOS option where you can select Linux?
> 
> I'm intending to have the BIOS queried at the earliest convenient
> time. Its not near me so I can't get this information right away, but I
> will.

Take your time. I've a full plate of bugs to deal with :-(
> 
> > So MA Young found an interesting issue where the IRQs below 16 don't get programmed
> > correctly, which was fixed in .. some unstable version, but before we go that route
> > 1) Go in the serial console and hit Ctrl-A, hit '?' and hit '*' and send the
> > output. I am curious to see if the IO-APIC ends up having a proper vector as during
> > bootup it looks to be set to nothing but it should have by now have a good value.
> 
> Ok, here it is:
.. snip ..
> (XEN)    IRQ:   5 affinity:00000000,00000000,00000000,00000001 vec:38 type=IO-APIC-edge    status=00000010 in-flight=0 domain-list=0:  5(-S--),

So it thinks it is an edge, but if that IRQ is shared it has to be level. You have
a really weird motherboard..

One more thing - can you boot the kernel as baremetal (so, no Xen) and "cat /proc/interrupts"
and also attach the full serial bootup log? I am curious to see how the irq 5 is setup.
It should show up as level.