On Thu, Mar 08, 2007 at 10:55:55AM -0700, Moore, Eric wrote:
> > It does point to the line you mentioned.
> > 
> 
> Most likely the culprit was at the beginning of the function, when we
> set hd.  I think ioc->sh was probably a NULL pointer when this is
> called.  I was wondering if you could do something like this:
> 
> (1) don't set hd at the top of the function.
> 
> (2) replace the if (!hd || !hd->ioc) check with :
> 
> if (!ioc->sh || !ioc->sh->hostdata)
>       goto out
> hd = ioc->sh->hostdata;
> if (!hd->ioc)
>       goto out;
> 

I tried that out and it got rid of the oops.   The disk mounted 
fine and was written to for a sysrq-c test. However, for one crash point 
(INT_HARDWARE_ENTRY, panic) on doing the insmod mtpsas, it seemed to find 
an extra disk that was not really there.  I am not sure if this problem is 
related.  I expected "/dev/sdb" and "/dev/sdc", but also got a "/dev/sdd".  
I saw this before but thought it was the firmware on the one older disk I 
was using (which I removed).  Once I rebooted, I saw only the two disks, 
but after manually doing rmmod and modprobe I have three again where the 
"/dev/sdb" and "/dev/sdd" are the same disk which was formerly "/dec/sdc".  
Multipath is not on.

> 
> That should eliminate the oops, but we need to debug why interrupts are
> not getting enabled.   A similiar issue was reported yesterday from
> Dell, (but not involving kdump/kexec), so I'm trying to see if we can
> replicate it.

That seems like a good start to me.  If I can help with this kexec testing, 
please ask.

> 
> 
> BTW, do you work with Mark Haverkamp?    I know him.    I notice you
> have the osdl domain name.

Yes, Mark and I are both working on the drivers testing.  I am testing on
the adapters you (LSI) sent to us.:)

Judith

> 
> Eric
> 
> 
> 
> 
> 
> 
_______________________________________________
fastboot mailing list
[email protected]
https://lists.osdl.org/mailman/listinfo/fastboot

Reply via email to