[openib-general] Re: mstflint failing on sparc64

[Michael S. Tsirkin]

Hello!
Quoting r. Tom Duffy (tduffy at sun.com) "mstflint failing on sparc64":
> I have just tried running the mstflint utility on Linux/sparc64:
> 
> tat:~# lspci
> 0000:80:00.0 Host bridge: Sun Microsystems Computer Corp. Psycho PCI Bus
> Module
> 0000:80:01.0 PCI bridge: Mellanox Technology MT23108 PCI Bridge (rev a1)
> 0000:81:00.0 InfiniBand: Mellanox Technology MT23108 InfiniHost (rev a1)
> tat:~# ./mstflint -d 0000:81:00.0 q
> Can not parse device 0000:81:00.0
> *** ERROR *** Invalid argument

I wander what 0000: is for? Anyway, I added support for names format
0000:xx:xx.x just for convenience.
But its not a real issue.

By the way, looking at drivers/pci/proc.c it seems that domains>0 are not
really supported in linux yet. Do you see 8100 in a first column in
/proc/bus/pci/devices, too?
If not, could you cat /proc/bus/pci/devices please?
Please try sending it as an attachment, your mailer seems to wrap lines.

> tat:~# ./mstflint -d 81:00.0 q
> Bus error

Interesting. Maybe mmap does not work as it should?
Could you run it under gdb and do a backtrace? I also added
a sanity checks on mmap result, maybe that will help us
see what is the problem.

> tat:~# ./mstflint -d /proc/bus/pci/0000\:81/00.0 q
> *** ERROR *** cannot obtain GPIO semaphore (63)
> tat:~# ./mstflint -d /proc/bus/pci/0000\:81/00.0 q
> *** ERROR *** Flash gateway timeout: CMD doesn't become zero
> 
> Any ideas?
> 
> -tduffy

Hmm. I think I forgot about endian-ness for the /proc/bus/pci case.
I fixed that, could you try again please?

If that does not help, could you please locate the line 

#if  1

around flint.cpp line 297 , and change this to #if 0?

This will turn on lots of messages on each memory access.

MST