Re: forcing a pci nic to use a different irq?

[james terris <shinden@sympatico.ca>]

Sorry, I'm joining this conversation a little late so forgive
me if this has already been said...

Is your ethernet card an ISA one?
If so there should be an configuration application
that will let you move it to a different irq.

Otherwise in your BIOS somewhere there should be a setting
to assign irqs to individual PCI slots from which you
can give each slot it's own irq. The card in that slot
will use that irq of course.

Irq sharing is possible though. My interrupt table looks like
the following and I have no problems:

           CPU0       CPU1       
  0:   11296449   11304831    IO-APIC-edge  timer
  1:          0          2    IO-APIC-edge  keyboard
  2:          0          0          XT-PIC  cascade
  3:    7682376    7694587    IO-APIC-edge  eth0
  8:          2          0    IO-APIC-edge  rtc
  9:    4766503    4764709   IO-APIC-level  advansys, aic7xxx, eth1
 10:       5842       5890   IO-APIC-level  advansys, aic7xxx
 11:      18862      18793   IO-APIC-level  advansys
 12:          4          3   IO-APIC-level  advansys
 13:          1          0          XT-PIC  fpu
 14:     183701     180103    IO-APIC-edge  ide0
 15:    2003423    2014463    IO-APIC-edge  ide1
NMI:          0
ERR:          0

ttyl,
james

Paul Mackinney wrote:
> 
> Matthew Garman muttered:
> > According to the Ethernet HOWTO, the most common cause of this problem is
> > an IRQ conflict.  This seems believable, because...
> >
> > cat /proc/interrupts
> >
> >            CPU0
> >   0:      42726          XT-PIC  timer
> >   1:       2024          XT-PIC  keyboard
> >   2:          0          XT-PIC  cascade
> >   4:      12012          XT-PIC
> >   5:          3          XT-PIC  soundblaster
> >  11:       5986          XT-PIC  sym53c8xx, eth0
> >  12:       4799          XT-PIC  PS/2 Mouse
> >  14:          2          XT-PIC  ide0
> > NMI:          0
> > ERR:          0
> >
> > You can see that both my SCSI controller and my ethernet card live on IRQ
> > 11.
> I'm not an expert on IRQs, but my (vague!) understanding is that it's
> actually the PCI controller that uses the IRQ, so if these devices are
> on the same bus, it doesn't necessarily indicate a conflict.
> 
> Joost is absolutely on the right track: what changed between when it
> worked and when it quit working? This should be a clue. Also, before
> pinging it from another host, make sure that ifconfig returns good info
> for eth0 and that it can ping itself at eth0's TCP/IP address--if either
> of these fail then you already know it's not talking.
> 
> One standard trouble-shooting technique for PCI devices is to swap the
> cards into different slots.