> On Sat, 31 Mar 2001, Richard Earnshaw wrote:
> 
> > > > Last time I checked, the IOMD IRQ handler completely ignored those fields,
> > > > which is why podulebus_irq_establish() doesn't bother filling them in.  My
> > > > recommendation would thus be to do the same unless you have a good reason
> > > > to do otherwise (in which case, I want to know about it).
> > >
> > > They look to be used in poduleirqhandler, but only if something has gone
> > > really wrong with the IRQ handling.  This is why I asked, as I noted that
> > > some drivers don't set them any more. (eg the dev/podbus/if_ie.c doesn't,
> > > while the arm32 one does)
> >
> > Hmm, I wonder if that explains why I get panics with the new if_ie
> > (aka if_ei) driver in dev/podulebus...
> 
> What kind of panics?  As you might expect, the driver works for me.  I'd
> be most surprised if the lack of the interrupt mask bits was having any
> effect.

Well, and abort really, as I posted before:


rearnsha@arm.com said:
> Unfortunately, adding this driver seems to be causing my RPC to abort
> in  irq_enter when (checking the SPL masks, I think).  One possibility
> is that  this is related to the fact that I have both an Ether1 and an
> Acorn SCSI  card on the same podule bus (and that both have to share
> the same irq  entry).  Didn't have time to look further last night, or
> try backing off  to the old driver to see if it might be something
> unrelated.

> R.