> > > Someone swapped & and && in ide-dma.c:
> >
> > Thanks again, but found-it already -- Linus has a patch for 2.1.71
> >
> > (harmless but annoying bug).
>
> More significant question: is this the correct behavior? It seems like we
> should be jumping out of the ide-dma setup somehow if we find that
Well. actually, ide-dma.c really handles setup for PCI chipsets in
general,
not just bus-master DMA capable ones. That's why the code continues,
even without DMA capability. The benefit becomes apparent when one has
multiple PCI IDE interfaces in a single system -- ide-dma.c will find
and configure the i/o ports for all of them (one user had *six* IDE
i/fs).
And the pciirq problem is "old" news now, with a fix in Linus's queue.
-- mlord@pobox.com The Linux IDE guy