[lug] A bit confused

[D. Stimits]

Alan Robertson wrote:
> 
> "D. Stimits" wrote:
> >
> > Rob Riggs wrote:
> > >
> > > Also, contrary to what others have suggested, I don't
> > > think you will find any sbpcd card built that use PNP.
> > > Everyone I have seen used jumpers. (Think *really* cheap.)
> >
> > PCI is of course all PNP. I have one I can bring to the meeting, at
> > least the box with specs. It is listed as SB-2770, soundblast 16 by
> > creative labs. They were sold by Electronics Boutique. On the creative
> > labs web site it is listed as sound blaster 16 pci (the PCI and software
> > configured soundblaster):
> > http://www.soundblaster.com/products/sb16pci/
> >
> > In terms of irq, this is about the kernel config if it isn't a module,
> > not the other line that was mentioned. I've always preferred the
> > non-module for sb. In the case of module support, I would assume one
> > must still specify irq?
> 
> The web page you refer to doesn't mention a CDROM interface at all.  It
> mentions a CDROM *audio* interface, like one you'd hook to an IDE CDROM.

I've never even looked to see about a cdrom interface on the pci one.
However, the box does say it has it (holding it in my hand as we speak,
it claims it has such an interface...I tend to save old boxes). The
thing about pci is that all pci is plug-n-play, that all values must be
assigned, but normally it is the bios that sets that up (saving us and
programs from having to do it). The very gold of pci is that all
components on the bus are assignable resources, with a pseudo bios
capable of describing the device so software and/or the bios can do the
setup, without user intervention. Non-pci isn't smart enough without
intervention of some kind; jumpers, kernel compile, isapnp, so on.

Depending on the PCI info in the bios of the device, a conflict might or
might not be resolvable by the bios; the part about "device or resource
busy" is a very strong indication of something not being initialized,
and with a pci device, that is quite often a conflict that the bios
can't resolve. User intervention (e.g., parameters of modules or kernel
compile) or changing slot position sometimes helps. Only part of the
point is irrelevant if the device isn't the pci soundblaster; the
"device or resource busy" part would indicate either that it is set to
use a conflicting hardware resource, or one that it can't work with. If
it is moved to an earlier slot, it is possible the card will work
because another device that had taken the resource will only get second
crack at it after that. My real point is that changing to the first slot
is an easy test, and regardless of who configures the card (bios or
kernel or jumpers), it won't work if those resources are taken by
something else before it gets a chance.

> 
> AFAIK You never have to assign IRQs to PCI devices.
> 
>         -- Alan Robertson
>            alanr at suse.com
> 
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug