Netgear MA301 works

[Martijn van Oosterhout]

On Thu, Nov 15, 2001 at 11:28:11AM +0100, Bjoern Giesler wrote:
> > I'm also concerned about the i386 specificness of the patch but someone
> > who knows about this can fix that.
> 
> Hmm. Do you think changing the outl/inl to outb/inb would help? We're 
> only fiddling with bit 6 on the lowest byte anyway, and the endianness / 
> word order of the PCI bus is defined.

Actually, I'm referring to the fact that non-i386 architectures don't have
in/out instructions, everything is memory-mapped. Like I said, I'm not sure
how that's supposed to work.

> OK, so if pci_resource_start() is 0, that doesn't mean anything special? 
> Is it pci_resource_len() that is important? Where do the PCI resources 
> get set anyway?

Actually, now that you put it that way, yes. A zero field in the PCI table
means that's it's not a valid address. Valid fields can never be set to zero
(IIRC) due to the fact that read-only bits define various properties.

PCI resources are usually set by the BIOS although they are in principle
alterable on the fly by anyone. Nobody usually bothers since all that
matters is that they don't overlap with anything else and the BIOS usually
does that fine.

But that's your responsibility to check. ioremap() has no real issue with
the zero address and will happily map it.

Am I going back on what I said earlier?

-- 
Martijn van Oosterhout <kleptog at svana.org>
http://svana.org/kleptog/
> Magnetism, electricity and motion are like a three-for-two special offer:
> if you have two of them, the third one comes free.