Re: [PATCH] fix ppc ioremap prototype

From: Deepak Saxena
Date: Thu Sep 04 2003 - 10:51:15 EST


On Sep 04 2003, at 14:57, Geert Uytterhoeven was caught saying:
> On Thu, 4 Sep 2003, Paul Mackerras wrote:
> > Geert Uytterhoeven writes:
> > > `ioremap is meant for PCI memory space only'
> >
> > Did I say that, or someone else? :) ioremap predates PCI support by a
> > long way IIRC...
>
> inb() and friends are for ISA/PCI I/O space
> isa_readb() and friends are for ISA memory space
> readb() and friends are for PCI memory space (after ioremap())
>
> That's why other buses (e.g. SBUS and Zorro) have their own versions of
> ioremap() and readb() etc.).
>
> Life would be much easier with bus-specific I/O ops...

What happens if I have a device that can be either ISA or connected
directly to a local memory bus? The driver should be able to
ioremap(some resource) and then read/write the device without
having to have ugly #ifdefs to deal with different bus types.
Example in point is the CS8900a device which is hooked up directly
to a FPGA on the local memory bus with the bytelanes backwards.
The ammount of hacking done in the driver to get around that is
ugly. It would be much nicer if the driver still just did read*/write*
and the platform level code could deal with all the translation
issues. This requires a generic API for all I/O devices.

~Deepak

--
Deepak Saxena
MontaVista Software - Powering the Embedded Revolution - www.mvista.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/