Re: [PATCH] Clarify pci_iomap() usage for MMIO-only devices

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

On Tue, 2007-09-18 at 12:03 -0700, Linus Torvalds wrote:
> 
> On Tue, 18 Sep 2007, Luis R. Rodriguez wrote:
> > 
> > Alright, here is the same patch inline with s/recommended/required/ language:
> 
> Well, the thing is, I'm not at all sure that I agree with this.
> 
> If you use ioport_map/unmap, then you really *should* access them with the 
> proper iomem accessors (ioread/iowrite). The fact that it may happen to 
> work (when using the default lib/iomap.c implementation, at least) on 
> some architectures and with the current implementation still doesn't mean 
> that you should necessarily use readb/writeb.
> 
> After all, you cannot use "inb/outb" on it, even if would happen to be an 
> IO address.
> 
> So what is this usage that wants to use the bogus accessor? Why not fix 
> that instead of adding documentation for something that is very arguably 
> something we want to *avoid* having people do!

Agreed.

ioremap -> readb/writeb
ioport_map/pci_iomap -> ioread/iowrite

and nothing else should be allowed.

Cheers,
Ben.