[RFC] ARM DMA mapping TODO, v1

[Russell King - ARM Linux]

On Wed, Apr 27, 2011 at 10:21:48PM +0200, Arnd Bergmann wrote:
> On Wednesday 27 April 2011 22:16:05 Russell King - ARM Linux wrote:
> > > As for making the default being to do cache handling, I'm not completely
> > > sure how that would work on architectures where most devices are coherent.
> > > If I understood the DRM people correctly, some x86 machine have noncoherent
> > > DMA in their GPUs while everything else is coherent.
> > 
> > Well, it sounds like struct device needs a flag to indicate whether it is
> > coherent or not - but exactly how this gets set seems to be architecture
> > dependent.  I don't see bus or driver code being able to make the necessary
> > decisions - eg, tulip driver on x86 would be coherent, but tulip driver on
> > ARM would be non-coherent.
> > 
> > Nevertheless, doing it on a per-device basis is definitely the right
> > answer.
> 
> The flag would not get set by the driver that uses the device but
> the driver that found it, e.g. the PCI bus or the platform code,
> which should know about these things and also install the appropriate
> iommu or mapping operations.

As I said above, I don't think bus code can do it.  Take my example
above of a tulip pci device on x86 and a tulip pci device on ARM.  Both
use the same PCI code.

Maybe something in asm/pci.h - but that invites having lots of bus
specific header files in asm/.

A better solution imho would be to have an architecture callback for
struct device which gets registered, which can inspect the type of
the device, and set the flag depending on where it appears in the
tree.