[RFC] ARM: dma_map|unmap_sg plus iommu

Marek Szyprowski m.szyprowski at samsung.com
Tue Aug 9 03:04:26 EDT 2011


Hello,

On Monday, August 08, 2011 5:22 PM Ramirez Luna, Omar wrote:

> On Sun, Jul 31, 2011 at 7:57 PM, KyongHo Cho <pullip.cho at samsung.com> wrote:
> > On Fri, Jul 29, 2011 at 11:24 PM, Marek Szyprowski
> ...
> >> Right now I have no idea how to handle this better. Perhaps with should be
> >> possible
> >> to specify somehow the target dma_address when doing memory allocation, but
I'm
> >> not
> >> really convinced yet if this is really required.
> >>
> > What about using 'dma_handle' argument of alloc_coherent callback of
> > dma_map_ops?
> > Although it is an output argument, I think we can convey a hint or
> > start address to map
> > to the IO memory manager that resides behind dma API.
> 
> I also thought on this one, even dma_map_single receives a void *ptr
> which could be casted into a struct with both physical and virtual
> addresses to be mapped, but IMHO, this starts to add twists into the
> dma map parameters which might create confusion.

Nope, this is completely wrong approach. DMA-mapping is kernel wide, 
architecture independent API and you should not define any exceptions from it.

> > DMA API is so abstract that it cannot cover all requirements by
> > various device drivers;;
> 
> Agree.



More information about the linux-arm-kernel mailing list