Re: More DMA API junk

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

> On your platform, dma_sync_single_range() has a defined meaning for
> streaming mappings.  dma_cache_sync will be a nop, so they're not
> interchangeable.

Ok, so we should make that clear in DMA-API.txt, currently it's
definitely not ;) I'll propose a patch later on if nobody beats me
on this.

> You're confused about what it's used for.  It's designed only to be
> called on memory allocated by dma_alloc_noncoherent() and tells you if
> that API actually returned coherent memory or not.  This was designed
> for ARM which has a limited range of allocateable coherent memory and
> then would need to fail dma_alloc_coherent() or, in the case of
> dma_alloc_noncoherent() begin handing out ordinary kmalloc'd memory.

I still have a problem with the argument. We should pass at least both
the virtual and physical address then. Or also make it clear in the
DMA-API.txt that it is only valid on the result of dma_alloc_noncoherent
in which case it becomes legal for me to hard-wire a result of 1.

Ben