[RFC 0/2] fix dma_map_sg not to do barriers for each buffer

[rdunlap@xenotime.net (Randy Dunlap)]

On 02/10/10 16:39, FUJITA Tomonori wrote:
> On Wed, 10 Feb 2010 13:27:47 -0800
> Randy Dunlap <rdunlap@xenotime.net> wrote:
> 
>> On 02/10/10 12:37, adharmap at codeaurora.org wrote:
>>> From: Abhijeet Dharmapurikar <adharmap@quicinc.com>
>>>
>>> Please refer to the post here
>>> http://lkml.org/lkml/2010/1/4/347
>>>
>>> These changes are to introduce barrierless dma_map_area and dma_unmap_area and
>>> use them to map the buffers in the scatterlist. For the last buffer, call
>>> the normal dma_map_area(aka with barriers) effectively executing the barrier
>>> at the end of the operation.
>>>
>>> Note that the barrierless operations are implemented for few arm
>>> architectures only and I would implement for others once these are okayed by the
>>> community.
>>
>> So when you add these interfaces for other architectures, you will also
>> update Documentation/DMA-API.txt,  right??
> 
> Seems that you misunderstand him.
> 
> He is talking about other "arm" architectures. His patchset improves
> arm's internal implementation (dma_map_area and dma_unmap_area are not
> the DMA API; not exported for driver writers). He meant that the
> patchset doesn't cover all arm architectures.
> 
> This is about arm's implementation details and not related with other
> non arm architectures. So no need to update Documentation/DMA-API.txt.

OK, in that case I did misunderstand.  Thanks for the info.

-- 
~Randy