[spi-devel-general] Re: [PATCH 2.6-git] SPI core refresh

[Vitaly W. <vw...@ru...>]

Stephen Street wrote:

>On Wed, 2005-11-30 at 13:36 -0800, David Brownell wrote:
>  
>
>>>- it is DMA-safe
>>>      
>>>
>>Which as I pointed out is incorrect.  The core API (async) has always
>>been fully DMA-safe.  And a **LOT** lower overhead than yours, which
>>allocates buffers behind the back of drivers, and encourages lots of
>>memcpy rather than just doing DMA directly to/from the buffers that
>>are provided by the SPI protocol drivers.
>>    
>>
>
>Minimal (or no) core intervention on the DMA code path is a good thing.
>I need to fix some broken hardware with software and must to move 96
>bytes from one SPI device to another on the same SPI bus every for 4ms.
>Needless memcpy's will cause substantial performance problems in my
>application. Thinner is definitely better.
>  
>
Oh yep, I must agree with you here, thanks.
However, it's not a big thing to change memcpy to spi_memcpy which will 
copy the data only when necessary.
That's what I definitely had been doing but didn't include in the patch 
sent, oops. :(
I'll come up with that shortly.

Vitaly