Re: [PATCH] spi: fix m25p80 when the cs_change hint is honored

[David Brownell <david-b-yBeKhBN/0LDR7s880joybQ@public.gmane.org>]

On Tuesday 19 May 2009, Baruch Siach wrote:
> Thanks for the clarification.
> 
> If I understood correctly then, cs_change has two opposite meanings:

No, it has exactly one meaning -- *change* normal behavior -- but that
gets applied in two different contexts, where "normal" is two things:


> 1. If cs_change != 0 in spi_transfer that is not the last in the current 
>    spi_message, then deactivate the chip select between this spi_transfer and 
>    the next.
> 
> 2. If cs_change != 0 in spi_transfer that is that last in the current 
>    spi_message, then the SPI controller is allowed to NOT deactivate the chip 
>    select after the current spi_message.
> 
> Is this correct? 

Modulo nuances on #2, yes.  See the kerneldoc for more info.


> If so would a clarification of the relevant text at  
> include/linux/spi/spi.h be accepted?

In what way is the existing language unclear?  There are
two paragraphs in the "struct spi_transfer" kerneldoc.

 
> baruch
> 



------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables 
unlimited royalty-free distribution of the report engine 
for externally facing server and web deployment. 
http://p.sf.net/sfu/businessobjects