Re: [PATCH 1/1][RESEND] Driver for the Atmel on-chip SSC on AT32AP and AT91.

[Hans-Christian Egtvedt <hcegtvedt@atmel.com>]

Hi Andrew,

On Fri, 2007-07-20 at 08:55 +0200, Andrew Victor wrote:
> > Just an example from sound/soc/at91/at91-ssc.c
> > 
> > rcmr =	  (( ssc_p->rcmr_period		<< 24) & AT91_SSC_PERIOD)
> > 			| (( 1				<< 16) & AT91_SSC_STTDLY)
> > 			| (( AT91_SSC_START_FALLING_RF	     ) & AT91_SSC_START)
> > 			| (( AT91_SSC_CK_RISING		     ) & AT91_SSC_CKI)
> > 			| (( AT91_SSC_CKO_NONE		     ) & AT91_SSC_CKO)
> > 			| (( AT91_SSC_CKS_DIV		     ) & AT91_SSC_CKS);
> 
> Well, I didn't write the above, so it's more complex than it needs to
> be.
> 
> For bitfields where the user can input any value we would usually add
> to
> the header:
>    #define AT91_SSC_STTDLY_(x)   ((x) << 16)
>    #define AT91_SSC_PERIOD_(x)   ((x) << 24)
> 
> Then it can simply be re-written as:
> 
> rcmr = AT91_SSC_PERIOD(ssc_p->rcmr_period)
> 	| AT91_SSC_STTDLY(1)
> 	| AT91_SSC_START_FALLING_RF
> 	| AT91_SSC_CK_RISING
> 	| AT91_SSC_CKO_NONE
> 	| AT91_SSC_CKS_DIV;

This is more sane and readable.

> > Would with the header style for atmel-ssc be:
> > 
> > rcmr = SSC_BF(RCMR_PERIOD, ssc_p->rcmr_period)
> > 		| SSC_BF(RCMR_STTDLY, 1)
> > 		| SSC_BF(RCMR_START, 4)
> > 		| SSC_BF(RCMR_CKI, 1)
> > 		| SSC_BF(RCMR_CKO, 0)
> > 		| SSC_BF(RCMR_CKS, 0);
> > 
> > I find the latter more readable and compact, the user also does not
> need
> > to know the offset of the different bit-fields.
> 
> But the user does then constantly have to refer to the datasheet to
> determine what CKI = 1 or CKS = 0 means.

As I said two lines furter down, I think these defines are nice. Will
adding bit-field descriptors make the SSC driver more usable by the AT91
people?

I guess there is also the possibility to include the ARM headers for
that architecture and use the definitions in the header file for AVR32.

-- 
With kind regards,

Hans-Christian Egtvedt, siv.ing. (M.Sc.)
Applications Engineer - AVR32 System Solutions - Atmel Norway