[gta02-core] ECN0001 comments+patch on SPI

[Werner Almesberger]

Rask Ingemann Lambertsen wrote:
>    Figure 22-2 of the SC32442B manual shows SPI data changing on SPI clock
> edges.

Yeah, I wouldn't be at all surprised if the "disabled" accelerometer
misfired every once in a while. This kind of design is practically
begging for it ...

>    The direction of shift doesn't matter so much if you send software SPI
> clock pulses, just ask the card for e.g. firmware revision and discard bits
> until the alignment is right.

Hmm, I'm not entirely sure, but I think the game starts again for
each SDIO transfer.

Our main user of fast SPI would be WLAN, which we'll never see in
this form again after gta02-core anyway. So I wouldn't put much
effort into debugging 2442 SPI, considering that a real product
would use a different CPU anyway.

E.g., even if we want to stay very close to the gta02-core design,
the 2450 would be a much nicer choice than the 2442:

http://origin2.samsung.com/global/system/business/semiconductor/product/2008/8/20/401324ptb_s3c2450.pdf

>    The thing is that the 2442 SPI interface is poorly documented in places. 

Indeed. Quite a lot about the 2442 looks a bit like a rush job.

> Does it mean that gta02-core WLAN SPI is effectively limited to
> 66 MHz / 3 = 22 MHz rather than 66 MHz / 2 = 33 MHz?

Good point. I guess we just have to try and see what happens :-)

- Werner