WLAN from SDIO to SPI: battle plan

[Werner Almesberger]

Andy Green wrote:
> In terms of proving viability of s3c2442 SPI, maybe it is more critical
> to prove s3c24xx_spi can work at all

Yup, that's what I'm experimenting with at the moment. Ben Dooks told
me that the 24xx SPI driver should work, even with DMA, but I've
noticed that it's easy to feed the SPI stack parameters that cause
confusion. (E.g., the odd CS timing is probably such a thing.)

> stack.  Because shifting the rest of the big stack over to SPI is going
> to be a very phyrric victory if it turns out that fast SPI is broken and
> we can't really use it on 2442 anyway.

Yeah, that would be a lot of sweat for nothing ;-)

> I recall sitting there with Willie trying to get the motion sensors
> working with s3c24xx_spi and having to switch to s3c24xx_spi_gpio about
> 6 months ago.

Did you examine what went wrong at the signal level ?

- Werner