> The Dallas device is a 1-wire bus, not I2C, and I doubt it'll 
> interoperate gracefully with I2C, so it's off on its own. 
> I think that credible numbers have been extracted from the Dallas 
> device, but I'm not sure that the I2C device(s) ave been tested at 
> all - however, they're connectd, and can't really be wrong. 

ID from the Dallas chip is working fine on its 1-wire nasty timeframe dependant bus. See kernel/arch/arm/mach-sa1100/balloon_id.c. I read the ID at kernel wakeup where I can suspend interrupts with less system damage and cache the result that is returned via proc. Bootldr also can read the ID.

I2C. I did try modifying the linux I2C bit banging interface (the existing l3-bit-sa1100.c assumes the I2C GPIOs are the same as the L3 bus so they needed separating out) to attach the RTC but progress was not fruitful (see kernel/arch/arm/mach_sa1100/balloon_rtc.c). It may be that the I2C stuff worked but the RTC interface was flawed or the hardware may be broken or my code was worse than expected, not sure. I mentioned this as a useful thing for someone to look at a while back but there was no takeup.

Nick

> 
> Steve
> 
> 
> 
> 
> _______________________________________________
> Balloon mailing list
> Balloon@balloonboard.org
> http://balloonboard.org/mailman/listinfo/balloon
>