BeagleBone questions (was Re: Towards an ARM system-building script)

[Warner Losh]

On Oct 30, 2012, at 10:58 AM, Adrian Chadd wrote:

> On 30 October 2012 09:31, Warner Losh <imp at bsdimp.com> wrote:
>> 
>> On Oct 30, 2012, at 10:20 AM, Adrian Chadd wrote:
>> 
>>> .. ah. armv5 has no sleep "halt" function, so it's just spinning.
>>> Chances are that's a good candidate for why things are running too
>>> warm.
>> 
>> Each SoC has its own flavor of this with varying degrees of latency...
> 
> Yeah, but the armv5 code doesn't do this, it just defaults to
> cpufunc_null or whatever it's called. Hence, there's no idle halt /
> wait-for-interrupt being done.

that's right.  My point was more 'each SoC needs to do something special, and we support a lot of SoCs' with a little bit of a dig at there not being a latency hint that the SoC could use to pick between different options.

Warner