On Fri, May 18, 2012 at 11:55:53AM +0200, Guennadi Liakhovetski wrote: > On Thu, 3 May 2012, Mark Brown wrote: > > should it be being factored out into the core? > Weeeell, it's not a lot of code. We could make this a library function, > provided by the core, but I don't think it would be very straight forward > to also automate its calling, thus freeing individual drivers from this > task. As you can see in these two drivers, one of them turns the card on > already upon MMC_POWER_UP, the other one only in MMC_POWER_ON. I'm not > sure off the top of my head why this difference is there, but in any case > different drivers might have different ideas about this. So, I'm not sure > whether this micro-optimisation is worth it. My take on this would be that we shouldn't even have to think about what the differences are and that if there are genuine tradeoffs here it's likely that we should be offering them at runtime or to boards or something rather than picking something in an individual driver. If there are meaningful differences I'd at least expect to be able to tell what they are.
Attachment:
signature.asc
Description: Digital signature
