Messages in this thread | | | Date | Sun, 18 Jan 2015 16:29:02 +0100 | From | Hans de Goede <> | Subject | Re: [PATCH v4 4/4] ARM: mvebu: Armada 385 GP: Add regulators to the SATA port |
| |
Hi,
On 18-01-15 13:35, Mark Brown wrote: > On Sat, Jan 17, 2015 at 03:28:39PM +0100, Hans de Goede wrote: >> On 17-01-15 14:14, Mark Brown wrote: > >>> Following your argument to the logical conclusion means we can never >>> turn any regualtor off - we always have the risk that there's another >>> shared user which is going to get a power bounce if we power down. More >>> directly we'll also get people complaining that we're burning power >>> pointlessly on their systems for devices they've not even got drivers >>> enabled for. This powering down is something there's been user demand >>> for. > >> Right, note I'm only advocating to not turn off regulators marked as >> regulator-boot-on. I would expect any regulator to have such a >> marking to have at least one user with an actual driver. If people decide >> to not build that driver, and then complain we can simply tell them to >> build the driver ... > > Right, but that's not what regulator-boot-on actually means (and I'm not > sure why you would think it would TBH)
Well, the meaning of regulator-boot-on is not clearly defined really, to begin we need with fixing that, currently all the bindings file says is:
- regulator-boot-on: bootloader/firmware enabled regulator
One could easily argue that the bootloader likely has a good reason to turn the regulator on, and that unless there is a specific driver which claims the regulator and thus knows what to do with it it is best left alone ...
> so this will disrupt existing > users who are expecting the current behaviour. We could try adding a > new property but it doesn't feel very idiomatic for DT which isn't very > nice. > > Telling people not to build the driver doesn't in general work any > better than telling them to build it in I fear, it seems like it's > essentially just shuffling things around so people have to change their > kernel config in a different way to avoid issues.
Regards,
Hans
| |