On 23/3/2019 11:51 AM, Pavel Machek wrote: > On Sat 2019-03-23 11:38:04, Daniel Mack wrote: >> On 23/3/2019 11:34 AM, YueHaibing wrote: >>> On 2019/3/23 18:25, Daniel Mack wrote: >>>> On 23/3/2019 11:15 AM, Pavel Machek wrote: >> >>>>> If I look at code below, there's no difference between DEFSTATE_KEEP >>>>> and DEFSTATE_ON, right? I can't see how it works. >>>> >>>> Ah, you're right. Yes, the "keep" branch needs to go away entirely. >>>> >>>> As the chip can't be queried for it's current state, the "keep" option >>>> doesn't make sense. The only option in DT should be turn the LED on or >>>> off at probe time. YueHaibing, can you add that to your patch? >>> >>> Ok. I will send v2 with this. >> >> Note that Jacek just applied a patch of mine that removes all pdata >> handling from this driver. In order to avoid merge conflicts, you might >> want to base your patch on top of that. >> > > Actually, Daniel, at this point I'd prefer you to take over this > patch. You have the hardware and you are already working on the > code... YueHaibing's patch is fine except for the coding style issue. I can send a 2nd patch on top of that that removes the handling of "keep". Thanks, Daniel