On Tue, Mar 12, 2019 at 01:39:50PM -0700, Stephen Boyd wrote: > Quoting Patrick Wildt (2019-03-12 00:36:54) > > On Fri, Mar 08, 2019 at 07:29:05AM -0800, Stephen Boyd wrote: > > > It's mostly about making sure that any existing dtbs don't have their > > > numbers shifted around. So hopefully any overlapping identifiers aren't > > > in use yet and then those ids can be changed while leaving the ones that > > > are in use how they are. > > > > In practice I bet no one uses Linux 5.0's i.MX8M device trees since they > > lack too much support. It's so basic it's not useful. You'd still run > > your existing non-mainline bindings until it is. Thus I would argue > > changing the ABI right now would be the only chance there is. > > > > If you think that chance is gone, then I guess the reasonable thing is > > to keep the numbers and only move those (to the end) which overlap. Or > > put them into that erreneous number gap. > > > > The chance is quickly slipping away because we're going to be at -rc1 > soon. I'm not the one to decide what is and isn't being used by people > out there, so I'm happy to apply this patch now before the next -rc1 > comes out as long as it doesn't break anything in arm-soc area. The > confidence I'm getting isn't high though. Has anyone from NXP reviewed > this change? Maybe I can get an ack from someone else that normally > looks after the arm-soc/dts side of things here indicating that nothing > should go wrong? That would increase my confidence levels. The person that supplied the diff apparently is from NXP, which should be enough to say that NXP reviewed it? It's a bit of a shame that the ones that are CC'd keep quiet. I would take this chance and go ahead with it. After 5.1/rc1 there will be no chance to rectify this in a sane way.
