Re: v4.17-rc1: regressions on N900, N950

[Tony Lindgren]

* Pavel Machek <pavel@xxxxxx> [180523 20:14]:
> On Wed 2018-05-23 16:06:15, Pavel Machek wrote:
> > On Wed 2018-05-23 00:56:38, Aaro Koskinen wrote:
> > > Hi,
> > > 
> > > On Tue, May 22, 2018 at 10:58:26PM +0200, Pavel Machek wrote:
> > > > On Tue 2018-05-22 22:41:39, Aaro Koskinen wrote:
> > > > > My device worked with v4.17-rc1 (haven't found time to test newer kernels),
> > > > > but if you say the probe order is random then we must find some proper way
> > > > > to express the dependency.
> > > > 
> > > > I started bisect, but.. that will probably not be useful.
> > > > 
> > > > If your device works ok in v4.17-rc1, it probably works in newer -rcs,
> > > > too.
> > > 
> > > Actually, my statement may be bogus... Now I tried again with -rc1
> > > (and also -rc6) and it fails... But v4.16 works.
> > > 
> > > > Thanks for the ordering hint, I'll try to figure out what is going on
> > > > there.
> > > 
> > > My bisection pointed to 6fa7324ac5489ad43c4b6351355b869bc5458bef which
> > > doesn't seem to make any sense...?! So maybe there really is something
> > > random stuff going on? :-(
> > 
> > So... I did some experiments on v4.16.
> > 
> > Swapping tsc2005@0 and lcd: acx565akm@2 entries in the dts does break
> > stuff.
> > 
> > I thought it might be due to vio regulator, but it does not appear
> > so... screen still works with tsc2005 driver disabled in .config. (so
> > there's noone to enable vio regulator).
> 
> Deleting tsc2005@0 entry also breaks screen.
> 
> Replacing tsc2005@0 entry with 
> 
>         foobar@0 {
> 	               compatible = "not really with anything";
>  		       spi-max-frequency = <6000000>;
>  		       reg = <0>;
>  };
> 
> still results in working touchscreen, removing any of the three fields
> breaks it again.

I bisected one regression down, see "omapdrm regression in v4.17-rc series"
not sure if it covers all the issues being discussed here though.

Regards,

Tony