On Tue, May 05, 2020 at 05:31:59PM +0300, Andy Shevchenko wrote: > On Tue, May 5, 2020 at 5:15 PM Mattia Dongili <malattia@xxxxxxxx> wrote: ... > > Another (possibly better) fix is to keep a list of which functions > > successfully initialized and only attempt to resume those instead of > > blindly going through all the known ones. > > I'm not against this last one but it's quite more invasive. > > OK, can we go with the series out of two (besides another fix for > ACPICA changed behaviour): > - (temporary) fix for the failing initialization (so, can be backported) > - real invasive fix as described by you > ? In all fairness (and after sleeping on it), the thermal_profile function is the only ones that depends on pointers from module load-time initialization. I'm going to send a v2 with an early exit in sony_nc_thermal_resume to avoid doing any attempt at resuming if the handle pointer is NULL. -- mattia :wq!
