On Wed, Sep 09, 2015 at 04:41:04PM +0100, Chris Wilson wrote: > On Wed, Sep 09, 2015 at 06:28:50PM +0300, Ville Syrjälä wrote: > > On Tue, Sep 08, 2015 at 09:16:11PM +0100, Chris Wilson wrote: > > > Would it make sense to disable dvfs after a failure as well, > > > then the user is shown a single DRM_ERROR at runtime and we should > > > recover (by not going to the full WM next time)? > > > > I wouldn't expect any failures after we've determined that it works. > > That would indicate Punit going belly up or something, and then I'm > > not sure anything would work anymore. > > We didn't expect any before either :) And it sounds like you are arguing > that we should be reducing the noise from the victims as well :)
Well, I think I'll still leave it as is. People have generally been opposed to adding code to deal with conditions that should never happen in real life. Should I be proven wrong, you can smack me on the head with a big "told you so!" sign ;) -- Ville Syrjälä Intel OTC _______________________________________________ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx
