On Thu, Aug 20, 2015 at 06:00:02PM +0300, Ville Syrjälä wrote: > On Thu, Aug 20, 2015 at 11:29:29AM -0300, Paulo Zanoni wrote: > > 2015-08-20 10:58 GMT-03:00 Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>: > > > Once it's otherwise known to be solid, > > > then it might make sense, although a much cooler thing would be if we > > > could actually detect when the display has failed entirely and recover > > > it somehow. > > > > > > Oh and to make the protection mechanism actually kick in reliably you > > > would somehow need to address the problems with the underrun interrupts. > > > > Which problems? > > The fact that we disable them as soon as one occurs, and even worse the > error interrupt is a shared one on many platforms, so one underrun any > pipe or some other unrelated error kills underrun detection everywhere. E.g. on hsw with vga + hdmi/dp (which now happens easily since igt creates a fake vga output, just run kms_pipe_crc_basic) underruns happen reliably on modeset and then don't work anywhere. We also seem to fail to restore them somehow :( -Daniel -- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
