On Wed, Apr 09, 2014 at 05:59:30PM +0300, Imre Deak wrote: > On Wed, 2014-04-09 at 11:36 -0300, Paulo Zanoni wrote: > > 2014-04-08 13:57 GMT-03:00 Imre Deak <imre.deak@xxxxxxxxx>: > > > On VLV we depend on RC6 saving the GT render and media HW context before > > > going to D3 state, so disable RPM if RC6 is not enabled. > > > > Maybe this problem affects other platforms too, and we just didn't > > notice because RC6 is always enabled on them. Did we test this on the > > other platforms? > > No, tbh I wanted to submit this as early as possible. I think now with > patch 10/15 RPM should be disabled for all GEN6+ platforms. > > > Do we have a way to disable RC6 at runtime? If yes, then we could > > probably try to write some test at pm_pc8.c. > > It's a module option, but I can look into adding a test for it. Doing a test for this is fairly tricky and since disabling rc6 isn't a supported option really not that useful imo. What might be genuinely useful though is checking that the entire render state gets restored correctly by doing - rendercpy batch - make sure we go into runtime suspend - re-emit the rendercpy 3D_PRIM command _without_ all the state setup This obviously needs hw contexts to work. If we somehow fail to restore some of the render state then this should nicely blow up all over the place. Of course it's a crude check, but better than just doing a blit or something which essentially just checks that the rings and ptes still work. -Daniel -- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
