On Thu, Dec 13, 2012 at 03:17:38PM +0100, Daniel Vetter wrote: > On Wed, Dec 12, 2012 at 06:15:27PM +0200, ville.syrjala at linux.intel.com wrote: > > Another month, another massive atomic patchset. > > > > I managed to clean up warts left over from the modeset-rework rebase, > > but other than that I haven't really found the time to touch this too > > much since the last time I posted patches from this set. > > > > Seeing as my schedule isn't getting any less busy in the forseeable > > future, it would be nice to get this monster merged sooner rather > > than never. > > > > Ever since the code started to resemble something sane, I've tried > > to avoid squashing patches, just in case someone was actually trying > > to follow what's changed. But clearly some of the patches can > > be squashed, and that would probably allow some of the earlier ones > > to be dropped entirely. > > > > What's clearly needed is a test tool that will stresss the mode > > setting side of the code more. My glplane test just changes the > > modes in the beginning, and then just does flips and other plane > > stuff. I'll try to get something done on that front before Christmas. > > > > Repo is here: > > https://gitorious.org/vsyrjala/linux/commits/drm_atomic_23 > > > > The repo also contains 64bit get_user() implementation for x86-32, > > which I sent to lkml and x86 people for review. So this should now > > be testable even on a 32bit system. > > Can you please upload your testcase somewhere so that I can play around > with this without too much hassle. It's here: https://gitorious.org/vsyrjala/glplane It responds to a bunch of keys in some way. Look near the end of main() in plane.c to see which keys do what. You'll need libdrm too: https://gitorious.org/vsyrjala/drm/commits/drm_atomic_7 And there's an older test here, but I'm not sure it works anymore: https://gitorious.org/vsyrjala/plane -- Ville Syrj?l? Intel OTC
