On Fri, 2014-10-17 at 11:13 +0100, Steven Newbury wrote: > On Mon, 2014-09-01 at 12:09 +0100, Steven Newbury wrote: > > I tried building the xorg intel ddx driver with only DRI3 support, > > with DRI1 and DRI2 disabled. > > > > glxinfo says direct rendering is enabled, but gives no core > > contexts*. > > > > gnome-shell appears to be using software fallback, generally there > > seems to be no hardware acceleration. I'm wondering if DRI3 isn't > > ever initialising successfully, and now there's no DRI2 to fall > > back > > to. > > > > What is the minimal kernel version for DRI3? I'm currently stuck > > on > > the 3.14 stable series due to an unresolved kernel bug (which I > > need > > to bisect...), is it too old? > > > > I remember before the intel (SNA) support landed > > LIBGL_DEBUG=verbose > > used to mention DRI3 failing, I see nothing mentioning DRI3 at all > > now. Is there some way of determining which DRI extension is in > > use? > > DRI3 is properly listed in xdpyinfo. > > > > > > * I also get no core contexts also with xwayland/glx, related? > > As a follow-up, I think this is related to > https://bugs.freedesktop.org/show_bug.cgi?id=85064 > > Something tells me DRI3 isn't really working with i965. This is > largely going unnoticed since when DRI2 is available it's usually > used > instead. webkit-gtk is not falling back to DRI2 without DRI3 being > specifically disabled with LIBGL_DRI3_DISABLE=1. > Sorry for replying to myself again! I should have mentioned, I'm back on the current kernel now, so that can be eliminated. I get the same behaviour with uxa/glamor backends as with sna. It's still suspicious to me that core contexts are unavailable with DRI3, either with the intel driver build with --disable-dri2, or from xwayland.
Attachment:
signature.asc
Description: This is a digitally signed message part
_______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
