Created attachment 122269 [details] Piglit patch to add failing test On Intel Haswell (at least: I don't have other hardware to test), on Mesa git aa3b85f, when using the isolines tessellation primitive mode, the outer tessellation levels are interpreted the wrong way around. The GL_ARB_tessellation_shader specification says, under the header "Isoline Tessellation": The number of isolines generated is derived from the first outer tessellation level; the number of segments in each isoline is derived from the second outer tessellation level. However, the number of isolines is currently taken from gl_TessLevelOuter[1], and the number of segments in each from gl_TessLevelOuter[0]. I've attached a Piglit patch that adds a test for this.
Thanks a ton for tracking this down and writing a test! This patch should fix the problem: https://lists.freedesktop.org/archives/mesa-dev/2016-March/109935.html
Oh, this landed a while ago, along with some other fixes: commit 5b2d8c2273c6f48e764a1386240ec674cb4aa4ad Author: Kenneth Graunke <kenneth@whitecape.org> Date: Mon Mar 14 14:22:39 2016 -0700 i965: Fix gl_TessLevelOuter[] for isolines. commit a3dc99f3d48df3da1e997d95961747daed0ba6b9 Author: Kenneth Graunke <kenneth@whitecape.org> Date: Thu May 26 00:29:56 2016 -0700 i965: Fix the passthrough TCS for isolines. commit 25e1b8d366a6131bc9d46fe27f6bc476f05a7a58 Author: Kenneth Graunke <kenneth@whitecape.org> Date: Tue May 31 21:00:43 2016 -0700 i965: Fix isoline reads in scalar TES.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.