System Environment: -------------------------- Platform: HSW Libdrm: (master)libdrm-2.4.59-33-gfcff9e21652c94444be565e915b117ee069227a9 Mesa: (master)e5cd42ed9abadac8be085eea67049dc1e19ade09 Xserver: (master)xorg-server-1.17.0-16-g3a06faf3fcdb7451125a46181f9152e8e59e9770 Xf86_video_intel:(master)2.99.917-183-gaefedd2baff8beaa5a5548bcf7a126ab6fca310b Libva: (master)a14498bc78cd245184e27e9810adf7a8f723d031 Libva_intel_driver: (master)4c0fda6179f9eaa5ae5887a26826af260bc5ee72 Kernel: (drm-intel-nightly)c09a3b7f343bc87a132ef107cb263e57f8246b88 Bug detailed description: --------------------------- It fails on IVB+ platforms with mesa master branch, works well on 10.5 branch. Bisect shows:1ee000a0b6737d6c140d4f07b6044908b8ebfdc7 is the first bad commit commit 1ee000a0b6737d6c140d4f07b6044908b8ebfdc7 Author: Laura Ekstrand <laura@jlekstrand.net> AuthorDate: Wed Mar 4 10:49:55 2015 -0800 Commit: Laura Ekstrand <laura@jlekstrand.net> CommitDate: Mon Mar 9 13:33:53 2015 -0700 main: _mesa_cube_level_complete checks NumLayers. _mesa_cube_level_complete now verifies that a cube map texture object actual has six texture images before proceeding. Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com> output: dEQP Core GL-CTS-2.0 (0x0052484b) starting.. target implementation = 'X11' Test case 'ES3-CTS.gtf.GL3Tests.npot_textures.npot_gen_mipmap'.. ERROR:npot_gen_mipmap: GL_INVALID_OPERATION, internalformat: GL_LUMINANCE, level: 0, target: - Fail (Fail) DONE! Test run totals: Passed: 0/1 (0.00%) Failed: 1/1 (100.00%) Not supported: 0/1 (0.00%) Warnings: 0/1 (0.00%) ==Reproduce steps== ---------------------------- 1. xinit 2. ./glcts --deqp-case=ES3-CTS.gtf.GL3Tests.npot_textures.npot_gen_mipmap
We suspect that 1ee000a was in error. Try running the test with 1ee000a reverted.
(In reply to Laura Ekstrand from comment #1) > We suspect that 1ee000a was in error. Try running the test with 1ee000a > reverted. Revert this commit, it works well.
I just check the code of mesa. It seems the token *texObj->NumLayers* which is used to judge result in patch "1ee000a", is only used in function *_mesa_TextureView*(For glTextureView in textureview.c), *_mesa_set_texture_view_state*(For glTexStorage and teximagemultisample in textureview.c), *create_texture_for_pbo*(in meta_tex_subimage.c). Since the token is not tracked by all function which is related with texture such as *glTexImage* and so on. I think the token should not be used to judge result in the function *_mesa_cube_level_complete* now, unless the token also tracked by all function which is related with Texture. So, I think this patch should be revert. Or, need I miss something else?
1ee000a was reverted, please test again
I confirm that with 1ee000a reverted the: "ES3-CTS.gtf.GL3Tests.npot_textures.npot_gen_mipmap" will pass
Fixed by 44ecf0793d872e771edc448436f7a2fd7c3390f5
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.