Created attachment 33825 [details] piglit test summary I get 65 piglit tests fail with r600 driver, see below. Here are the software version: 1) Mesa: 7.8 branch, commit ba8f4c5d36e6c9cd806a4c3360a8202b1f81ee05 Build command: ./configure --prefix=/opt/xorg --with-dri-drivers=r600,swrast --disable-gallium --with-dri-driverdir=/opt/xorg/lib/dri --enable-glx-tls 2) piglit: master branch, commit 3ff22d685dfacfb517058773ec5584ba8d47c8d5 3) Xorg 7.5, and xf86-video-ati 6.12.191. 4) Kernel: Linux debian 2.6.33 #10 SMP PREEMPT Thu Feb 25 11:48:12 EET 2010 x86_64 GNU/Linux CONFIG_DRM_RADEON=y CONFIG_DRM_RADEON_KMS=y # CONFIG_FB_RADEON is not set dmesg output is attached in attachment "log". 5) Video card 01:00.0 VGA compatible controller: ATI Technologies Inc RV730 PRO [Radeon HD 4650] 6) Compiler: gcc version 4.4.3 (Debian 4.4.3-3) 7) glxinfo | grep OpenGL OpenGL vendor string: Advanced Micro Devices, Inc. OpenGL renderer string: Mesa DRI R600 (RV730 9498) 20090101 TCL DRI2 OpenGL version string: 2.0 Mesa 7.8-devel OpenGL shading language version string: 1.10 Full glxinfo is attached in attachment "log". 8) Piglit tests See attached "summary" for details, here are some segfaults I've seen in dmesg while running the tests: [ 7255.092525] glean[19775]: segfault at 28 ip 00007f5936966ada sp 00007fff1b9741d0 error 4 in r600_dri.so[7f5936927000+26f000] [ 7741.425984] vp-bad-program[20450]: segfault at 28 ip 00007fbf9e017ada sp 00007fffce2c8570 error 4 in r600_dri.so[7fbf9dfd8000+26f000] [ 7747.659629] asmparsertest[20471]: segfault at cd ip 0000003b3a445411 sp 00007fff9613a280 error 4 in libc-2.10.2.so[3b3a400000+14a000] [ 8663.112725] vp-bad-program[21816]: segfault at 28 ip 00007f4a8709ba9a sp 00007fffe087c430 error 4 in r600_dri.so[7f4a8705c000+26f000] [ 8669.317341] asmparsertest[21837]: segfault at cd ip 0000003b3a445411 sp 00007fff0940d5c0 error 4 in libc-2.10.2.so[3b3a400000+14a000] piglit tests, passed: 614/679 Failed tests: glean/texture_srgb glean/fbo glean/paths glean/texUnits glean/polygonOffset glean/glsl1 glean/exactRGBA glean/pointAtten glean/vertProg1 glean/fragProg1 glean/logicOp glslparsertest/dataType13.frag glslparsertest/dataType5.frag glslparsertest/CorrectFull.frag glslparsertest/CorrectConstFolding2.vert glslparsertest/struct9.frag glslparsertest/matrix.V110.frag glslparsertest/CorrectConstruct.vert glslparsertest/CorrectFull.vert glslparsertest/TextureRectangleSamplers.frag glslparsertest/conditional3.frag glslparsertest/ConstantConversions.frag glslparsertest/comma2.vert glslparsertest/function9.frag glslparsertest/comma1.vert glslparsertest/CorrectConstFolding1.vert glslparsertest/dataType19.frag glslparsertest/dataType4.frag glslparsertest/function5.frag glslparsertest/function3.frag glslparsertest/constFunc.frag glslparsertest/ParseTest3.frag glslparsertest/preprocess1.frag glslparsertest/constructor3.V110.frag glslparsertest/CorrectUnsizedArray.frag glslparsertest/comma3.vert general/scissor-bitmap general/linestipple shaders/glsl-bug-22603 shaders/glsl-vs-mov-after-deref shaders/fp-indirections2 shaders/fp-lit-mask shaders/vpfp-generic/fogcoord-dph shaders/vpfp-generic/nv-arl shaders/vpfp-generic/nv-mov shaders/vpfp-generic/nv-init-zero-addr shaders/vpfp-generic/fogcoord-dp3 shaders/vpfp-generic/vp-arl-constant-array-huge-offset-neg shaders/vpfp-generic/fogcoord-dp4 shaders/vpfp-generic/nv-init-zero-reg shaders/vpfp-generic/vp-arl-neg-array shaders/vpfp-generic/vp-exp-alias shaders/vpfp-generic/nv-add shaders/vp-address-01 shaders/vp-bad-program shaders/glsl-derivs shaders/glsl-vs-arrays shaders/glsl-deriv-varyings asmparsertest/ARBvp1.0/param-08.txt texturing/depth-tex-modes texturing/depth-tex-compare texturing/tfp texturing/texdepth fbo/fbo-copyteximage fbo/fbo-readpixels
Created attachment 33826 [details] log Full output from dmesg, Xorg.0.log, and glxinfo.
I've run the tests like this: ./piglit-run.py tests/r500.tests results/r500_nopbo.results Where the only difference in tests/r500.tests is that I enabled glean quick mode (was taking too long otherwise): + +GleanTest.globalParams += [ "--quick" ]
(In reply to comment #0) > glslparsertest/dataType13.frag > glslparsertest/dataType5.frag > glslparsertest/CorrectFull.frag > glslparsertest/CorrectConstFolding2.vert > glslparsertest/struct9.frag > glslparsertest/matrix.V110.frag > glslparsertest/CorrectConstruct.vert > glslparsertest/CorrectFull.vert > glslparsertest/TextureRectangleSamplers.frag > glslparsertest/conditional3.frag > glslparsertest/ConstantConversions.frag > glslparsertest/comma2.vert > glslparsertest/function9.frag > glslparsertest/comma1.vert > glslparsertest/CorrectConstFolding1.vert > glslparsertest/dataType19.frag > glslparsertest/dataType4.frag > glslparsertest/function5.frag > glslparsertest/function3.frag > glslparsertest/constFunc.frag > glslparsertest/ParseTest3.frag > glslparsertest/preprocess1.frag > glslparsertest/constructor3.V110.frag > glslparsertest/CorrectUnsizedArray.frag > glslparsertest/comma3.vert These failures are all in common Mesa code and are not specific to r600. There are already bugs reported for most, if not all, of them. Search bugzilla for "glslparser", and it will turn up quite a few.
(In reply to comment #3) > > These failures are all in common Mesa code and are not specific to r600. There > are already bugs reported for most, if not all, of them. Search bugzilla for > "glslparser", and it will turn up quite a few. > I compared my r600 piglit results with swrast (LIBGL_ALWAYS_SOFTWARE=1), and here are the tests that failed in r600, but passed with swrast (there are tests that fail with swrast but pass with r600, or are skipped, I don't list those below): fbo/fbo-copyteximage fbo/fbo-readpixels general/linestipple glean/fbo glean/logicOp glean/paths glean/pointAtten shaders/fp-indirections2 shaders/glsl-bug-22603 shaders/glsl-deriv-varyings shaders/glsl-vs-arrays shaders/glsl-vs-mov-after-deref shaders/vp-address-01 shaders/vp-bad-program shaders/vpfp-generic/vp-arl-constant-array-huge-offset-neg shaders/vpfp-generic/vp-arl-neg-array texturing/depth-tex-modes texturing/texdepth texturing/tfp I also compared with the r300/r500 on the piglit site, and here are the tests that passed for those, but not for r600: glean/logicOp glean/pointAtten shaders/vp-bad-program (this one segfaults!) texturing/texdepth texturing/tfp Should I open bugreports for these, and make them block this one?
See bug 27141 for the reason that vertProg1 and vp-bad-program segfault.
Note: classic r600 driver has been abandoned. Please use r600g (gallium driver) instead. Is this still an issue with a newer driver/kernel?
The classic r600 driver has been abandoned long ago. It was replaced by the Gallium driver r600g. If you have issues with r600g please file a new bug report with component Drivers/Gallium/r600 Thanks.
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.