Created attachment 144650 [details] The shader containing code that produces an error and the stack trace by catchsegv. I ran a piglit shader test with the following fragment shader which should render the red solid color image. However, I got an error: Segmentation fault (core dumped). The attachment are the shader and the log obtained when running shader runner with catchsegv tool (by using command “catchsegv ./bin/shader_runner_gles3 shaders/sample.shader_test”). void main() { for( int _injected_loop_counter = 0; 1 != int((injectionSwitch.y)); 1 ) { if(gl_FragCoord.y < 0.0) { return; } } do { _GLF_color = vec4(1.0, 0.0, 0.0, 1.0); if(gl_FragCoord.x < 0.0) { discard; } } while(0.0 > injectionSwitch.y); } [test] uniform vec2 injectionSwitch 0.0 1.0 draw rect -1 -1 2 2 Steps to reproduce: ------------------------------------------------------------------------------- 1. Obtain and build piglit, the Mesa OpenGL test suite runner: https://gitlab.freedesktop.org/mesa/piglit 2. Download the attached archive. 3. From a terminal, execute the supplied tests with the piglit GLES3 shader runner: $ bin/shader_runner_gles3 result.shader_test Build & PC specs: ------------------------------------------------------------------------------- CPU: Intel Core i7-2860QM GPU: Intel Sandybridge Mobile OS: Ubuntu 18.04.2 LTS libdrm: git-46cb2aa1ede16380df2d965ca125d855c4278823 (most recent as of this writing) Mesa: git-d19fe5e67a3945582448748c72b10fd5029d2ed5 (most recent as of this writing) Linux kernel version: 4.18.0-20-generic This bug was found with GraphicsFuzz: https://github.com/google/graphicsfuzz
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/pavel/mesa_versions_64/19.1.0/lib/x86_64-linux-gnu export LIBGL_DRIVERS_PATH="/home/pavel/mesa_versions_64/19.1.0/lib/x86_64-linux-gnu/dri" Hi Jiradet I've also reproduced the issue on my KBL (Intel® UHD Graphics 620) on Ubuntu 18.04 and 5.1.2 version of Kernel. Also, I've checked that the issue isn't actual on 18.3.5 version of the mesa and I bisected the issue: 9dcf90d7ba966cba904266209087a87ea9c018b9 is the first bad commit commit 9dcf90d7ba966cba904266209087a87ea9c018b9 Author: Kenneth Graunke <kenneth@whitecape.org> Date: Wed Apr 3 14:24:31 2019 -0700 intel/fs: Don't emit empty ELSE blocks. While we can clean this up later, it's trivial to not generate the stupid code in the first place, which saves some optimization work. Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Reviewed-by: Matt Turner <mattst88@gmail.com> :040000 040000 6b7fc44def2bb5b72b9470ed3fea74d276c9845f 23d3807532dd2c5c3316807a11996e77bab207f4 M src
will try to look on that
https://gitlab.freedesktop.org/mesa/mesa/merge_requests/1257
Hi all I've tested the issue and it was fixed by this patch - https://patchwork.freedesktop.org/patch/319488/ My configuration is KBL (Intel® UHD Graphics 620) on Ubuntu 18.04, kernel 4.18, patch was applied to the latest master version of the mesa.
Since the patch hasn't merged to the master for now - I reopen the issue until patch merges.
My mistake, I saw that the merge request was closed, and thought it had been merged. Paul, can you try Curro's new patch? https://lists.freedesktop.org/archives/mesa-dev/2019-July/221634.html
Should be fixed in master now. Closing.
I've checked the issue on current master version - fixed.
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.