Created attachment 122158 [details] Makefile Hello. I'm using the following setup: Extended renderer info (GLX_MESA_query_renderer): Vendor: Intel Open Source Technology Center (0x8086) Device: Mesa DRI Intel(R) Sandybridge Mobile (0x116) Version: 11.1.2 Accelerated: yes Video memory: 1536MB Unified memory: yes Preferred profile: core (0x1) Max core profile version: 3.3 Max compat profile version: 3.0 Max GLES1 profile version: 1.1 Max GLES[23] profile version: 3.0 OpenGL vendor string: Intel Open Source Technology Center OpenGL renderer string: Mesa DRI Intel(R) Sandybridge Mobile OpenGL core profile version string: 3.3 (Core Profile) Mesa 11.1.2 OpenGL core profile shading language version string: 3.30 OpenGL core profile context flags: (none) OpenGL core profile profile mask: core profile Linux copperhead 4.4.3-1-ARCH #1 SMP PREEMPT Fri Feb 26 15:09:29 CET 2016 x86_64 GNU/Linux The following small program will trigger an assertion failure when calling glDrawArrays(). The problem seems to be triggered by declaring a vec2 fragment shader output, and writing to it, but without having an associated color attachment on the framebuffer and with all framebuffer draw buffers set to GL_NONE. I've included sources as attachments, but you may prefer the git repository here: https://github.com/io7m/mesa-intel-bug-20160308 Just "make && ./bug". I've also included an apitrace.
Created attachment 122159 [details] Main bug source
Created attachment 122160 [details] Bug API trace
Created attachment 122161 [details] Header for example
Created attachment 122162 [details] Small GLFW frontend for example
The actual output of the program is: GL_VERSION: 3.3 (Core Profile) Mesa 11.1.2 ASSERT: Scalar FS validation failed! load_payload(8) vgrf2:UD, (null):UD(null):UD(null):UD, vgrf0+3.0:F brw_fs_validate.cpp:53: inst->src[i].reg_offset + inst->regs_read(i) <= (int)alloc.sizes[inst->src[i].nr] Aborted (core dumped)
This seems to have been fixed for a while. I tracked the fix down to this commit: commit 7f53fead5cf9a85c74a94d359dd5fccfbb87856c Author: Timothy Arceri <timothy.arceri@collabora.com> Date: Mon May 23 16:48:05 2016 +1000 i965: enable component packing for vs and fs Rather than trying to work out the total number of components used at a location we simply treat all outputs as vec4s. This removes the need for complex code looping over varyings to match packed locations and the need for storing the total number of components used at each location. Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com> Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
I'm now on: Extended renderer info (GLX_MESA_query_renderer): Vendor: Intel Open Source Technology Center (0x8086) Device: Mesa DRI Intel(R) Sandybridge Mobile (0x116) Version: 12.0.3 Accelerated: yes Video memory: 1536MB Unified memory: yes Preferred profile: core (0x1) Max core profile version: 3.3 Max compat profile version: 3.0 Max GLES1 profile version: 1.1 Max GLES[23] profile version: 3.0 OpenGL vendor string: Intel Open Source Technology Center OpenGL renderer string: Mesa DRI Intel(R) Sandybridge Mobile OpenGL core profile version string: 3.3 (Core Profile) Mesa 12.0.3 OpenGL core profile shading language version string: 3.30 OpenGL core profile context flags: (none) OpenGL core profile profile mask: core profile Linux copperhead 4.8.4-1-ARCH #1 SMP PREEMPT Sat Oct 22 18:26:57 CEST 2016 x86_64 GNU/Linux ... and I can still reproduce this problem: $ ./bug GL_VERSION: 3.3 (Core Profile) Mesa 12.0.3 ASSERT: Scalar FS validation failed! load_payload(8) vgrf3:UD, (null):UD(null):UD(null):UD, vgrf0+3.0:F brw_fs_validate.cpp:53: inst->src[i].reg_offset + inst->regs_read(i) <= (int)alloc.sizes[inst->src[i].nr] Aborted (core dumped) Has that commit actually made it into a release yet?
(In reply to org.freedesktop from comment #7) > Has that commit actually made it into a release yet? It's in mesa-13.0.0 which was just released yesterday. :)
Ah, thanks. I was going to say that I'd wait for my distro to catch up, but it seems like they just pushed new 13.0.0 packages out today.
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.