| Summary: | geometry shader with optional out variable segfaults | ||
|---|---|---|---|
| Product: | Mesa | Reporter: | pavol |
| Component: | glsl-compiler | Assignee: | Ian Romanick <idr> |
| Status: | RESOLVED FIXED | QA Contact: | Intel 3D Bugs Mailing List <intel-3d-bugs> |
| Severity: | normal | ||
| Priority: | medium | CC: | eero.t.tamminen, pavol |
| Version: | git | ||
| Hardware: | x86-64 (AMD64) | ||
| OS: | Linux (All) | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: |
Demo of bug
Valgrind output GLSL dump and backtrace Geometry shader input Piglit test Typos patch Patch that may fix the problem |
||
Please provide more details about how to reproduce the problem. Ideal would be if you could attach a simple test program which we can compile and run, or at least a command line to reproduce the problem with a standard tool and the geometry shader source code you provided. Created attachment 105265 [details]
Demo of bug
Program should be valid according GL spec. I'm hitting this assertion failure in the GLSL IR to TGSI translation code first: ../../../src/mesa/state_tracker/st_glsl_to_tgsi.cpp:2157:visit: Assertion `var->data.location != -1' failed. Created attachment 105268 [details] Valgrind output This is output from valgrind, mesa is from here https://launchpad.net/~oibaf/+archive/ubuntu/graphics-drivers Created attachment 105270 [details]
GLSL dump and backtrace
With Intel we get following backtrace .. I don't have good enough clue to tell what's wrong but I can see that the reg and reg_offset values clearly go over the size of attribute_map.
--- 8< -----
(gdb)
#0 0x00007ffff73683e2 in brw::vec4_visitor::lower_attributes_to_hw_regs (this=0xffff, this@entry=0x7fffffffcdf0, attribute_map=attribute_map@entry=0x7fffffffc770,
interleaved=false) at brw_vec4.cpp:1501
#1 0x00007ffff736f6a0 in brw::vec4_gs_visitor::setup_payload (this=0x7fffffffcdf0) at brw_vec4_gs_visitor.cpp:134
#2 0x00007ffff73691ec in brw::vec4_visitor::run (this=this@entry=0x7fffffffcdf0) at brw_vec4.cpp:1813
#3 0x00007ffff737077f in brw::brw_gs_emit (brw=brw@entry=0x7ffff6daf040, prog=prog@entry=0xe0e070, c=c@entry=0x7fffffffda50, mem_ctx=mem_ctx@entry=0xae59d0,
final_assembly_size=final_assembly_size@entry=0x7fffffffda4c) at brw_vec4_gs_visitor.cpp:648
#4 0x00007ffff736e7b9 in do_gs_prog (brw=brw@entry=0x7ffff6daf040, prog=prog@entry=0xe0e070, gp=gp@entry=0xdab920, key=key@entry=0x7fffffffdd30) at brw_vec4_gs.c:264
#5 0x00007ffff736ebb8 in brw_gs_precompile (ctx=0x7ffff6daf040, prog=0xe0e070) at brw_vec4_gs.c:401
#6 0x00007ffff73607ff in brw_shader_precompile (prog=0xe0e070, ctx=0x7ffff6daf040) at brw_shader.cpp:76
#7 brw_link_shader (ctx=0x7ffff6daf040, shProg=0xe0e070) at brw_shader.cpp:269
#8 0x00007ffff7213a26 in _mesa_glsl_link_shader (ctx=0x7ffff6daf040, prog=0xe0e070) at ../../src/mesa/program/ir_to_mesa.cpp:3038
This is a separate shader objects bug, and is probably related to 79783. If you modify the demo application to not use SSO, it works fine. I have attached two patches, one adds piglit test and another fixes some minor typos. Created attachment 109813 [details] [review] Geometry shader input Piglit test Created attachment 109814 [details] [review] Typos patch Created attachment 110323 [details] [review] Patch that may fix the problem I just sent better patches to the mesa-dev mailing list: http://lists.freedesktop.org/archives/mesa-dev/2014-December/071748.html http://lists.freedesktop.org/archives/mesa-dev/2014-December/071749.html Fixed in master by the following commits. It now occurs to me that I should have tagged these for stable branches. :( commit a909b995d95892798a189818454905fdefd4bc9b Author: Ian Romanick <ian.d.romanick@intel.com> Date: Mon Dec 1 14:07:30 2014 -0800 linker: Assign varying locations geometry shader inputs for SSO Previously only geometry shader outputs would be assigned locations if the geometry shader was the only stage in the linked program. Signed-off-by: Ian Romanick <ian.d.romanick@intel.com> Cc: pavol@klacansky.com Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=82585 Reviewed-by: Jordan Justen <jordan.l.justen@intel.com> commit 5eca78a00a5de442aaf541a1095d5cfa6b4f45de Author: Ian Romanick <ian.d.romanick@intel.com> Date: Mon Dec 1 14:16:24 2014 -0800 linker: Wrap access of producer_var with a NULL check producer_var could be NULL if consumer_var is not NULL and consumer_is_fs is false. This will occur when the producer is NULL and the consumer is the geometry shader for a program that contains only a geometry shader. This will occur starting with the next patch. Signed-off-by: Ian Romanick <ian.d.romanick@intel.com> Cc: pavol@klacansky.com Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=82585 Reviewed-by: Jordan Justen <jordan.l.justen@intel.com> Thank you Ian. Where should I submit the piglit patch please? (In reply to pavol from comment #15) > Thank you Ian. Where should I submit the piglit patch please? The piglit mailing list: piglit@lists.freedesktop.org. |
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.
Running this shader in geometry stage, it segfaults: #version 330 layout(points) in; layout(points, max_vertices = 1) out; in gl_PerVertex { vec4 gl_Position; float gl_PointSize; float gl_ClipDistance[]; } gl_in[]; in vec4 v_colour[]; out vec4 g_colour; void main() { for(int i = 0; i < 1; i++) { gl_Position = gl_in[i].gl_Position; gl_PrimitiveID = gl_PrimitiveIDIn; /* TODO report bug */ g_colour = v_colour[i]; EmitVertex(); } EndPrimitive(); } Following message is from Valgrind: EE ../../../../../../src/gallium/drivers/r600/r600_shader.c:353 tgsi_is_supported - unsupported src 0 (dimension 1) EE ../../../../../../src/gallium/drivers/r600/r600_shader.c:157 r600_pipe_shader_create - translation from TGSI failed ! EE ../../../../../../src/gallium/drivers/r600/r600_state_common.c:750 r600_shader_select - Failed to build shader variant (type=2) -22