(r_smp) Crash when running Pulse on Material system

[illwieckz]

Crash when running Pulse on Material system, I was running this branch:

• Rework material system material and surface generation #1645

But haven't compared with master yet.

Thread 19 "render thread" received signal SIGSEGV, Segmentation fault.
0x000056adcd76a9b0 in Poly2dIndexedCommand::ExecuteSelf (this=0x79adb84b7aac) at src/engine/renderer/tr_backend.cpp:4905
4905			tess.verts[ tess.numVertexes ].xyz[ 1 ] = verts[ i ].xyz[ 1 ] + translation[ 1 ];

Thread 19 (Thread 0x79ada6a006c0 (LWP 3545100) "render thread"):
#0  0x000056adcd76a9b0 in Poly2dIndexedCommand::ExecuteSelf (this=0x79adb84b7aac) at src/engine/renderer/tr_backend.cpp:4905
#1  0x000056adcd76c599 in RB_ExecuteRenderCommands (data=<optimized out>) at src/engine/renderer/tr_backend.cpp:5590
#2  0x000056adcd76c622 in RB_RenderThread () at src/engine/renderer/tr_backend.cpp:5621
#3  0x000056adcd81b9d6 in GLimp_RenderThreadWrapper () at src/engine/sys/sdl_glimp.cpp:236
#4  0x000079ade5f14296 in ?? () from /lib/x86_64-linux-gnu/libSDL2-2.0.so.0
#5  0x000079ade509caa4 in start_thread (arg=<optimized out>) at ./nptl/pthread_create.c:447
#6  0x000079ade5129c3c in clone3 () at ../sysdeps/unix/sysv/linux/x86_64/clone3.S:78

[illwieckz]

Also this:

Thread 19 "render thread" received signal SIGSEGV, Segmentation fault.
0x0000644a397362bb in Tess_AddQuadStampExt (origin=0x7477066d4f80, left=left@entry=0x7476f9dff7a0, up=up@entry=0x7476f9dff7b0, color=..., s1=s1@entry=0, t1=t1@entry=0, s2=s2@entry=1, t2=t2@entry=1) at src/engine/renderer/tr_surface.cpp:235
235		tess.indexes[ tess.numIndexes ] = ndx;

Thread 19 (Thread 0x7476f9e006c0 (LWP 3582178) "render thread"):
#0  0x0000644a397362bb in Tess_AddQuadStampExt (origin=0x7477066d4f80, left=left@entry=0x7476f9dff7a0, up=up@entry=0x7476f9dff7b0, color=..., s1=s1@entry=0, t1=t1@entry=0, s2=s2@entry=1, t2=t2@entry=1) at src/engine/renderer/tr_surface.cpp:235
#1  0x0000644a39736ac2 in Tess_AddQuadStamp (color=..., up=0x7476f9dff7b0, left=0x7476f9dff7a0, origin=<optimized out>) at src/engine/renderer/tr_surface.cpp:304
#2  Tess_SurfaceSprite () at src/engine/renderer/tr_surface.cpp:633
#3  Tess_SurfaceEntity () at src/engine/renderer/tr_surface.cpp:1351
#4  0x0000644a396927c8 in RB_RenderDrawSurfaces (fromSort=fromSort@entry=shaderSort_t::SS_ENVIRONMENT_NOFOG, toSort=toSort@entry=shaderSort_t::SS_POST_PROCESS, drawSurfFilter=drawSurfFilter@entry=DRAWSURFACES_ALL) at src/engine/renderer/tr_backend.cpp:950
#5  0x0000644a396a0b30 in RB_RenderView (depthPass=<optimized out>) at src/engine/renderer/tr_backend.cpp:4542
#6  DrawViewCommand::ExecuteSelf (this=0x7477072b6234) at src/engine/renderer/tr_backend.cpp:5378
#7  0x0000644a39695599 in RB_ExecuteRenderCommands (data=<optimized out>) at src/engine/renderer/tr_backend.cpp:5590
#8  0x0000644a39695622 in RB_RenderThread () at src/engine/renderer/tr_backend.cpp:5621
#9  0x0000644a397449d6 in GLimp_RenderThreadWrapper () at src/engine/sys/sdl_glimp.cpp:236
#10 0x0000747734dde296 in ?? () from /lib/x86_64-linux-gnu/libSDL2-2.0.so.0
#11 0x0000747733e9caa4 in start_thread (arg=<optimized out>) at ./nptl/pthread_create.c:447
#12 0x0000747733f29c3c in clone3 () at ../sysdeps/unix/sysv/linux/x86_64/clone3.S:78

[illwieckz]

I reproduce it on master.

[slipher]

I can't repro. Which version of Pulse do you have?

[VReaperV]

Hmm, I don't think I touched any of the related code anytime recently.

[illwieckz]

I will paste the link there, with my config as well.

I just reproduced it on v0.55.2 tag anyway, so this doesn't block the release. It was already broken.

[VReaperV]

I would also be surprised if this depends on the map because AFAIK that code is only used for 2D drawing, so things like UI.

[illwieckz]

Interesting, anyway there is no urge anymore. I just noticed I had a stray r_smp 1.

With r_smp 0 I don't reproduce the crash on v0.55.2 and on master, and r_smp is disabled by default. So we don't have to fix it for the point release. It is not blocking.

I backuped my config, I will post my config and the map later when I'll have time for that.

[VReaperV]

It seems... unlikely that this is caused by material system, as there aren't any functions used by it in the call stacks posted above.