Crash when repeatedly using `SDL_RenderCopy`

SDL Development
1

I’m having an issue that after a certain number of calls to SDL_RenderCopy I get a segfault. The program is working fine up until this point: it draws the expected elements without any glitches.

The stack trace is deep within the driver unfortunately:

#0  0x00007fffe548a167 in ?? () from /usr/lib/x86_64-linux-gnu/dri/r600_dri.so
...
#11 0x00007fffe5155f87 in ?? () from /usr/lib/x86_64-linux-gnu/dri/r600_dri.so
#12 0x00007fffece6bc06 in ?? () from /usr/lib/x86_64-linux-gnu/libSDL2.so
#13 0x00007fffece633b1 in ?? () from /usr/lib/x86_64-linux-gnu/libSDL2.so
#14 0x00007ffff7ff5877 in _init_module_sdl_render_blit_18 ()
#15 0x00007ffff7ff4a9e in _init_module_run_26 ()
#16 0x00007ffff7ff348d in _init_module_main_38 ()
...

Perhaps I’m missing some kind of “free” operation? My general flow of events, in the main loop is:

  • while SDL_PollEvent(e) != 0
  • SDL_SetRenderDrawColor
  • SDL_RenderClear
  • SDL_GetWindowSize
  • SDL_RenderCopy
  • SDL_RenderPresent

My actual code is written in Leaf, my experimental language. Given the consistency and location of the failure though, I don’t think I have an API/ABI interface issue.

Is there something I’m missing? If not, any ideas on how I can go about debugging such an issue?

Platform: Ubuntu 16.04
Device: AMD JUNIPER (DRM 2.43.0 / 4.4.0-116-generic, LLVM 5.0.0) (0x68bf)

2

Note, there’s still a > 90% chance this is my compiler’s fault. I’m just hoping somebody may have insight in the odd chance this happens to be an SDL issue.

3

Sorry, you can disregard this. I’ve identified an errant alloca inside the loop. Thus I eventually run out of stack space. The consistent backtrace appearing inside the driver/SDL is most likely because it’s a deep stack and the first thing to notice there is no stack left.