I found this segfault/crash launching kov2p, in versions 0.2.97.40 and 0.2.97.42. The stack trace is from the libretro version.
I reported it to the libretro port maintainer but I think I confused them, and it might be helpful if people who are more familiar with the codebase can look at this, since it's likely a memory safety issue or overflow in the emulation code.
Thread 1 "retroarch" received signal SIGSEGV, Segmentation fault.
0x00007fffe6778497 in Arm7ReadByte (addr=939524096) at src/cpu/arm7_intf.cpp:208
208 return membase[READ][addr >> PAGE_SHIFT][addr & PAGE_BYTE_AND];
(gdb) bt
#0 0x00007fffe6778497 in Arm7ReadByte (addr=939524096) at src/cpu/arm7_intf.cpp:208
#1 0x00007fffe677fb1d in arm7_cpu_read8 (addr=939524096) at src/cpu/arm7/arm7core.c:187
#2 0x00007fffe6788eaa in Arm7Run (cycles=277) at src/cpu/arm7/arm7exec.c:802
#3 0x00007fffe61a09c0 in pgmFrame () at src/burn/drv/pgm/pgm_run.cpp:873
#4 0x00007fffe5d4eade in BurnDrvFrame () at src/burn/burn.cpp:719
#5 0x00007fffe68ea5b0 in ForceFrameStep () at src/burner/libretro/libretro.cpp:795
#6 0x00007fffe68ebcb0 in retro_run () at src/burner/libretro/libretro.cpp:1317
#7 0x000000000040d819 in core_run () at core_impl.c:415
#8 0x0000000000411e35 in runloop_iterate (sleep_ms=sleep_ms@entry=0x7fffffffe8ec) at retroarch.c:3051
#9 0x000000000040c732 in rarch_main (argc=1, argv=<optimized out>, data=0x0) at frontend/frontend.c:129
#10 0x00007ffff55f64ca in __libc_start_main () from /usr/lib/libc.so.6
#11 0x0000000000409fba in _start ()
I can reproduce this consistently on one of my machines, but I've found that reproducing it in general is highly dependent on the operating system/host CPU, among other factors. kov and kovplus don't trigger it, but kov2p, kovsh, and theglad do.