ar16 wrote: > known issue? if not, how can I coax helpful trace info out of this? Not a known issue. I have no explanation for that "table overflow". Maybe you have some LUA_* environment variables set? Try: env - /usr/local/bin/luajit Or try to find out where this error is triggered: gdb luajit break lj_err_throw run bt Otherwise look for mmap failures here: strace luajit [Note that Valgrind 3.9.0 has chosen to block MAP_32BIT, which breaks LuaJIT on Linux/x64: https://bugs.kde.org/show_bug.cgi?id=324181 Older versions of Valgrind still work. It wasn't that useful to debug LuaJIT on x64, anyway, due to the mandatory use of the builtin allocator. But none of that relates to your problem.] --Mike