Mirolyub Hristov wrote: > but is there a way to use a "far jump" for the first entry and > not rely on the address being 25 bits away from lj_vm_exit_handler? That's going to be complicated since JIT-compiled code calls various C helper functions, too. So, either move the code up or lift the restriction on mmap()'ing. If this is a BSD-derived OS, then rlimit(RLIMIT_DATA, &rlim) might help (see lj_alloc.c which has a similar issue with FreeBSD/x64). --Mike