That is quite likely an unimplemented FPU opcode triggered by a call to udivsi in the libc. Meaning: you compiled with -ffast-math, hence the compiler will generate FPU opcodes whenever it sees it feasible and will not use any mathematics crutch like a software
I doubt -ffast-math generates FPU code unless -m68020-60 or an equivalent option is given on the commandline. If -ffast-math has something to do with that, it's probably some math rounding error. It's definitely worth a try, to remove it.