http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57455
--- Comment #2 from Artyom <theartlav at gmail dot com> --- > Can you try to get at a backtrace using gdb? Looks like i found the problem. gcc -wrapper gdb,--args cmathmodule.i ... Program received signal SIGFPE, Arithmetic exception. 0xb7f17c31 in __gmpn_divrem_2 () from /usr/lib/libgmp.so.10 (gdb) bt #0 0xb7f17c31 in __gmpn_divrem_2 () from /usr/lib/libgmp.so.10 #1 0xed12ce80 in ?? () #2 0xbcfb951c in ?? () #3 0x6dd89843 in ?? () #4 0x4004abe0 in ?? () #5 0x55b8c549 in ?? () #6 0x8b827a40 in ?? () #7 0x00000001 in ?? () #8 0xb7fef9c9 in ___tls_get_addr_internal () from /lib/ld-linux.so.2 #9 0xb7f75618 in mpfr_mul () from /usr/lib/libmpfr.so.4 Backtrace stopped: previous frame inner to this frame (corrupt stack?) Rebuilt GMP, no difference. Rebuilt MPFR, bug gone. Previously MPFR was built in a chroot on another machine with AMD CPU, maybe it picked up some wrong processor optimizations, that fail on Intel's? In any case, rebuilding MPFR natively removed the problem. Sorry for a false alarm.
