gmp 4.3.0 tests fail on Mac OS X 10.5

Torbjorn Granlund tg at
Tue May 5 22:37:36 CEST 2009

Marcus Calhoun-Lopez <marcuscalhounlopez at> writes:

  Even with the new version,  t-asmtype (and others) fails in the same
      make  check-TESTS
      /bin/sh: line 1: 41402 Bus error               ${dir}$tst
      FAIL: t-asmtype
  As requested in a previous message, make try && ./try mpn_add_n gives:
      /usr/bin/gcc-4.0 -DHAVE_CONFIG_H -I. -I. -I../.. -I../.. -I../../
  tests  -I/opt/local/include  -m32 -O2 -pedantic -fomit-frame-pointer - 
  mtune=k8 -march=k8 -c try.c
      /bin/sh ../../libtool --mode=link /usr/bin/gcc-4.0  -m32 -O2 - 
  pedantic -fomit-frame-pointer -mtune=k8 -march=k8  -L/opt/local/lib -o
  try  try.o ../../tests/ ../../
      mkdir .libs
      /usr/bin/gcc-4.0 -m32 -O2 -pedantic -fomit-frame-pointer - 
  mtune=k8 -march=k8 -o .libs/try try.o  -L/opt/local/lib ../../
  tests/.libs/libtests.a /opt/local/var/macports/build/
  libgmp.dylib ../../.libs/libgmp.dylib
      creating try
      Segmentation fault
  After applying the first workaround (changing mpn/x86/p6/aors_n.asm to
  all tests pass.

I cannot reproduce this on my core2 Mac, using the same options as you

/SRCPATH/configure ABI=32 && make && make check

I am at loss guessing what workaround to use here.

  There is no longer a need to apply the second workaround (rm mpn/x86/
If I understand correctly, the snapshot made the popcount problem go


More information about the gmp-bugs mailing list