4 tests failed on gmp 5.0.3 using mingw64 gcc 4.7

[Sisyphus]

----- Original Message ----- 
From: "Rene Sugar" <rene.sugar at gmail.com>
To: <gmp-bugs at gmplib.org>
Sent: Wednesday, February 08, 2012 9:00 AM
Subject: 4 tests failed on gmp 5.0.3 using mingw64 gcc 4.7


>
> I got the following 4 errors on gmp 5.0.3 when running "make check" on a
> mingw64 build on Windows 7 64-bit.
>
> ./configure --host=x86_64-w64-mingw32 --build=x86_64-w64-mingw32
> --target=x86_64-w64-mingw32 --prefix=/mingw
> make
> make check
>
> My CPU is an AMD Phenom II X4 940 processor. The configure script picked
> "k8" instead of "amdfam10".
>
>
> mpz_mul_si wrong
>   x=1
>   y=-2147483648 (0x80000000)
>   got =-2147483648
>   want=-18446744071562067968
>
> This application has requested the Runtime to terminate it in an unusual
> way.
> Please contact the application's support team for more information.
> FAIL: t-mul_i.exe
>
>
> mpz_set_si wrong on data[5]
>
> This application has requested the Runtime to terminate it in an unusual
> way.
> Please contact the application's support team for more information.
> FAIL: t-set_si.exe
>
>
> mpz_fits_slong_p (INT_MIN) got 0 want 1
>  z size -1
>  z dec -18446744071562067968
>  z hex -ffffffff80000000
> mpz_fits_sint_p (INT_MIN) got 0 want 1
>  z size -1
>  z dec -18446744071562067968
>  z hex -ffffffff80000000
> mpz_fits_slong_p (LONG_MIN) got 0 want 1
>  z size -1
>  z dec -18446744071562067968
>  z hex -ffffffff80000000
>
> This application has requested the Runtime to terminate it in an unusual
> way.
> Please contact the application's support team for more information.
> FAIL: t-fits.exe
>
>
> mpz_cmp_si wrong on data[12]
>   a=-2147483648
>   b=-2147483648
>   got=1
>   want=0
>
> This application has requested the Runtime to terminate it in an unusual
> way.
> Please contact the application's support team for more information.
> FAIL: t-cmp_si.exe
>
>
> Here is the build of gcc I used:
>
> (mingw64 project also lets you cross-compile mingw32 and mingw64 builds
> on Linux)
>
> mingw64 toolchain sources with build scripts to create cross-compilers:
>
> http://sourceforge.net/projects/mingw-w64/files/Toolchain%20sources/Personal%20Builds/rubenvb/
>
> mingw64 toolchain binaries:
>
> http://sourceforge.net/projects/mingw-w64/files/Toolchains%20targetting%20Win64/Personal%20Builds/rubenvb/4.7.0-1/
>

I have similar to you (but, thankfully,  no 'make check' failures).
I'm using Vista 64-bit, and my cpu is AMD Athlon X2 4600+.
I, too, have a 64-bit build of gcc-4.7.0, but it's one of the pre-compiled 
mingw-w64 "automated builds" - where the names of the executables (gcc, g++, 
ar, ld, etc.) are prefixed with "x86_64-w64-mingw32-".

I even re-built gmp-5.0.3 using essentially the same configure command as 
you:

$ 
./configure --host=x86_64-w64-mingw32 --build=x86_64-w64-mingw32 --target=x86_64-w64-mingw32 
CC=x86_64-w64-mingw32-gcc CXX=x86_64-w64-mingw32-g++ 
AR=x86_64-w64-mingw32-ar LD=x86_64-w64-mingw32-ld NM=x86_64-w64-mingw32-nm 
RANLIB=x86_64-w64-mingw32-ranlib OBJDUMP=x86_64-w64-mingw32-objdump 
STRIP=x86_64-w64-mingw32-strip

All of that additional stuff is just to ensure that the right CC, AR, LD, 
etc. get found.

Comparing our config.log files, I didn't see anything that would account for 
your failing 'make check' results.

Near the end of your config.log I did see some things that I didn't expect - 
though it takes "a fair stretch of the imagination" to relate them to the 
test failures.

You got:
AR='x86_64-w64-mingw32-ar'
AS='x86_64-w64-mingw32-as'
LD='c:/mingw64/x86_64-w64-mingw32/bin/ld.exe'
NM='/mingw/bin/nm'
OBJDUMP='x86_64-w64-mingw32-objdump'
RANLIB='x86_64-w64-mingw32-ranlib'
STRIP='x86_64-w64-mingw32-strip'

For your build of gcc-4.7.0 I expected to see:
AR='ar'
AS='as'
LD='ld'
NM='nm'
OBJDUMP='objdump'
RANLIB='ranlib'
STRIP='strip'

I thought that the "personal builds" didn't use the "x86_64-w64-mingw32-" 
prefix ?
(And it's a bit odd that LD and NM saw fit to specify a path.)

Cheers,
Rob