GCC-4.4.0 fails 7 tests unexpectedly

[Matthew Burgess]

My test logs show GCC-4.4.0 failing the following tests:

gcc.target/i386/avx-vmovntdq-256-1.c (test for excess errors)
gcc.target/i386/avx-vmovntpd-256-1.c (test for excess errors)
gcc.target/i386/avx-vmovntps-256-1.c (test for excess errors)

If fails to produce an executable in each case.

It also fails the following 4 mudflap tests:

libmudflap.c++/pass41-frag.cxx execution test
libmudflap.c++/pass41-frag.cxx (-static) execution test
libmudflap.c++/pass41-frag.cxx (-O2) execution test
libmudflap.c++/pass41-frag.cxx (-O3) execution test

These may be covered under the current note about libmudflap being particularly problematic, but confirming its due to a timeout as the bug we link to suggests would be good.

[Matthew Burgess]

Looks like the avx tests fail due to the version of Binutils we use. See ​http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38906. So, I think a note to this effect is pertinent for 6.5.

[bdubbs@…]

I've been building manually and got the following for gcc:

                === g++ Summary ===

# of expected passes            19702
# of expected failures          142
# of unsupported tests          99
/sources/gcc-build/gcc/testsuite/g++/../../g++  version 4.4.0 (GCC)

                === gcc tests ===
--
                === gcc Summary ===

# of expected passes            56715
# of unexpected failures        3
# of expected failures          188
# of unsupported tests          436
/sources/gcc-build/gcc/xgcc  version 4.4.0 (GCC)

--
                === libgomp Summary ===

# of expected passes            1007
                === libmudflap tests ===


Running target unix
FAIL: libmudflap.c++/pass41-frag.cxx execution test
--
                === libmudflap Summary ===

# of expected passes            1890
# of unexpected failures        4
                === libstdc++ tests ===


Running target unix
--
                === libstdc++ Summary ===

# of expected passes            5848
# of unexpected successes       1
# of expected failures          80
# of unsupported tests          334

FAIL: gcc.target/i386/avx-vmovntdq-256-1.c (test for excess errors)
FAIL: gcc.target/i386/avx-vmovntpd-256-1.c (test for excess errors)
FAIL: gcc.target/i386/avx-vmovntps-256-1.c (test for excess errors)

FAIL: libmudflap.c++/pass41-frag.cxx execution test
FAIL: libmudflap.c++/pass41-frag.cxx (-static) execution test
FAIL: libmudflap.c++/pass41-frag.cxx (-O2) execution test
FAIL: libmudflap.c++/pass41-frag.cxx (-O3) execution test

I think this is OK. Compare to: ​http://gcc.gnu.org/ml/gcc-testresults/2009-05/msg02136.html

However, the time to run make check is *really* long. Somewhere around 40 SBU on my test system. I recommend closing this ticket, but using alfs logs to update all the SBU times and build sizes.

[Matthew Burgess]

Closing, as Bruce's link to upstream test results should be enough for folks not to be too scared by the additional avx* failures. Build logs & SBU times are to be updated by Bruce following his next jhalfs build.