Interpreting gcc-3.3 "make check" results

[Dan Kegel <dkegel at ixiacom dot com>]

Ahh.  I feel *much* better now.  I actually gcc's "make check"
against a ppc750 system remotely.  Life is wonderful!

But now I have to interpret the results, and they're not all good.

I'm testing gcc-3.3/glibc-2.3.2, and I forgot to use the -k option
to 'make check', so it terminated on the first failure.

On my x86 system, I got

FAIL: 27_io/istream_unformatted.cc execution test

I should look into that, but I need to focus on the ppc stuff first.

On my ppc-750 system, I got

gcc/testsuite/g++.log:FAIL: g++.dg/bprob/g++-bprob-1.C compilation,  -g  -fprofile-arcs
gcc/testsuite/g++.log:FAIL: g++.dg/bprob/g++-bprob-1.C compilation,  -O0  -fprofile-arcs
gcc/testsuite/g++.log:FAIL: g++.dg/bprob/g++-bprob-1.C compilation,  -O1  -fprofile-arcs
gcc/testsuite/g++.log:FAIL: g++.dg/bprob/g++-bprob-1.C compilation,  -O2  -fprofile-arcs
gcc/testsuite/g++.log:FAIL: g++.dg/bprob/g++-bprob-1.C compilation,  -O3  -fprofile-arcs
gcc/testsuite/g++.log:FAIL: g++.dg/bprob/g++-bprob-1.C compilation,  -O3 -g  -fprofile-arcs
gcc/testsuite/g++.log:FAIL: g++.dg/bprob/g++-bprob-1.C compilation,  -Os  -fprofile-arcs
gcc/testsuite/g++.log:FAIL: g++.dg/abi/mangle17.C (test for excess errors)

That first batch of problems was a link failure:
/tmp/cchpc6Vx.o(.text+0x18dc): In function `_GLOBAL__I_for_val1GCOV':
gcc-3.3/gcc/testsuite/g++.dg/bprob/g++-bprob-1.C:234: undefined reference to `__bb_init_func'

So maybe profiling isn't implemented.  Tsk.  Guess I'll check older versions
and see if this is a regression.

The mangle17.C failure was the following:
gcc-3.3/gcc/testsuite/g++.dg/abi/mangle17.C:8: sorry, unimplemented: real-valued template parameters when cross-compiling
gcc-3.3/gcc/testsuite/g++.dg/abi/mangle17.C:8: warning: the mangled name of `void f(S<(e + (int)(3.70000000000000017763568394002504646778106689453e+0))>) [with int I = 7]' will change in a future 
version of GCC
gcc-3.3/gcc/testsuite/g++.dg/abi/mangle17.C:11: warning: the mangled name of `void g(S<(e + (int)(3.70000000000000017763568394002504646778106689453e+0))>) [with int I = 7]' will change in a future 
version of GCC

What madman uses real-valued template parameters?  That one I won't worry about,
and I'm happy it showed up, since it gives me a chance to learn how to
exclude test cases :-)

Anyone run into the above?  Or should I take this level of detail to the
main gcc list?

- Dan



------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com