6.12. GCC-4.0.3 - "make -k check" failed

Ken Moffat ken at linuxfromscratch.org
Thu Feb 22 07:18:02 PST 2007


On Thu, Feb 22, 2007 at 10:02:21AM +0100, Kai Ulrich wrote:
> Some further information
> Before I started to build the for the Basic System
> 
> 1. I removed the old /sources/gcc-build and the /sources/gcc-4.0.3 the 
> direktory.
> 1. I unpack the gcc-4.0.3 sources fresh.
> 
> Kai

 Hi Kai,

  it's good to know that you did that, but the results are so bad
that I've no idea what is wrong.

> > Lot of messages like (see III.) looks like all tests fail.

 I _have_ seen this sort of catastrophic failure, but it was a long
while ago on a rather unusual machine, and I'm not sure if I ever got
to the bottom of it.  In that case, I suspected it was a problem with
devices, so let's start by trying to rule that out.

 In chroot, what do you get for 'ls -l /dev' ?

> > III.)
> > [...]
> > FAIL: gcc.c-torture/execute/20041114-1.c compilation,  -O0
> > FAIL: gcc.c-torture/execute/20041114-1.c compilation,  -O1
> > FAIL: gcc.c-torture/execute/20041114-1.c compilation,  -O2
> > FAIL: gcc.c-torture/execute/20041114-1.c compilation,  -O3 
> > -fomit-frame-pointer
> > FAIL: gcc.c-torture/execute/20041114-1.c compilation,  -O3 -g
> > FAIL: gcc.c-torture/execute/20041114-1.c compilation,  -Os
> > [...]
> > or
> > [...]
> > FAIL: g++.old-deja/g++.jason/member.C  (test for errors, line 8)
> > FAIL: g++.old-deja/g++.jason/member.C  (test for errors, line 15)
> > FAIL: g++.old-deja/g++.jason/member.C  (test for errors, line 21)
> > FAIL: g++.old-deja/g++.jason/member.C  (test for errors, line 27)
> > FAIL: g++.old-deja/g++.jason/member.C  (test for errors, line 28)
> > FAIL: g++.old-deja/g++.jason/member.C  (test for errors, line 33)
> > FAIL: g++.old-deja/g++.jason/member.C  (test for errors, line 34)
> > FAIL: g++.old-deja/g++.jason/member.C (test for excess errors)
> > [...]
> > and ends with
> > make[3]: Leaving directory `/sources/gcc-build/i686-pc-linux-gnu/libmudflap'
> > make[2]: Leaving directory `/sources/gcc-build/i686-pc-linux-gnu/libmudflap'
> > make[1]: *** [check-recursive] Error 1
> > make[1]: Target `check' not remade because of errors.
> > make[1]: Leaving directory `/sources/gcc-build/i686-pc-linux-gnu/libmudflap'
> > make: *** [check-target-libmudflap] Error 2
 That message is normal, even when you get it down to a few errors
it will end up with 'Error 2'.
> > IV.
> > root:/sources/gcc-build# ../gcc-4.0.3/contrib/test_summary | grep -A7 Summ
> >                 === g++ Summary ===
> >
> > # of expected passes            244
> > # of unexpected failures        10222
> > # of unexpected successes       37
> > # of expected failures          30
> > # of unresolved testcases       169
> > # of untested testcases         54
> > --
> >                 === gcc Summary ===
> >
> > # of expected passes            817
> > # of unexpected failures        26885
> > # of unexpected successes       30
> > # of expected failures          20
> > # of unresolved testcases       6771
> > # of untested testcases         504
> > --
> >                 === libmudflap Summary ===
> >
> > # of unexpected failures        412

 On the bright side, gcc and g++ seem to pass a few of the tests,
although libmudflap failed them all.

 If you run the 'test_summary' script without the grep, does it give
any detail, e.g. an 'Error:' message, perhaps because a program is
missing or broken ?  (I suspect it doesn't)

> >                 === libstdc++ tests ===
> >
> >
> > Running target unix
> > ERROR: tcl error sourcing 
> > /sources/gcc-4.0.3/libstdc++-v3/testsuite/libstdc++-abi/abi.exp.

 I know I've seen the abi tests give up in the past, particularly on
powerpc, but i686 usually runs.  Is it possible you missed tcl, or
expect, or dejagnu in chapter 5 ?

 On the face of it, the compiler should not be totally broken,
because gcc appeared to pass 817 tests.  But, maybe they just failed
to give the errors that the testsuite was looking for.  Can you try
the following test, please.

 echo 'main(){}' > dummy.c
 xgcc dummy.c > dummy.log 2>&1
 more dummy.log

 Does it have any interesting error messages in dummy.log ?

ĸen
-- 
das eine Mal als Tragödie, das andere Mal als Farce



More information about the lfs-support mailing list