This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Large increase in testsuite failures in optimized 3.0 20010613 (prerelease)
- To: gcc-bugs at gcc dot gnu dot org
- Subject: Large increase in testsuite failures in optimized 3.0 20010613 (prerelease)
- From: David Ronis <ronis at ronispc dot chem dot mcgill dot ca>
- Date: Thu, 14 Jun 2001 09:31:51 -0400
- Reply-To: ronis at onsager dot chem dot mcgill dot ca
I just finished make check for an i686-linux-gnu box using optimized
CFLAGS for the bootstrap build (specifically CFLAGS(set before
configure)=-O3 -march=i686 -funroll-loops -ffast-math
-fomit-frame-pointer -fforce-mem -fforce-addr -malign-jumps=3
-malign-functions=3 -malign-loops=3 -mpreferred-stack-boundary=3).
There were a huge number of testsuite failures, including many in
libstdc++, which may explain most of the errors in g++-v3. While
there have consistently been more failures in the optimized built gcc,
there haven't been this many in some time. I've sent the summary to
the testsuite list and it can be seen at
http://gcc.gnu.org/ml/gcc-testresults/2001-06/msg00330.html.
The testsuite logs show many of the following warning:
system /usr/i686-pc-linux-gnu/include -ggdb3 -DDEBUG_ASSERT -ffunction-sections
-fdata-sections -nostdinc++ -I/home/ronis/Project/notar/objdir/i686-pc-linux-gnu
/libstdc++-v3/include -I/home/ronis/Project/notar/gcc/libstdc++-v3/include -I/ho
me/ronis/Project/notar/gcc/libstdc++-v3/include/std -I/home/ronis/Project/notar/
gcc/libstdc++-v3/include/c_std -I/home/ronis/Project/notar/gcc/libstdc++-v3/libs
upc++ -I/home/ronis/Project/notar/gcc/libstdc++-v3/libio -I/home/ronis/Project/n
otar/gcc/libstdc++-v3/testsuite -I/home/ronis/Project/notar/gcc/libstdc++-v3/inc
lude/backwards -I/home/ronis/Project/notar/gcc/libstdc++-v3/include/ext /home/ro
nis/Project/notar/gcc/libstdc++-v3/testsuite/17_intro/header_cassert.cc -DDEB
UG_ASSERT -lm -o ./header_cassert (timeout = 300)
/usr/i686-pc-linux-gnu/bin/ld: warning: libgcc_s.so.0, needed by /usr/lib/gcc-li
b/i686-pc-linux-gnu/3.0/../../../libm.so, may conflict with libgcc_s.so.1
output is:
/usr/i686-pc-linux-gnu/bin/ld: warning: libgcc_s.so.0, needed by /usr/lib/gcc-li
b/i686-pc-linux-gnu/3.0/../../../libm.so, may conflict with libgcc_s.so.1
FAIL: 17_intro/header_cassert.cc (test for excess errors)
Excess errors:
/usr/i686-pc-linux-gnu/bin/ld: warning: libgcc_s.so.0, needed by /usr/lib/gcc-li
b/i686-pc-linux-gnu/3.0/../../../libm.so, may conflict with libgcc_s.so.1
PASS: 17_intro/header_cassert.cc execution test
Would a make install fix these problems? Shouldn't be necessary, of course.
Finally, I had some trouble with the version of dejagnu I was running
and upgraded (first to 1.4.1 and then back to 1.4.0, cf Joseph Meyers'
comment at http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00639.html), so I
suppose there could still be a problem with dejagnu. I'm running a
nonoptimized bootstrap now and will post the results soon.
David