Optimized Build Failure in gcc-ss-20000925

[David Ronis]

An optimized build of the current snapshot fails in the compare stage
(I think this has been happening in the last 3 or so snapshots).  The
build fails in the compare step.  I'm on a
i686-linux(2.2.17)-gnu(2.1.3) box.  Using the following CFLAGS (set
before configuring):

-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

The compare fails in java/parse.c.  The files differ in size:

{} ls -l java/parse.o
-rw-r--r--   1 ronis    ronis      431012 Oct  3 10:57 java/parse.o
{} ls -l stage2/java/parse.o
-rw-r--r--   1 ronis    ronis      431164 Oct  3 10:25 stage2/java/parse.o

They were compiled as:

../stage1/xgcc -B../stage1/ -B/usr/i686-pc-linux-gnu/bin/ -c -DIN_GCC
-O2 -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 -W -Wall -Wwrite-strings
-Wstrict-prototypes -Wmissing-prototypes -Wno-long-long -W -Wall
-I. -I.. -I../../../gcc/gcc/java
-I../../../gcc/gcc/java/.. -I../../../gcc/gcc/java/../config
-I../../../gcc/gcc/java/../../include
-I../../../gcc/gcc/java/../../zlib ../../../gcc/gcc/java/parse.c


and 

../stage2/xgcc -B../stage2/ -B/usr/i686-pc-linux-gnu/bin/ -c -DIN_GCC
-O2 -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 -W -Wall -Wwrite-strings
-Wstrict-prototypes -Wmissing-prototypes -Wno-long-long -W -Wall
-I. -I.. -I../../../gcc/gcc/java
-I../../../gcc/gcc/java/.. -I../../../gcc/gcc/java/../config
-I../../../gcc/gcc/java/../../include
-I../../../gcc/gcc/java/../../zlib ../../../gcc/gcc/java/parse.c

Finally, a build with default parameters works, although there are a
very large number of g++ testsuite failures.  I posted the results at:

     http://gcc.gnu.org/ml/gcc-testresults/2000-10/msg00016.html

These actually confuse me quite a bit, since others on similar
machines get far fewer failures.  My suspicion is that most of these
are related to the large failure rates in the libio and libstdc++
(100%) tests.  In turn, these are probably due to some problem with
system libraries, etc., at my end.  Here's what I've got installed
that seems relevant:

Kernel modutils (2.3.7) insmod version 2.3.17
Gnu C (2.7.2.3) 2.95.2
Binutils  GNU ld version 2.10.90 (with BFD 2.10.90)
Linux libc6 C /lib/libc.so.6 -> libc-2.1.3.so
Dynamic Linker  ldd (GNU libc) 2.1.3
procps version 2.0.6
Procinfo version 17 (1999-05-16)
Psmisc  psmisc version 18
Sh-utils basename (GNU sh-utils) 1.16
Bash version 2.04.0(1)-release (i386-slackware-linux-gnu)

David