------- Comment #4 from malitzke at metronets dot com 2007-06-04 21:56 ------- Here is the build machinery used on the powerpc: There were two changes made to prior runs that caused no boot failures:
BUILD was incresed from 11 to 12 form enable-languages c++, fortran were removed as being pointless to demonstrate the boot failure. Note that disable-checking is active. Onthe other two machines the build machinery is architecture and number of processor specific other pretty much the same. #!/bin/sh VERSION=4.3.0 ARCH=${ARCH:-powerpc} TARGET=$ARCH-unknown-linux-gnu BUILD=12 TMP=/var/tmp/gcc_r43 # `mcookie` cd $TMP # build gcc ( mkdir -p build-$BUILD; mkdir -p destdir-$BUILD; cd build-$BUILD; ../gcc-$VERSION/configure \ --prefix=/usr \ --infodir=/usr/share/info \ --mandir=/usr/share/man \ --host=$TARGET \ --build=$TARGET \ --enable-__cxa_atexit \ --enable-threads=posix \ --enable-altivec \ --enable-shared \ --enable-clocale=gnu \ --enable-bootstrap \ --enable-languages=c \ --disable-nls \ --disable-checking \ --disable-werror \ --disable-multilib \ --with-ibmlongdouble \ --with-cpu=7450 \ --enable-clocale=gnu \ --with-system-zlib 2>&1 | tee .Conf # --with-gxx-include-dir=/usr/lib/gcc/$TARGET/$VERSION/include/g++-v4 \ # --enable-version-specific-runtime-libs \ # --includedir=/usr/lib/gcc/$TARGET/$VERSION/include \ # --bindir=/usr/$TARGET/gcc-bin/$VERSION \ # --enable-libffi \ # --enable-ffi \ # --target=$TARGET \ # --enable-mudflap \ # --enable-libgcc-math \ # --enable-secureplt \ # --enable-libgfortran \ # --disable-libmudflap gentoo\ # --datadir=/usr/share/gcc-data/$TARGET \ # --disable-dss \ # --disable-libssp gentoo\ # --disable-libunwind-exceptions \ # --enable-libssp \ # --program-suffix=-$VERS \ # --enable-libgcj-multifile \ # --with-java-home=/usr \ # --disable-libgcj \ # --enable-java-awt=gtk \ # --with-java-home=/usr/lib/jvm/java-1.4.2/jre \ # --enable-languages=c,c++,fortran,objc,obj-c++,java \ # --enable-languages=c,c++,fortran \ # --libexecdir=/usr/libexec/gcc-$VERSION \ # --with-slibdir=/usr/lib/$TARGET/$VERSION \ # --libdir=/usr/lib/$TARGET/$VERSION \ # --enable-checking=release \ # Start the build: nice --adjustment=15 make -j2 BOOT_CFLAGS="-O2 -pipe" STAGE1_CFLAGS="-O -pipe" LIBCFLAGS="-O2 -pipe" \ LIBCXXFLAGS="-O2 -pipe -fno-implicit-templates" \ bootstrap 2>&1 |tee .Build # LIBPATH=/$TARGET/$VERSION bootstrap 2>&1 |tee -a .Build make install DESTDIR=/$TMP/destdir-$BUILD 2>&1 | tee .Inst nice --adjustment=18 make -j2 -i check 2>&1 |tee .Check ) echo echo "powerpc-linux-gnu GCC-$VERSION-$BUILD package build complete!" echo In an earlier instance, which lasted two or three month, it was finally admitted that a bas file had for a short time been on the GCC svn repositiry. this file was not eradicated during further svn updates and so stayed on my system. Given that I intend (when load on the net is lower) to download a complete fresh copy of gcc-4.3.0 to see if this kills the problem. Sorry I overlooked your earlier posting the las too must have crossed each other. -- malitzke at metronets dot com changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |andreast at gcc dot gnu dot | |org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32209