Thanks for your help Dennis. It's taken me a little while, but I finally got GCC 4.7.0 built on Solaris 10 (and on CentOS 6.2 too). My script is much simplified now, I'm using contrib/download_prerequisites, and my configure looks like this: ../configure --prefix=/opt/gcc-4.7.0 --without-gnu-as --with-as=/usr/ccs/bin/as --without-gnu-ld --with-ld=/usr/ccs/bin/ld --enable-shared --enable-languages=c,c++ So it all seems to be working good, and everything it builds seems to run just fine :) BTW, the cause of the segmentation fault? A dodgy 'touch'! If I deleted /opt/sfw/bin/touch (installed from the companion CD's SFWcoreu package I believe) then gmake would then succeed - very strange. Regards, Graham On 30 March 2012 15:15, Dennis Clarke <dclarke@xxxxxxxxxxxxx> wrote: > >> Hi, >> >> I'm attempting to build GCC 4.6.3 on a virtual Solaris 10 machine >> (installed using sol-10-u10-ga2-x86-dvd.iso), but gmake fails pretty >> early on in the compilation. The machine's pretty clean, I've only >> installed a few packages from the companion CD, nothing else. > > OKay, good looking thus far. > >> >> Here's what my build script does: >> >> cd $HOME >> mkdir gccbuild/gcc463/output >> cd gccbuild/gcc463 >> export GCCBUILDOUTPUT=`pwd`/output > > * nod * > >> >> gtar -xjf $HOME/downloads/gcc463/gcc-4.6.3.tar.bz2 >> gtar -xzf $HOME/downloads/gcc463/gmp-4.2.1.tar.gz >> gtar -xjf $HOME/downloads/gcc463/mpfr-2.4.0.tar.bz2 >> gtar -xzf $HOME/downloads/gcc463/mpc-0.8.2.tar.gz > > I always build gmp/mpfr/mpc "in-tree" with no issue at all unless I am > on a ppc970 PowerMac or DecAlpha or some other odd beast. On sparc or > x86_64 Solaris 10 it will "just work". However you should make > some symlinks like so : > > ln -s ./gmp-4.2.1 ./gmp > ln -s ./mpfr-2.4.0 ./mpfr > ln -s ./mpc-0.8.2 ./mpc > > Hey, why are you using those versions ? > > I had just recently bootstrapped gcc-4.7.0 on Solaris 8 sparc and > i386 with no issues and I use : > > lrwxrwxrwx 1 root other 11 Mar 23 12:28 gmp -> ./gmp-5.0.4/ > drwxr-xr-x 14 1001 root 2048 Feb 10 10:23 gmp-5.0.4/ > lrwxrwxrwx 1 root other 9 Mar 23 12:29 mpc -> ./mpc-0.9/ > drwxr-xr-x 6 1000 1000 512 Feb 21 2011 mpc-0.9/ > lrwxrwxrwx 1 root other 12 Mar 23 12:29 mpfr -> ./mpfr-3.1.0/ > drwxr-xr-x 9 1000 1000 1024 Oct 3 09:43 mpfr-3.1.0/ > > > See the results here ( until someone updates the successful build page ) : > > http://gcc.gnu.org/ml/gcc-testresults/2012-03/msg02928.html > > >> cd gmp-4.2.1 >> ./configure --prefix=$GCCBUILDOUTPUT --without-gnu-ld --enable-shared >> --enable-static >> gmake >> gmake install >> cd .. > > okay .. whoa. Why build these before doing your bootstrap? > Not really required unless you do really want them installed > prior to the actual bootstrap. > > >> cd gcc-4.6.3 >> mkdir objdir >> cd objdir >> ../configure --prefix=$GCCBUILDOUTPUT --with-gmp=$GCCBUILDOUTPUT >> --with-mpfr=$GCCBUILDOUTPUT --with-mpc=$GCCBUILDOUTPUT --with-gnu-as >> --with-as=/usr/sfw/bin/gas --without-gnu-ld --with-ld=/usr/ccs/bin/ld >> --enable-shared --enable-languages=c,c++ LDFLAGS="-R >> $GCCBUILDOUTPUT/lib:/usr/sfw/lib" >> gmake ###FAILURE### >> gmake -k check >> gmake install > > OKay, do not use the binutils gas or gld. Just use the ones in the > regular /usr/ccs/bin directory. > > I can get into a long long discussion why but for now take my word > for it that you are going down a deep and old rabbit hole there. > > >> When the GCC gmake fails, here's the end of the output: >> >> config.status: executing default commands >> gmake[3]: Entering directory >> `/home/grahamb/gccbuild/gcc463/gcc-4.6.3/objdir/libiberty' >> if [ x"-fpic" != x ] && [ ! -d pic ]; then \ >> mkdir pic; \ >> else true; fi >> touch stamp-picdir >> gmake[3]: *** [stamp-picdir] Segmentation Fault (core dumped) >> gmake[3]: *** Deleting file `stamp-picdir' >> gmake[3]: Leaving directory >> `/home/grahamb/gccbuild/gcc463/gcc-4.6.3/objdir/libiberty' >> gmake[2]: *** [all-stage1-libiberty] Error 2 >> gmake[2]: Leaving directory `/home/grahamb/gccbuild/gcc463/gcc-4.6.3/objdir' >> gmake[1]: *** [stage1-bubble] Error 2 >> gmake[1]: Leaving directory `/home/grahamb/gccbuild/gcc463/gcc-4.6.3/objdir' >> gmake: *** [all] Error 2 >> > > Segmentation Fault (core dumped) ? > > nice. > > Do you have a output core file ? Do you have coreadm setup? > > Type coreadm as rroot and if you see this : > > > # coreadm > global core file pattern: > global core file content: default > init core file pattern: core > init core file content: default > global core dumps: disabled > per-process core dumps: enabled > global setid core dumps: disabled > per-process setid core dumps: disabled > global core dump logging: disabled > > > You don't have core tracking setup. > > Here is my coreadm config : > > # coreadm -e global -e global-setid -e log -e process -e proc-setid \ >> -g > /var/crash/jupiter/coredump/node_%n-host_%m-zone_%z-time_%t-pid_%p-uid_%u-gid_%g-fid_%f.core > \ >> -G all > # coreadm -e global -e global-setid -e log -e process -e proc-setid \ >> -i > /var/crash/jupiter/coredump/node_%n-host_%m-zone_%z-time_%t-pid_%p-uid_%u-gid_%g-fid_%f.per-process-core > \ >> -I all > # coreadm > global core file pattern: > /var/crash/jupiter/coredump/node_%n-host_%m-zone_%z-time_%t-pid_%p-uid_%u-gid_%g-fid_%f.core > global core file content: all > init core file pattern: > /var/crash/jupiter/coredump/node_%n-host_%m-zone_%z-time_%t-pid_%p-uid_%u-gid_%g-fid_%f.per-process-core > init core file content: all > global core dumps: enabled > per-process core dumps: enabled > global setid core dumps: enabled > per-process setid core dumps: enabled > global core dump logging: enabled > # > > So a while back ( before Oracle fixed it ) luxadm would toss a core on > a regular basis and I caught all that like so : > > # luxadm probe > Found Enclosure(s): > SENA Name:array2 Node WWN:5080020000027f00 > Logical Path:/dev/es/ses0 > Logical Path:/dev/es/ses1 > Segmentation Fault - core dumped > > Sep 27 00:52:27 jupiter genunix: NOTICE: core_log: luxadm[2575] core dumped: > /var/crash/jupiter/coredump/node_jupiter-host_sun4u-zone_global-time_1285548746-pid_2575-uid_0-gid_0-fid_luxadm.core > > Here is a shell script that will set it up for you : > > #!/bin/ksh > # Only for Solaris 10 > mkdir -m 0700 -p /var/crash/`/usr/bin/uname -n`/coredump > coreadm -e global -e global-setid -e log -e process -e proc-setid -g > /var/crash/`/usr/bin/uname > -n`/coredump/node_%n-host_%m-zone_%z-time_%t-pid_%p-uid_%u-gid_%g-fid_%f.core > -G all > coreadm -e global -e global-setid -e log -e process -e proc-setid -i > /var/crash/`/usr/bin/uname > -n`/coredump/node_%n-host_%m-zone_%z-time_%t-pid_%p-uid_%u-gid_%g-fid_%f.per-process-core > -I all > coreadm > > > Now you can boldly go forwards and catch every time a core happens .. which > should be NEVER. However if you do get one at least you can then try a dbx or > even basic debug on it to see wtf happened. > >> Has anybody seen this before, and knows how to fix it? > > Seen before : no > fix it : yes > >> I've actually tried GCC 4.3.6 the same way, with gmp and mpfr (but >> without mpc, it's not required), with the same result. >> >> I have very limited experience with Solaris, so please bear with me! > > I have loccked in mortal battle with it a few times .. so stay in > touch. also, on this list we may get lucky and hear from a man > named Rainer. He is a real guru on these matters. > > Dennis > > > -- > -- > http://pgp.mit.edu:11371/pks/lookup?op=vindex&search=0x1D936C72FA35B44B > +-------------------------+-----------------------------------+ > | Dennis Clarke | Solaris and Linux and Open Source | > | dclarke@xxxxxxxxxxxxx | Respect for open standards. | > +-------------------------+-----------------------------------+ >
