| Message ID | yddhabgief1.fsf@lokon.CeBiTec.Uni-Bielefeld.DE |
|---|---|
| State | New |
| Headers | show |
Il 13/11/2013 15:06, Rainer Orth ha scritto: > > This happens because there's no installed amd64 libgcc_s.so.1 on the > system, and toplevel Makefile only sets LD_LIBRARY_PATH for the default > multilib. Initially, I thought that there were something special going > on, but it turned out that other runtime libs containing C++ code don't > have this problem, which can easily be cured by the following patch. It > allowed a C++-only i386-pc-solaris2.10 bootstrap to finish, make check > currently running. I'm currently also running an > x86_64-unknown-linux-gnu bootstrap to make sure nothing breaks there. > > Ok for mainline if those pass? Yes. > Btw., I noticed a couple of other anomalies: > > * configure.ac has > > GCC_LIBSTDCXX_RAW_CXX_FLAGS > > but does nothing with the result: Makefile.in substitutes the results, > but that's it. Also, toplevel Makefile.tpl should set raw_cxx=true it > this were useful, which it doesn't do as well. Yes, looks like cut-and-paste. Paolo > * MAINTAINERS doesn't list a maintainer for libcilkrts. > > * I believe there should be a libcilkrts bugzilla category, intead of > having to use other.
# HG changeset patch # Parent 8810468d355365c1e486a34054abb823b17aad4b Enable libcilkrts multilib build on Solaris diff --git a/libcilkrts/configure.ac b/libcilkrts/configure.ac --- a/libcilkrts/configure.ac +++ b/libcilkrts/configure.ac @@ -46,8 +46,8 @@ AM_MAINTAINER_MODE # Build a DLL on Windows # AC_LIBTOOL_WIN32_DLL +AC_PROG_CC AC_PROG_CXX -AC_PROG_CC # AC_PROG_LIBTOOL # AC_CONFIG_MACRO_DIR([..]) AC_CONFIG_FILES([Makefile])
Balaji, currently configuring the non-default libcilkrts multilib fails on Solaris/x86 like this: configure:3139: checking whether the C++ compiler works configure:3148: ./a.out ld.so.1: a.out: fatal: /var/gcc/regression/trunk/10-gcc/build/./gcc/libgcc_s.so.1: wrong ELF class: ELFCLASS32 /vol/gcc/src/hg/trunk/local/libcilkrts/configure[3150]: 26758 Killed configure:3152: $? = 137 configure:3159: error: in `/var/gcc/regression/trunk/10-gcc/build/i386-pc-solaris2.10/amd64/libcilkrts': configure:3161: error: cannot run C++ compiled programs. If you meant to cross compile, use `--host'. See `config.log' for more details. This happens because there's no installed amd64 libgcc_s.so.1 on the system, and toplevel Makefile only sets LD_LIBRARY_PATH for the default multilib. Initially, I thought that there were something special going on, but it turned out that other runtime libs containing C++ code don't have this problem, which can easily be cured by the following patch. It allowed a C++-only i386-pc-solaris2.10 bootstrap to finish, make check currently running. I'm currently also running an x86_64-unknown-linux-gnu bootstrap to make sure nothing breaks there. Ok for mainline if those pass? Btw., I noticed a couple of other anomalies: * configure.ac has GCC_LIBSTDCXX_RAW_CXX_FLAGS but does nothing with the result: Makefile.in substitutes the results, but that's it. Also, toplevel Makefile.tpl should set raw_cxx=true it this were useful, which it doesn't do as well. * MAINTAINERS doesn't list a maintainer for libcilkrts. * I believe there should be a libcilkrts bugzilla category, intead of having to use other. Thanks. Rainer 2013-11-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE> * configure.ac (AC_PROG_CXX): Move after AC_PROG_CC. * configure: Regenerate.