This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: $(build_tooldir)/lib (was Re: http://gcc.gnu.org/ml/gcc-patches/2000-05/msg01104.html)
- To: Jason Merrill <jason at redhat dot com>
- Subject: Re: $(build_tooldir)/lib (was Re: http://gcc.gnu.org/ml/gcc-patches/2000-05/msg01104.html)
- From: Jim Wilson <wilson at cygnus dot com>
- Date: Fri, 21 Jul 2000 15:28:40 -0700
- cc: Geoff Keating <geoffk at cygnus dot com>, hjl at lucon dot org, aoliva at cygnus dot com, gcc-patches at gcc dot gnu dot org, binutils at sourceware dot cygnus dot com
When you build an ia64 toolchain, where does your crt0.o live? How does
the compiler find it?
The libraries are copied to $prefix/$target/lib (aka $tooldir/lib) at
configure time. That is one of the paths that gcc searches by default.
That happens in gcc.c when it adds tooldir_prefix/lib to startfile_prefixes.
Looking at H.J. messages, he says this doesn't work because tooldir is
really $prefix/lib/gcc-lib/$target/$version/../../../../$target/lib
and $prefix/lib/gcc-lib/$target/$version doesn't exist at build time.
But it does. We deliberately create it so that $tooldir/lib and
$tooldir/sys-include will work. See the code in the stmp-fixinc rule in the
Makefile. Nothing needs $libsubdir before then. If something does need it,
then we should create it earlier.
I believe one of H.J.'s other comments was that he defines prefix to point
somewhere in /usr, and since he isn't root when he is building the toolchain,
the creation of $libsubdir fails. I think that is a little weak. But if
this is a concern, then the new scheme doesn't help fix it, because it is
still installing stuff in $prefix at configure/build time.
Jim