This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Avoid use of libgcc_s and libgcc_eh when building glibc
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Roland McGrath <roland at hack dot frob dot com>
- Cc: Thomas Schwinge <thomas at codesourcery dot com>, libc-alpha at sourceware dot org, ryan dot arnold at gmail dot com, Andreas Krebbel <krebbel at linux dot vnet dot ibm dot com>, Mike Frysinger <vapier at gentoo dot org>, "H.J. Lu" <hjl dot tools at gmail dot com>
- Date: Tue, 12 Jun 2012 19:30:20 +0000 (UTC)
- Subject: Re: Avoid use of libgcc_s and libgcc_eh when building glibc
- References: <Pine.LNX.4.64.1205282140020.2726@digraph.polyomino.org.uk><20120529235232.E0F742C0B8@topped-with-meat.com><Pine.LNX.4.64.1205301052380.25659@digraph.polyomino.org.uk><877gvt8hi0.fsf@schwinge.name> <87txyoksdr.fsf@schwinge.name><20120606221104.AEA162C096@topped-with-meat.com><Pine.LNX.4.64.1206081142100.3499@digraph.polyomino.org.uk><20120612172703.921CB2C07D@topped-with-meat.com>
On Tue, 12 Jun 2012, Roland McGrath wrote:
> But I think all those --disable-*/-with* options are really just intended
> to affect libgcc, right? So I wonder if there might not be a better
They are intended to affect anything in GCC they are relevant to; I don't
think there's a clear definition that they are only relevant to some parts
and not others.
> procedure in which one builds gcc proper with the normal/final set of
> configuration options and only specially configures a libgcc build to be
> used for the libc build.
I hope it will indeed eventually be possible to configure and build GCC
only once with only libgcc having separate builds for bootstrap purposes.
The present patch (now presumably for glibc 2.17) aims at a more modest
incremental improvement on the glibc side. I believe I have had reports
of successful builds for SPARC, SH, Tile and GNU Hurd. I don't appear to
have had any reports of build results for Power Architecture, S390 or IA64
(and I also need to do an i386 build; x32 testing would also be good).
There was some discussion of IA64-specific bootstrap issues, but while
interesting I don't think that's relevant to this patch; what I want for
IA64 is to make sure that the patch
<http://sourceware.org/ml/libc-alpha/2012-05/msg01920.html> (with or
without the possible sysdeps makefile additions I suggested) does not
break things, rather than that, given IA64-specific issues, it makes
things any better there.
--
Joseph S. Myers
joseph@codesourcery.com