[SM-Discuss] RFC: combining all gcc frontends into a single spell

Arwed von Merkatz v.merkatz at gmx.net
Wed Jul 22 14:07:17 EDT 2009


On Tue, Jul 21, 2009 at 05:12:39AM +0200, Ladislav Hagara wrote:
> 
> > For gmp and mpfr gcc docs specify one can include the sources into the
> > tree and gcc will roll it's own. As mpfr seems to be only used by it I
> > don't see a problem with that and as gmp is needed by gcc as well might
> > as well include it. Of course any dependency on gmp would then change to
> > config_query_option.
> >   
> 
> If mpfr and gmp are included in gcc, every update (or only security
> update?) of mpfr or gmp should be followed by re-casting of all gcc.
> For example mpfr's HISTORY:
> 
> 2009-02-26 Ladislav Hagara <hgr at vabo.cz>
>         * DETAILS: 2.4.1, SECURITY_PATCH=1, buffer overflow
> 
> This would certainly causes re-casting of gcc  (3 hours....).
> Mpfr alone last only 1 or 2 minutes.
> 
> # ldd /usr/libexec/gcc/i686-pc-linux-gnu/4.3.3/cc1
>         linux-gate.so.1 =>  (0xffffe000)
>         libmpfr.so.1 => /usr/lib/libmpfr.so.1 (0xb7fde000)
>         libgmp.so.3 => /lib/libgmp.so.3 (0xb7f8a000)
>         libc.so.6 => /lib/libc.so.6 (0xb7e49000)
>         /lib/ld-linux.so.2 (0xb805a000)
> 

Good point. Can we resolve the circular dependency with only the
combined gcc and mpfr/gmp as separate spells? Otherwise security updates
for those will get quite painful.

-- 
Arwed v. Merkatz                              Source Mage GNU/Linux developer
                                                    http://www.sourcemage.org


More information about the SM-Discuss mailing list