This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH][LTO][C++] Fix PR42762, fix GTY issue


On Fri, 5 Feb 2010, Diego Novillo wrote:

> On Fri, Feb 5, 2010 at 08:48, Richard Guenther <rguenther@suse.de> wrote:
> 
> > 2010-02-05 ÂRichard Guenther Â<rguenther@suse.de>
> >
> > Â Â Â ÂPR lto/42762
> > Â Â Â Â* tree-eh.c (execute_cleanup_eh_1): Copy from execute_cleanup_eh.
> > Â Â Â Â(execute_cleanup_eh): Clear DECL_FUNCTION_PERSONALITY if it is
> > Â Â Â Âno longer needed.
> > Â Â Â Â* lto-streamer-in.c (get_resolution): Deal with references
> > Â Â Â Âto undefined functions.
> >
> > Â Â Â Âcp/
> > Â Â Â Â* Make-lang.in (cp/cp-lang.o): Depend on gt-cp-cp-lang.h.
> > Â Â Â Â* cp-lang.c: Include gt-cp-cp-lang.h.
> > Â Â Â Â* config-lang.in (gtfiles): Add cp/cp-lang.c.
> 
> Now, if we're clearing DECL_FUNCTION_PERSONALITY, why do we need to
> deal with undefined functions in get_resolution?  Are we seeing this
> function in any other expression?

No, but at -O0 eh-optimize does not run before emitting LTO bytecode
but it does run before expansion, so gold does not see the
cxx_personality routine or references to it but lto1 does (and
misses a resolution from gold for it).

> Not a big deal, anyway.  I'm OK with the patch as-is.

Thanks,
Richard.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]