[PATCH][LTO] register char_type_node

Richard Guenther rguenther@suse.de
Fri Aug 14 14:42:00 GMT 2009


On Fri, 14 Aug 2009, Diego Novillo wrote:

> On Fri, Aug 14, 2009 at 10:25, Rafael Espindola<espindola@google.com> wrote:
> 
> > Shouldn't we change the ME to check for signed and unsigned char?
> 
> The meaning of char_type_node in the ME will change depending on what
> the FE originally wrote to the bytecode stream, so making
> char_type_node available for merging seems like the safest approach
> for now.

Yeah.  I was wondering whether we shouldn't register all types
we preload the streamer cache with early - that would also make
us canonicalize to the integer type variants with the pretty names,
not this <unnamed-signed:8> crap I see everywhere around ...

That would cover also the char_type_node case.  If you think that
is a good idea I'll rework the patch accordingly.

> As Richard says, the code that relies on char_type_node is pretty
> brittle.  More cleanup opportunities, I guess.

Indeed.

Richard.



More information about the Gcc-patches mailing list