[LTO] Fix C++ thunks
Tue Dec 30 00:36:00 GMT 2008
Richard Guenther wrote:
> Still better than not supporting the varargs case at all ;) OTOH - why don't we
> use thunks for the "real" clones? I suppose because of possible optimizations?
And because I couldn't figure out how to do it at the time. It's not
just a simple pointer adjustment; you need to set a variable. It would
be more of a true alternate entry-point scenario. And, yes, it's going
to make the functions slower; they'll have to check a variable that
right now is optimized away. But, this would be a good -Os optimization.
Part of the history was that we were originally trying to support both
the old G++ ABI and the new industry-standard ABI, and I was trying to
avoid tearing too much apart all at once.
(650) 331-3385 x713
More information about the Gcc-patches