This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: no_new_pseudos
Richard Sandiford <richard@codesourcery.com> writes:
> For the record, Alex puts my point of view perfectly here too.
> I gather from the follow-ups that there's resistance to
> s/no_new_pseudos/!BEFORE_RELOAD_P ()/ -- with BEFORE_RELOAD_P
> defined as "reload_in_progress || reload_completed" until Alex's
> follow-on change is implemented -- but I don't understand why.
I would probably approve that patch if somebody writes it.
Kenny's patch has not been submitted, and should not be submitted as
is since there are some errors in alpha.md:
> - "! no_new_pseudos || reload_completed"
> + "! (reload_completed || reload_in_progress)"
That transformation is incorrect--the current code is true when
reload_complete is true, and the new code is not. The correct patch
is probably to make this the empty string--i.e., always true (since
this is a split condition which will never be invoked during reload).
But after all this discussion I'm hoping that somebody else will write
a different patch, so for now I'm not going to continue reading
Kenny's patch.
Ian