This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH for alias set/stack temporary bug
- To: mark at markmitchell dot com
- Subject: Re: PATCH for alias set/stack temporary bug
- From: Jeffrey A Law <law at hurl dot cygnus dot com>
- Date: Thu, 11 Feb 1999 01:59:55 -0700
- cc: rth at cygnus dot com, egcs-patches at cygnus dot com
- Reply-To: law at cygnus dot com
In message <199902110854.AAA27249@adsl-206-170-148-33.dsl.pacbell.net>you wri
te:
> >>>>> "Richard" == Richard Henderson <rth@cygnus.com> writes:
>
> Richard> On Wed, Feb 10, 1999 at 11:01:59PM -0700, Jeffrey A Law
> Richard> wrote:
> >> Hmmm, what other ways can we try to deal with this?
>
> Richard> Can we represent set unions?
>
> Yes; we've already got that functionality.
>
> Richard> We could create a set for each stack slot; stack slots
> Richard> that get broken up get subsetted so that they don't
> Richard> conflict with one another but do with the `parent' slot.
> Richard> The memory returned for any particular stack memory would
> Richard> have a mem set that is the union of it's type and
> Richard> containing stack slot.
>
> That was my first idea too. But, it's not very easy to do. One of
> the cases where we get stack temps is for the call frames for inlined
> functions. It's hard to alter the MEMs used by the callee to
> incorporate this information.
Right. That's really the tough part -- finding all those mems which refer
to those slots and fixing them.
jeff