This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Make alias_sets_conflict_p less conservative
- From: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- To: matz at suse dot de
- Cc: dberlin at dberlin dot org, gcc at gcc dot gnu dot org, joseph at codesourcery dot com, rguenther at suse dot de
- Date: Wed, 05 Mar 2008 16:19:49 EST
- Subject: Re: [PATCH] Make alias_sets_conflict_p less conservative
- References: <Pine.LNX.firstname.lastname@example.org> <Pine.LNX.email@example.com> <10803051256.AA05161@vlsi1.ultra.nyu.edu> <Pine.LNX.firstname.lastname@example.org> <10803051426.AA05981@vlsi1.ultra.nyu.edu> <Pine.LNX.email@example.com> <10803051515.AA06678@vlsi1.ultra.nyu.edu> <Pine.LNX.firstname.lastname@example.org> <Pine.LNX.email@example.com> <10803051630.AA08112@vlsi1.ultra.nyu.edu> <firstname.lastname@example.org> <10803051924.AA09430@vlsi1.ultra.nyu.edu> <Pine.LNX.email@example.com> <10803052042.AA11211@vlsi1.ultra.nyu.edu> <Pine.LNX.firstname.lastname@example.org>
> alias_sets_conflict_p() is used to determine if an addressable variable
> and a type conflict also from the tree-ssa alias machinery (in effect to
> determine which virtual variables need to be clobbered). It just isn't
> feed COMPONENT_REFs, that's what you found very wrong, and I explained the
> reasons for not doing so. If we were to do that (and hence become more
> precise) we would have to increase memory use as just explained. You were
> asking how feeding COMPONENT_REFs would be connected with memory use, and
> I explained it.
Why can't you keep the vops as they are now, but when you call
alias_sets_conflict_p, pass the actual reference from the op, not the vop?