https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97403
--- Comment #4 from hubicka at kam dot mff.cuni.cz --- > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97403 > > --- Comment #3 from Martin Jambor <jamborm at gcc dot gnu.org> --- > (In reply to Jan Hubicka from comment #2) > > Martin, > > I think we can close this (possibly adding the testcase) > > Depends. ANCESTOR got generalized a bit but the propagation at IPA-CP level > still does not take place. That is not a problem with jump functions though, > for that we need: > > 1) When propagating, look into "aggregate" constants when encountering a pass > through function and and there is something useful there (there is in this > case). But we also need to somehow represent a reference to a simple constant > (which is not a CONST_DECL) in the lattices. > > 2) In the tree-replacement/transformation phase, find a mechanism to replace > applicable dereferences of the parameter with the constant. > > Ultimately, a future return function should also get you a constant > (post-ipa-cp). > > We can file another bug with the same testcase though. I guess we can keep this PR as a collection of stuff that could be improved on aggregates... Honza
