https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126906
--- Comment #7 from Aldy Hernandez <aldyh at gcc dot gnu.org> --- (In reply to Richard Biener from comment #5) > (In reply to Aldy Hernandez from comment #4) > > I’ve proposen a patch here: > > > > https://gcc.gnu.org/pipermail/gcc-patches/2026-August/728906.html > > > > With it we handle the Andrea’s test in comment #1, and if we disable the > > probably_never_executed_edge_p restriction, we also get the original test. > > > > So, however way we fix the backward’s threader or generic copier, all that’s > > left is tweaking the built-in restriction. > > > > Richard, what’s your take on the never executed restriction, mark it as a > > regression and track it, or adjust it somehow? > > The never-executed-edge-p restriction was put in to fix a regression IIRC, > so if you remove that you need to fix that differently ... > > r13-1891-g49ba4fdeb648c1 and r13-1924-gb9da686470d1c2 added them. BTW, neither of these commits have tests associated with them. How about we retroactively distill tests for them, commit them independently, so whatever fix we come up with does not regress those?
