https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113688
--- Comment #32 from uecker at gcc dot gnu.org --- (In reply to rguent...@suse.de from comment #31) > On Mon, 14 Apr 2025, ebotcazou at gcc dot gnu.org wrote: > > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113688 > > > > --- Comment #30 from Eric Botcazou <ebotcazou at gcc dot gnu.org> --- > > > Note that this is not directly related to C23 changes. C23 changes reuse > > > the > > > across TU rules also inside TU. But here the problem is that the across > > > TU > > > aliasing was already broken in some corner cases even before C23 which > > > causes subtle miscompilation errors. > > Do we have testcase for those? One is gcc.dg/pr114713.c which I think is annoying because it shows that aliasing between structures with old-style [0] and new FAMS [] is broken. > > > OK, but do we absolutely need to fix them in 14.3 while they (presumably) > > are > > present in the first two 14 releases? > > And even before GCC 14. I don't think so, reversal is IMO the best > option. The miscompilations are not regressions, so as pointed out, I think reversal is ok.
