On Fri, 27 Mar 2020, Jakub Jelinek wrote: > On Fri, Mar 27, 2020 at 09:32:40AM +0100, Richard Biener wrote: > > On Fri, 27 Mar 2020, Jakub Jelinek wrote: > > > > > On Thu, Mar 26, 2020 at 08:41:41PM -0300, Alexandre Oliva wrote: > > > > On Mar 26, 2020, Jakub Jelinek <ja...@redhat.com> wrote: > > > > > > > > > Or disable -gstatement-frontiers by default and declare it > > > > > -fcompare-debug > > > > > incompatible. > > > > > > > > I don't get what makes debug stmts introduced by -gstatement-frontiers > > > > special in this regard. I recall working a lot on making statement list > > > > management compatible with debug stmts for -fcompare-debug purposes back > > > > when introducing them, but not breaking new ground when extending them > > > > with markers rather than binds. > > > > > > The hardest issue I gave up completely on after trying 3 different > > > approaches is PR93786, I just don't know what to do in that case. > > > Gimplification of a STATEMENT_LIST is destructive, especially the > > > voidify_wrapper_expr part of it, and if the STATEMENT_LIST is used > > > multiple > > > times, that is a severe problem. > > > > The question for this case is wheter we can arrange for the STATEMENT_LIST > > to be present even for -g0 and "solve" the issue that way? > > That would just mean it ICEs at -g0 too, this PR isn't a -fcompare-debug > issue but a bug where we try to gimplify the same STATEMENT_LIST multiple > times and second time ICE on it as it has been voidified and emptied.
But only because we made STATEMENT_LISTs not having side-effects which they all did before all the changes. We could revert that... Richard.
