https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115528
--- Comment #18 from Richard Biener <rguenth at gcc dot gnu.org> --- (In reply to Jürgen Reuter from comment #17) > (In reply to Richard Biener from comment #16) > > (In reply to Jürgen Reuter from comment #15) > > > I fund the culprit commit in the gcc master, it is: > > > https://gcc.gnu.org/git/?p=gcc.git;a=commit; > > > h=1fe55a1794863b5ad9eeca5062782834716016b2 > > > by Richard Biener on the tree-optimization. Now I will try helping to get > > > a > > > smaller reproducer. > > > > Interesting, note the other commit had r15-1309-ge575b5c56137b1 as followup > > fix already. > > > > The loop at contrib/tauola/formf.f:599 should be the problem, adding > > > > !GCC$ novector > > > > to it should avoid the issue (and also serve as confirmation to that). > > > > I wonder if you can then attach enough to be able to compile the > > translation unit with the respective loop. > > Indeed, adding, as you suggested, !GCC$ novector before the double do-loop > cures the segmentation fault. I saw the follow-up commit but apparently the > daily bumps after that (e.g. from Monday morning, June 17, 0:18 am, or so, > still had this issue. I try to get a shorter reproducer, but not sure when I > will be able to do so. If the TU with the loop can be compiled in isolation can you attach that together with the GCC command-line triggering the issue?
