https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105950
--- Comment #13 from John Kanapes <jkanapes at yahoo dot com> --- (In reply to Jakub Jelinek from comment #12) > (In reply to John Kanapes from comment #11) > > (In reply to Richard Biener from comment #9) > > > I am trying to recreate this bug in a smaller, more concise context. > > It is not an obvious bug. This is valid code, and it takes a large chain of > > previous steps to get it wrong at runtime. It used to work with previous > > gccs, but it now seems broken:( > > Claiming it is valid code until it is analyzed is premature. It can very > well be undefined behavior in the code. True. Except that I have already analyzed it with my own tools. That means that the offending code, as reported by gdb, compiles and runs fine with -O6 optimization with a simpler code. I am not claiming anything, just stressing that this is not an obvious issue as reported by gdb, and requires a lot of previous steps to reproduce:(