https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124216
--- Comment #2 from Alexander Fisher <afisher at scitec dot com> --- (In reply to Andrew Pinski from comment #1) > I am not sure this is an errant error though. > > you assign k from things[i][j] and then access things[k][j]. > > Well GCC could figure out things.stuff is all zero but currently does not . > BUT I suspect the original code was much more complex than this. Yeah, the original code was a lot more complex, but it made sure to never go outside of the bounds of the array.
