https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126250

--- Comment #5 from Roger Sayle <roger at nextmovesoftware dot com> ---
Investigating... Clearly the data-flow here is completely broken/useless, as
there shouldn't be a chain containing a use and no def, which appears to be
caused/triggered by an uninitialized variable.

The one line change:
 _BitInt(95) deadx22 = 0;

resolves the problem.  The question is how can STV determine that the data-flow
information is bogus.  I believe STV assumes/requires all mentions of a pseudo
to appear in a single chain, which normally holds when the program is
well-formed.

Reply via email to