[Bug tree-optimization/92152] [10 Regression] Wrong code (Resurrection of PR53663)

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92152

--- Comment #14 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Created attachment 47901
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=47901&action=edit
another related testcase

> But that means we now miscompile the later two tests.

Good point, those are my testcases :) I have missed it is scan-not. I will make
updated patch.

I am attaching related testcase that reads same data using two different union
fields relying on C++ initial sequence rule (as Mark Williams explained me).

"In a standard-layout union with an active member of non-union class type T1,
it is permitted to read a non-static data member m of another union member of
non-union class type T2 provided m is part of the common initial sequence of T1
and T2 (except that reading a volatile member through non-volatile glvalue is
undefined)."

This seems to indicate that we really can not disambiguate after union
reference by access path unless we implement the "common initial sequence"
check