[Bug fortran/79929] [7 Regression] Bogus Warning: '__builtin_memset': specified size 4294967291 exceeds maximum object size 2147483647

[law at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79929

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P4
          Component|middle-end                  |fortran

--- Comment #16 from Jeffrey A. Law <law at redhat dot com> ---
So WRT c#4, that is not an equivalent testcase in C AFAICT.

The reduced fortran testcase fails at -O2, the C testcase does not.  The fact
that the C testcase fails at -O1 is simply uninteresting.

If we go back to the Fortran testcases I don't see anything that would allow us
to determine statically that the problematical block can not be reached.

Working with the reduced Fortran testcase we have this in the dumps:

  sizetype _1;
  integer(kind=4) _2;
  unsigned long _3;


  _1 = (sizetype) _yerrmsg_9(D);
[ ... ]
  _2 = _yerrmsg_9(D) + 5;
  _3 = (unsigned long) _2;
[ ... ]
  if (_1 > _3)
    goto <bb 6>; [36.64%]
  else
    goto <bb 7>; [63.36%]

We're dealing with unsigned types here, so overflow has well defined semantics.
 Thus we can not statically determine the result of that conditional.

This really is a Fortran problem AFAICT.