[Bug fortran/80696] New: [8 Regression] New Failures to do with matmul

[tkoenig at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80696

            Bug ID: 80696
           Summary: [8 Regression] New Failures to do with matmul
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: tkoenig at gcc dot gnu.org
  Target Milestone: ---

There are some new failures with matmul with -m32, reported e.g. at

https://gcc.gnu.org/ml/fortran/2017-05/msg00027.html (testsuite
failures) and https://gcc.gnu.org/ml/fortran/2017-05/msg00026.html
(SPEC failure).

These failures are _really_ strange.  One of the failing test
cases is generic_20.f90.

$ gfortran -O0 generic_20.f90
$ ./a.out
$ gfortran -m32 -O0 generic_20.f90
$ ./a.out
-559038737
0 2.000000

There isn't even a write statement in the patch, so this
points towards some kind of memory corruption, so it
is probably not the front end patch.

Investigating.