[Bug fortran/84615] [8 Regression] Executable Segfault for some tests compiled with -m32 after r256284

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84615

--- Comment #18 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> Preliminary tests show that all the failures reported in comment 0 are
> fixed by the patch except:
>
> % gfc -m32 -fdefault-integer-8
>  /opt/gcc/p_work/gcc/testsuite/gfortran.dg/assumed_rank_1.f90
> /opt/gcc/p_work/gcc/testsuite/gfortran.dg/assumed_rank_1_c.c

This failure is related to pr32957. What is surprising is that the test does
not fail with -m64.

Full test

                === gfortran Summary for unix/-m32 ===

# of expected passes            46873
# of expected failures          92
# of unsupported tests          178

                === gfortran Summary for unix/-m32/-fdefault-integer-8 ===

# of expected passes            46558
# of unexpected failures        296
# of expected failures          92
# of unresolved testcases       38
# of unsupported tests          178

Running target unix/-m64

                === gfortran Summary for unix/-m64 ===

# of expected passes            47090
# of expected failures          92
# of unsupported tests          75

                === gfortran Summary for unix/-m64/-fdefault-integer-8 ===

# of expected passes            46802
# of unexpected failures        280
# of expected failures          92
# of unresolved testcases       38
# of unsupported tests          75

                === gfortran Summary ===

# of expected passes            187323
# of unexpected failures        576
# of expected failures          368
# of unresolved testcases       76
# of unsupported tests          506
/opt/gcc/p_build/gcc/gfortran  version 8.0.1 20180320 (experimental) [trunk
revision 258691p2] (GCC) 

So the patch fixes this PR without regression.