Unknown testsuite failure

[NightStrike]

While running the fortran testsuite, I noticed a number of occurrences
of things like this in the log:

Executing on host:
/var/tmp/build/gcc-svn/build-x86_64-pc-linux/gcc/testsuite/gfortran/../../gfortran
-B/var/tmp/build/gcc-svn/build-x86_64-pc-linux/gcc/testsuite/gfortran/../../
/var/tmp/build/gcc-svn/gcc/gcc/testsuite/gfortran.dg/allocate_zerosize_3.f
  -O2   -pedantic-errors   -lm   -o ./allocate_zerosize_3.exe
(timeout = 300)
PASS: gfortran.dg/allocate_zerosize_3.f  -O2  (test for excess errors)
Executing on mingw: chmod +x /tmp/allocate_zerosize_3.exe.4981
(timeout = 300)
XYZ0ZYX
Executing on mingw: /tmp/allocate_zerosize_3.exe.4981    (timeout = 300)
Executing on mingw: rm -f  /tmp/allocate_zerosize_3.exe.4981.o
/tmp/allocate_zerosize_3.exe.4981    (timeout = 300)
XYZ0ZYX
FAIL: gfortran.dg/allocate_zerosize_3.f  -O2  execution test



The execution of the test fails, but there is no indication as to why
it failed in the log. There is no XYZ return code.  It just goes on to
removing the file.

How do I debug things like this?