Fix buidbot failure by disabling test

[tarunprabhu]

Summary

The bot fails on AArch64 because REAL(10) is not supported. See here:

https://lab.llvm.org/buildbot/#/builders/197/builds/7810/steps/13/logs/FAIL__test-suite__gfortran-regression-compile-regr

[luporl]

Thanks! All tests now pass on my AArch64 setup.

I think the following line in the test is responsible for detecting whether the target supports REAL(10) or not.
{ dg-require-effective-target fortran_real_10 }

[tarunprabhu]

In D154878#4486484, @luporl wrote:

I think the following line in the test is responsible for detecting whether the target supports REAL(10) or not.
{ dg-require-effective-target fortran_real_10 }

Thanks for looking into it. There are several DejaGNU directives that are currently not handled by the build system. It looks like this is another of them. I'll keep it in mind when tweaking the build system to handle some of the other currently failing tests.

[luporl]

Can we commit this fix? There are 5 AArch64 bots broken for more than a day now.

[tarunprabhu]

In D154878#4491204, @luporl wrote:

Can we commit this fix? There are 5 AArch64 bots broken for more than a day now.

Done. Sorry, I thought I had pushed this, but I hadn't.

Fortran/gfortran/regression/DisabledFiles.cmake

   # These tests are skipped because they cause flang to hang - possibly in an
   # infinite loop.
 
   goto_5.f90
   pr49540-1.f90 # Only on AArch64
 
   # --------------------------------------------------------------------------
   #
+  # These tests are skipped because REAL(10) is not supported on AAch64.
+
+  pr91497.f90
+
+  # --------------------------------------------------------------------------
+  #
   # These are skipped almost certainly because of a bug in the way multi-file
   # compile tests are built by the test-suite. This almost certainly has nothing
   # to do with flang, but they will be skipped until the test suite build
   # scripts are fixed.
 
   altreturn_9_0.f90
   bind_c_array_params_3.f90
   bind_c_usage_17.f90