This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

gcc.dg/weak/typeof-2.c test case assumptions


Hello,

this test case is failing on s390 due to what I believe to be
architecture-dependent assumptions made in the code.

Specifically, the test case expects that in the code generated for
  return foo3 (x) + foo3 (x) + foo3 (x) + foo3 (x) + foo3 (x) + foo3 (x);
(where foo3 is not const), the term 'foo3' (or, in this case, 'baz3'
as it is renamed via __asm) occurs 6 distinct times.

This is incorrect on s390, where the address of foo3 is loaded from
the constant pool; so you just have one occurrence of that term
even though 6 distinct function calls are properly generated.

Could this test case be rewritten to remove the architecture
specific assumptions?  Failing that, is there some mechanism
to mark it as expected to fail on s390?


Mit freundlichen Gruessen / Best Regards

Ulrich Weigand

--
  Dr. Ulrich Weigand
  Linux for S/390 Design & Development
  IBM Deutschland Entwicklung GmbH, Schoenaicher Str. 220, 71032 Boeblingen
  Phone: +49-7031/16-3727   ---   Email: Ulrich.Weigand@de.ibm.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]