Re: [PATCH]: PR29335 evaluate transcendentals at compile-time using MPFR [take 3]

["Kaveh R. GHAZI" <ghazi at caip dot rutgers dot edu>]

On Fri, 20 Oct 2006, Kaveh R. GHAZI wrote:

> Also, I'll figure out a testcase for vanilla 2.2.0 as a followup.


And here is the testcase for buggy MPFR.  I believe the results are only
valid when sizeof(double)==8, so I verified that in the test.

Tested on sparc-sun-solaris2.10 with the sin/cos/tan patch:
http://gcc.gnu.org/ml/gcc-patches/2006-10/msg01053.html
using vanilla MPFR 2.2.0 (where it fails as expected) and patched MPFR
2.2.0 (where it passes).

Is the testcase okay for mainline assuming that the sin/cos/tan patch is
approved?

		Thanks,
		--Kaveh



2006-10-23  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>

        * gcc.dg/torture/builtin-sin-mpfr-1.c: New test.

/* Version 2.2.0 of MPFR had bugs in sin rounding.  This test checks
   to see if that buggy version was installed.  The problem is fixed
   in the MPFR cumulative patch http://www.mpfr.org/mpfr-current and
   presumably later MPFR versions.

   Origin: Kaveh R. Ghazi 10/23/2006.  */

/* { dg-do link } */

extern void link_error (int, double, double);

#define TESTIT(ARG,RES) do { \
  if (sizeof (double) == 8 && __builtin_sin(ARG) != RES) \
    link_error(__LINE__, __builtin_sin(ARG), RES); \
  } while (0);

int main()
{
  TESTIT (-0x1.c0016155c4da3p-1, -0x1.88fc58bcf030dp-1);

  TESTIT (0x1.30654d85c2756p-2, 0x1.2beeb9de27a79p-2);

  TESTIT (0x1.fe68ccaa8e201p+2, 0x1.fc3f0c54e97a7p-1);

  /* This case should always pass.  */
  TESTIT (0.0, 0.0);

  return 0;
}