Re: [PATCH], Fix PR 71869, Correctly implement isgreater, etc. on PowerPC __float128

[Michael Meissner <meissner at linux dot vnet dot ibm dot com>]

On Tue, Jul 26, 2016 at 11:05:32AM -0500, Segher Boessenkool wrote:
> > --- gcc/testsuite/gcc.target/powerpc/float128-cmp.c	(revision 0)
> > +++ gcc/testsuite/gcc.target/powerpc/float128-cmp.c	(revision 0)
> > @@ -0,0 +1,17 @@
> > +/* { dg-do compile { target { powerpc*-*-linux* } } } */
> > +/* { dg-require-effective-target powerpc_float128_sw_ok } */
> > +/* { dg-skip-if "do not override -mcpu" { powerpc*-*-* } { "-mcpu=*" } { "-mcpu=power7" } } */
> > +/* { dg-options "-O2 -mcpu=power7 -mfloat128" } */
> > +
> > +#ifndef __FLOAT128__
> > +#error "-mfloat128 is not supported."
> > +#endif
> 
> This should never trigger, it is only there for debug, right?
> 
> > +int
> > +test_isgreater (__float128 a, __float128 b)
> > +{
> > +  return __builtin_isgreater (a, b);
> > +}
> > +
> > +/* { dg-final { scan-assembler "bl __\[gl\]ekf2"    } } */
> > +/* { dg-final { scan-assembler "bl __unordkf2" } } */
> 
> There are some extra spaces in that [gl]ekf2 line.
> 
> Could you test all five functions please?  Use multiple testcases, maybe.

I decided to write an executable test rather than do more assembly tests.  The
patch to rs6000.c is unchanged, and the test now tests all of the comparison
operators and functions with various values including quiet NaNs and signalling
NaNs.  It passes on power7 big endian 64-bit, power7 big endian 32-bit, power8
little endian 64-bit, and I ran it on the power9 simulator with hardware
float128 support.

[gcc]
2016-07-26  Michael Meissner  <meissner@linux.vnet.ibm.com>

	PR target/71869
	* config/rs6000/rs6000.c (rs6000_generate_compare): Rework
	__float128 support when we don't have hardware support, so that
	the IEEE built-in functions like isgreater, first call __unordkf3
	to make sure neither operand is a NaN, and if both operands are
	ordered, do the normal comparison.

[gcc/testsuite]
2016-07-26  Michael Meissner  <meissner@linux.vnet.ibm.com>

	PR target/71869
	* gcc.target/powerpc/float128-cmp.c: New test to make sure that
	IEEE built-in functions handle quiet and signalling NaNs
	correctly.

-- 
Michael Meissner, IBM
IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA
email: meissner@linux.vnet.ibm.com, phone: +1 (978) 899-4797

Attachment: pr71869.patch02b
Description: Text document