This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc 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]

Re: Use decimal constants in defining M_* in libm-test.inc


On 05/08/2013 09:38 PM, Joseph S. Myers wrote:
For IBM long double, GCC does not constant-fold expressions that are
not exact in 106 bits, resulting in errors for valid static
initializers<http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26374>.

Thus, to convert libm tests to being based on static data, such
expressions need to be avoided in test arguments and expected results.
Various values used in some tests are defined in libm-test.inc as
expressions that may not be exact.  This patch changes those M_*
macros in libm-test.inc to use decimal values (with 36 places after
the point, as in math.h) to avoid such problems.

Tested x86_64 and x86.  No ulps updates needed, but it's possible some
platforms might need ulps updates if this patch results in a more
accurate value of an expected result being used.

Ok, thanks,

Andreas
--
 Andreas Jaeger aj@{suse.com,opensuse.org} Twitter/Identica: jaegerandi
  SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
   GF: Jeff Hawn,Jennifer Guild,Felix Imendörffer,HRB16746 (AG Nürnberg)
    GPG fingerprint = 93A3 365E CE47 B889 DF7F  FED1 389A 563C C272 A126


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