math: support platforms with limited FP rounding or exception support
commit0af797def371ceb4f05586d7bcd25841653d2082
authorChris Metcalf <cmetcalf@tilera.com>
Thu, 17 May 2012 12:49:19 +0000 (17 08:49 -0400)
committerChris Metcalf <cmetcalf@tilera.com>
Thu, 17 May 2012 12:50:41 +0000 (17 08:50 -0400)
tree3aa5591836c95fba6b694fb45c1421bf0025de91
parente39745ffa030f685fbba13534a3023d52a27ead8
math: support platforms with limited FP rounding or exception support

For some tests, just claim that fetestexcept() always returns true,
so the rest of the test can be compiled.

For libm-test, provide known bogus values for unsupported rounding
modes, so fesetround() will return failure.

Elsewhere, just add some #ifdefs to avoid code that uses particular
FP exceptions if the exceptions aren't supported.
ChangeLog
math/bug-nextafter.c
math/bug-nexttoward.c
math/libm-test.inc
math/test-fenv.c
math/test-misc.c
stdlib/bug-getcontext.c