Allow lib/msun/logarithm_test to pass on ld128 platforms

For some reason the ld128 log1pl() implementation is less accurate than
logl(), but does at least guarantee precision >= the ld80 implementation.
Mark log1p_accuracy_tests as XFAIL for ld128 and increase the log1p tolerance
to the ld80 equivalent in accuracy_tests to avoid losing test coverage for
the other functions.

PR:		253984
Reviewed By:	ngie, dim
Differential Revision: https://reviews.freebsd.org/D29039
This commit is contained in:
Alex Richardson 2021-04-20 00:22:45 +01:00
parent ab147542b7
commit 1ad83445fc

View file

@ -212,6 +212,17 @@ ATF_TC_BODY(accuracy_tests, tc)
};
unsigned i;
long double log1p_ldbl_ulp = LDBL_ULP();
#if LDBL_MANT_DIG > 64
/*
* On ld128 platforms the log1p() implementation provides less accuracy,
* but does still match the ld80 precision. Use the ld80 LDBL_ULP()
* value for now to avoid losing test coverage for the other functions.
* Reported as https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=253984
*/
log1p_ldbl_ulp = ldexpl(1.0, 1 - 64);
#endif
for (i = 0; i < nitems(tests); i++) {
test_tol(log2, tests[i].x, tests[i].log2x, DBL_ULP());
test_tol(log2f, tests[i].x, tests[i].log2x, FLT_ULP());
@ -228,7 +239,7 @@ ATF_TC_BODY(accuracy_tests, tc)
test_tol(log1pf, tests[i].x - 1, tests[i].logex,
FLT_ULP());
test_tol(log1pl, tests[i].x - 1, tests[i].logex,
LDBL_ULP());
log1p_ldbl_ulp);
}
}
}
@ -236,6 +247,11 @@ ATF_TC_BODY(accuracy_tests, tc)
ATF_TC_WITHOUT_HEAD(log1p_accuracy_tests);
ATF_TC_BODY(log1p_accuracy_tests, tc)
{
#if LDBL_MANT_DIG > 64
if (atf_tc_get_config_var_as_bool_wd(tc, "ci", false))
atf_tc_expect_fail("https://bugs.freebsd.org/253984");
#endif
test_tol(log1pf, 0x0.333333p0F,
1.82321546859847114303367992804596800640e-1L, FLT_ULP());
test_tol(log1p, 0x0.3333333333333p0,