all: fix wildcard to work with make-4.x
In make-3.8x, the $(wildacrd) function would sort the entries,
while in make-4.x, it would just return the entries in any
unpredictable order [*]
Use the $(sort) function to get reproducible behaviour.
[*] Well, most probably the roder the entries appear when read
from readdir()
Reported-by: Andrew Ruder <andrew.ruder@elecsyscorp.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Tested-by: Andrew Ruder <andrew.ruder@elecsyscorp.com>
1 Original patch from: gentoo/src/patchsets/glibc/2.9/0090_all_glibc-math-tests.patch
3 -= BEGIN original header =-
4 From: "Joseph S. Myers" <joseph at codesourcery dot com>
5 To: libc-alpha at sourceware dot org
6 Date: Tue, 15 Jul 2008 14:43:04 +0000 (UTC)
9 If you test glibc on i686 with GCC 4.3, you get a test-ildoubl failure:
11 Failure: Test: expm1 (1) == M_El - 1.0
13 is: 1.71828182845904523532e+00 0xd.bf0a8b14576953500000p-3
14 should be: 1.71828182845904523543e+00 0xd.bf0a8b14576953600000p-3
15 difference: 1.08420217248550443401e-19 0x8.00000000000000000000p-66
18 Maximal error of `expm1'
22 What happens is that the inline expansion of expm1l uses __builtin_expm1l,
23 and GCC 4.3 optimizes calls to __builtin_expm1l with constant argument to
24 a correctly rounded result using MPFR. The result returned is thus the
25 value of e-1 rounded once to long double precision. However, the test
26 expects M_El - 1.0, and the result of rounding e to long double precision,
27 then subtracting 1, differs in the last place from the result of rounding
28 e-1 to long double precision (the latter has smaller exponent, and the
31 There are two obvious approaches possible to fixing this. The first patch
32 below changes the expectation to a decimal expansion for e-1 (taken from
33 that of M_El) rather than doing arithmetic in the expected value. This in
34 turn requires ulps to be set for the out-of-line version of expm1. It
35 might also need ulps to be set for the inline version for older compilers
36 if they should continue to pass the test, and possibly for other targets.
37 The second patch below takes the alternative approach of keeping the
38 existing expectation (which has the wrong bit in the last place) and
39 setting ulps for the inline expansion of expm1, which avoids the risk of
40 breaking the test for other targets.
42 2008-07-15 Joseph Myers <joseph@codesourcery.com>
44 * sysdeps/i386/fpu/libm-test-ulps: Add inline long double ulps for
47 -= END original header =-
49 diff -durN glibc-2_9.orig/sysdeps/i386/fpu/libm-test-ulps glibc-2_9/sysdeps/i386/fpu/libm-test-ulps
50 --- glibc-2_9.orig/sysdeps/i386/fpu/libm-test-ulps 2006-01-15 18:59:37.000000000 +0100
51 +++ glibc-2_9/sysdeps/i386/fpu/libm-test-ulps 2009-02-02 22:00:47.000000000 +0100
57 +Test "expm1 (1) == M_El - 1.0":
61 Test "gamma (-0.5) == log(2*sqrt(pi))":