[Bug math/17678] New: Testing failed about expm1

[Bug math/17678] New: Testing failed about expm1

[wuxiang at iscas dot ac.cn]

https://sourceware.org/bugzilla/show_bug.cgi?id=17678

            Bug ID: 17678
           Summary: Testing failed about expm1
           Product: glibc
           Version: 2.19
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
          Assignee: unassigned at sourceware dot org
          Reporter: wuxiang at iscas dot ac.cn

The following error was encountered when build the eglibc-2.19.

make[2]: Target `check' not remade because of errors.
make[2]: Leaving directory `/home/nano/zwx/eglibc-2.19'
make[1]: *** [check] Error 2
make[1]: Leaving directory `/home/nano/zwx/eglibc-2.19/build-tree/i386-libc'
#
# Testsuite failures, someone should be working towards
# fixing these! They are listed here for the purpose of
# regression testing during builds.
# Format: , Error  [(ignored)]
#
annexc.out, Error 1 (ignored)
run-conformtest.out, Error 1 (ignored)
test-double.out, Error 1
test-float.out, Error 1
test-idouble.out, Error 1
test-ifloat.out, Error 1
tst-cputimer1.out, Error 1
***************
Encountered regressions that don't match expected failures
(debian/testsuite-checking/expected-results-i686-linux-gnu-libc):
test-double.out, Error 1
test-float.out, Error 1
test-idouble.out, Error 1
test-ifloat.out, Error 1
TEST test-double.out:
testing double (without inline functions)
Failure: Test: expm1_towardzero (-0x1p-64)
Result:
 is:         -5.42101086242752217003e-20  -0x1.00000000000000000000p-64
 should be:  -5.42101086242752156818e-20  -0x1.fffffffffffff0000000p-65
 difference:  6.01853107621011204079e-36   0x1.00000000000000000000p-117
 ulp       :  1.0000
 max.ulp   :  0.0000
Failure: Test: expm1_downward (-0x1p-64)
Result:
 is:         -5.42101086242752337375e-20  -0x1.00000000000010000000p-64
 should be:  -5.42101086242752217004e-20  -0x1.00000000000000000000p-64
 difference:  1.20370621524202240815e-35   0x1.00000000000000000000p-116
 ulp       :  1.0000
 max.ulp   :  0.0000
Failure: Test: expm1_upward (-0x1p-64)
Result:
 is:         -5.42101086242752217003e-20  -0x1.00000000000000000000p-64
 should be:  -5.42101086242752156818e-20  -0x1.fffffffffffff0000000p-65
 difference:  6.01853107621011204080e-36   0x1.00000000000000000000p-117
 ulp       :  1.0000
 max.ulp   :  0.0000

Test suite completed:
  15788 test cases plus 15136 tests for exception flags and
    15136 tests for errno executed.
  3 errors occurred.
TEST test-float.out:
testing float (without inline functions)
Failure: Test: expm1_towardzero (-0x1p-64)
Result:
 is:         -5.42101086242752217003e-20  -0x1.00000000000000000000p-64
 should be:  -5.42101053931009539151e-20  -0x1.fffffe00000000000000p-65
 difference:  3.23117426778526435496e-27   0x1.00000000000000000000p-88
 ulp       :  1.0000
 max.ulp   :  0.0000
Failure: Test: expm1_downward (-0x1p-64)
Result:
 is:         -5.42101150866237572710e-20  -0x1.00000200000000000000p-64
 should be:  -5.42101086242752217004e-20  -0x1.00000000000000000000p-64
 difference:  6.46234853557052870993e-27   0x1.00000000000000000000p-87
 ulp       :  1.0000
 max.ulp   :  0.0000
Failure: Test: expm1_upward (-0x1p-64)
Result:
 is:         -5.42101086242752217003e-20  -0x1.00000000000000000000p-64
 should be:  -5.42101053931009539151e-20  -0x1.fffffe00000000000000p-65
 difference:  3.23117426778526435497e-27   0x1.00000000000000000000p-88
 ulp       :  1.0000
 max.ulp   :  0.0000

Test suite completed:
  10839 test cases plus 10240 tests for exception flags and
    10240 tests for errno executed.
  3 errors occurred.
TEST test-idouble.out:
testing double (inline functions)
Failure: Test: expm1_downward (-0x1p-64)
Result:
 is:         -5.42101086242752337375e-20  -0x1.00000000000010000000p-64
 should be:  -5.42101086242752217004e-20  -0x1.00000000000000000000p-64
 difference:  1.20370621524202240815e-35   0x1.00000000000000000000p-116
 ulp       :  1.0000
 max.ulp   :  0.0000

Test suite completed:
  15636 test cases plus 14984 tests for exception flags and
    14984 tests for errno executed.
  1 errors occurred.
TEST test-ifloat.out:
testing float (inline functions)
Failure: Test: expm1_downward (-0x1p-64)
Result:
 is:         -5.42101150866237572710e-20  -0x1.00000200000000000000p-64
 should be:  -5.42101086242752217004e-20  -0x1.00000000000000000000p-64
 difference:  6.46234853557052870993e-27   0x1.00000000000000000000p-87
 ulp       :  1.0000
 max.ulp   :  0.0000

Test suite completed:
  10733 test cases plus 10134 tests for exception flags and
    10134 tests for errno executed.
  1 errors occurred.
Encountered progressions that don't match expected failures:
tst-backtrace6.out, Error 1
tst-cpuclock2.out, Error 1
tst-mqueue5.out, Error 1
tst-timer.out, Error 139
tst-waitid.out, Error 1
make: *** [/home/nano/zwx/eglibc-2.19/stamp-dir/check_libc] Error 1
dpkg-buildpackage: error: debian/rules build gave error exit status 2

Is it the "difference" not small enough? 
My os is: 32 bit ubuntu 14.04
the processor is: VIA Nano X2 L4350
Compiler: gcc-4.8

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/17678] Testing failed about expm1

[wuxiang at iscas dot ac.cn]

https://sourceware.org/bugzilla/show_bug.cgi?id=17678

--- Comment #1 from zhangwuxiang <wuxiang at iscas dot ac.cn> ---
I use the apt-get source command to get my eglibc source code on unbunt 14.04.
And  use dpkg-buildpackage to build the source for the deb packages.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/17678] Testing failed about expm1

[wuxiang at iscas dot ac.cn]

https://sourceware.org/bugzilla/show_bug.cgi?id=17678

zhangwuxiang <wuxiang at iscas dot ac.cn> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wuxiang at iscas dot ac.cn

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/17678] Testing failed about expm1

[joseph at codesourcery dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=17678

--- Comment #2 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
libm-test-ulps is regenerated each release cycle (at least if there are 
significant libm changes justifying full regeneration).  If you see these 
failures with current git, then during the release freeze in January you 
should send a patch resulting from "make regen-ulps" - we need people with 
different x86 processors and compilers to take part in the regeneration 
process each release cycle so libm-test-ulps reflects the range of ulps 
seen on different processors.  Filing bugs in Bugzilla for small ulps 
errors is not useful.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug math/17678] Testing failed about expm1

[jsm28 at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=17678

Joseph Myers <jsm28 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from Joseph Myers <jsm28 at gcc dot gnu.org> ---
As these ulps already appear in the libm-test-ulps file for the relevant
rounding modes of expm1, and current glibc (2.20 and later; anything with
commit e6b6a85705be16373fb3f78ef998edc62150499c) only lists ulps for functions
not for individual tests, this is in fact already fixed.

-- 
You are receiving this mail because:
You are on the CC list for the bug.