5 x86-64 test failures on master, math/test-*-j0

Carlos O'Donell carlos@redhat.com
Fri Jul 24 01:09:55 GMT 2020


On 7/23/20 8:57 PM, Andreas K. Hüttel via Libc-alpha wrote:
> Ping pretty please. Still failing...

We fixed something similar for i686.

See: https://sourceware.org/glibc/wiki/Regeneration#Math_files

Can you please run `make regen-ulps` on that machine, update the ULPs file,
re-test to make sure it covers everything, and then submit the patch?

I'll review and apply it for the release.
 
> Am Freitag, 17. Juli 2020, 22:20:00 EEST schrieb Andreas K. Hüttel via Libc-
> alpha:
>> With today's master (e9422236a2dd4cf2b7e900af5e84706da68bd43e) I get the
>> following test failures:
>>
>> FAIL: math/test-double-j0
>> FAIL: math/test-float-j0
>> FAIL: math/test-float32-j0
>> FAIL: math/test-float32x-j0
>> FAIL: math/test-float64-j0
>>
>>
>> math/test-double-j0:
>>
>> testing double (without inline functions)
>> Failure: Test: j0_towardzero (0x2.602774p+0)
>> Result:
>>  is:          1.5262095721573004e-02   0x1.f41bbcff0cc83p-7
>>  should be:   1.5262095721573012e-02   0x1.f41bbcff0cc88p-7
>>  difference:  8.6736173798840354e-18   0x1.4000000000000p-57
>>  ulp       :  5.0000
>>  max.ulp   :  4.0000
>> Maximal error of `j0_towardzero'
>>  is      : 5 ulp
>>  accepted: 4 ulp
>>
>> Test suite completed:
>>   200 test cases plus 196 tests for exception flags and
>>     196 tests for errno executed.
>>   2 errors occurred.
>>
>>
>> math/test-float-j0:
>>
>> testing float (without inline functions)
>> Failure: Test: j0_towardzero (0x2.602774p+0)
>> Result:
>>  is:          1.52620896e-02   0x1.f41bb0p-7
>>  should be:   1.52620952e-02   0x1.f41bbcp-7
>>  difference:  5.58793544e-09   0x1.800000p-28
>>  ulp       :  6.0000
>>  max.ulp   :  5.0000
>> Maximal error of `j0_towardzero'
>>  is      : 6 ulp
>>  accepted: 5 ulp
>>
>> Test suite completed:
>>   160 test cases plus 156 tests for exception flags and
>>     156 tests for errno executed.
>>   2 errors occurred.
>>
>>
>> test-float32-j0:
>>
>> testing _Float32 (without inline functions)
>> Failure: Test: j0_towardzero (0x2.602774p+0)
>> Result:
>>  is:          1.52620896e-02   0x1.f41bb0p-7
>>  should be:   1.52620952e-02   0x1.f41bbcp-7
>>  difference:  5.58793544e-09   0x1.800000p-28
>>  ulp       :  6.0000
>>  max.ulp   :  5.0000
>> Maximal error of `j0_towardzero'
>>  is      : 6 ulp
>>  accepted: 5 ulp
>>
>> Test suite completed:
>>   160 test cases plus 156 tests for exception flags and
>>     156 tests for errno executed.
>>   2 errors occurred.
>>
>>
>> test-float32x-j0:
>>
>> testing _Float32x (without inline functions)
>> Failure: Test: j0_towardzero (0x2.602774p+0)
>> Result:
>>  is:          1.5262095721573004e-02   0x1.f41bbcff0cc83p-7
>>  should be:   1.5262095721573012e-02   0x1.f41bbcff0cc88p-7
>>  difference:  8.6736173798840354e-18   0x1.4000000000000p-57
>>  ulp       :  5.0000
>>  max.ulp   :  4.0000
>> Maximal error of `j0_towardzero'
>>  is      : 5 ulp
>>  accepted: 4 ulp
>>
>> Test suite completed:
>>   200 test cases plus 196 tests for exception flags and
>>     196 tests for errno executed.
>>   2 errors occurred.
>>
>>
>> test-float64-j0:
>>
>> testing _Float64 (without inline functions)
>> Failure: Test: j0_towardzero (0x2.602774p+0)
>> Result:
>>  is:          1.5262095721573004e-02   0x1.f41bbcff0cc83p-7
>>  should be:   1.5262095721573012e-02   0x1.f41bbcff0cc88p-7
>>  difference:  8.6736173798840354e-18   0x1.4000000000000p-57
>>  ulp       :  5.0000
>>  max.ulp   :  4.0000
>> Maximal error of `j0_towardzero'
>>  is      : 5 ulp
>>  accepted: 4 ulp
>>
>> Test suite completed:
>>   200 test cases plus 196 tests for exception flags and
>>     196 tests for errno executed.
>>   2 errors occurred.
>>
>>
>>
>> Machine:
>> Linux farino 5.4.50-gentoo-ah1 #1 SMP PREEMPT Fri Jul 3 18:00:55 EEST 2020
>> x86_64 Intel(R) Core(TM) i5-8265U CPU @ 1.60GHz GenuineIntel GNU/Linux
>>
>> *FLAGS: -march=native -O2 -pipe -ggdb -frecord-gcc-switches
>>
>> gcc (Gentoo 10.1.0-r2 p3) 10.1.0
>>
>> GNU ld (Gentoo 2.34 p4) 2.34.0
>>
>>
>> Thanks in advance,
>> Andreas
> 
> 


-- 
Cheers,
Carlos.



More information about the Libc-alpha mailing list