This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: glibc 2.21 -- ppc64le test results.


On 25-01-2015 19:44, Adhemerval Zanella wrote:
> On 24-01-2015 18:11, Carlos O'Donell wrote:
>> Adhemerval,
>>
>> If you have a chance I'd be interested in your opinion of the
>> ppc64le failures I see on master in my build environment and
>> if any of those are going to be serious for 2.21's release:
>>
>> https://sourceware.org/glibc/wiki/Release/2.21#Build_and_test_issues
>>
>> Cheers,
>> Carlos.
>>
> I tested on a Ubuntu 14.04 (kernel 3.13.0-35-generic, GCC 4.8.2, binutils 2.24)
> and I only see:
>
> FAIL: math/test-ildoubl
> FAIL: math/test-ldouble
> FAIL: nptl/tst-stack4
>
> The 'nptl/tst-stack4' seems to be a timing issue, since it is sporadic and running
> manually it does not show the issue:
>
> $ ./testrun.sh nptl/tst-stack4; echo $?
> 0

In fact I think it might be a race-condition issue, since now I am seeing 

$ ./testrun.sh nptl/tst-stack4; echo $?
Inconsistency detected by ld.so: dl-tls.c: 493: _dl_allocate_tls_init: Assertion `listp->slotinfo[cnt].gen <= _rtld_local._dl_tls_generation' failed!
127

Although it is not reproducible all the time.  I will take a look.

>
> Your 'math/test-double' and 'math/test-float' are a libgcc issue which are fixed in
> GCC 4.9 [1].
>
> The 'math/test-ldouble' issues are know from previous GLIBC version and it mainly
> related to libgcc issues as well [2] [3].  Joseph has advocate to fix some of them,
> but there not yet correct in GCC.  I also believe they might have some lingering 
> exception handling needed in GLIBC, but I didn't take a deep look yet.
>
> I do not see 'posix/tst-chmod' and 'posix/tst-getaddrinfo5'.  The 'tst-chmod' does
> look a kernel issue, which 'tst-getaddrinfo5' seems something related to you system
> IMOH.  The 'rt/tst-cpuclock2' I only see on a powerpc64 box (Fedora 19), which I 
> also think it is a kernel issue.
>
> [1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=57363
> [2] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59666
> [3] https://gcc.gnu.org/ml/gcc-patches/2014-01/msg00157.html
>


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]