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]

More tests failing under highly parallel builds.


During glibc-2.24 release testing for AArch64, ppc64, ppc64le,
and x86_64 I would, depending on the system load see:

FAIL: localedata/bug-iconv-trans
FAIL: localedata/tst-mbswcs6
FAIL: localedata/tst-trans
FAIL: localedata/tst-wctype
FAIL: localedata/tst_mblen
FAIL: localedata/tst_mbrlen
FAIL: localedata/tst_mbsrtowcs
FAIL: localedata/tst_mbstowcs
FAIL: localedata/tst_mbtowc
FAIL: localedata/tst_swscanf
FAIL: localedata/tst_wcrtomb
FAIL: localedata/tst_wcsrtombs
FAIL: localedata/tst_wcstombs
FAIL: localedata/tst_wctob
FAIL: localedata/tst_wctomb

They fail with SIGSEGV.

cat glibc/localedata/bug-iconv-trans.test-result 
FAIL: localedata/bug-iconv-trans
original exit status 139

Run manually the test passes just fine.

env GCONV_PATH=/root/build/glibc/iconvdata LOCPATH=/root/build/glibc/localedata LC_ALL=C   /root/build/glibc/elf/ld-linux-x86-64.so.2 --library-path /root/build/glibc:/root/build/glibc/math:/root/build/glibc/elf:/root/build/glibc/dlfcn:/root/build/glibc/nss:/root/build/glibc/nis:/root/build/glibc/rt:/root/build/glibc/resolv:/root/build/glibc/crypt:/root/build/glibc/mathvec:/root/build/glibc/nptl /root/build/glibc/localedata/bug-iconv-trans
output is "AEaeOEoeUEuess" which is OK

It looks like it could be a parallelism issue having to do with
the output of the shared objects which are used for conversions.

I reduced the parallelism from 24 down to 4 on x86_64 and the tests
pass, otherwise they just keep failing in a normal `make check`
sequence.

Has anyone else run across this?

I know we've seen conform failures related to parallelism.

-- 
Cheers,
Carlos.


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