Re: nl_langinfo-mt failure on Solaris 10 x86

[Dagobert Michelsen]

Hi Bruno,

Am 03.01.2020 um 23:28 schrieb Bruno Haible <address@hidden>:
>> I have one failing test on Solaris 10 x86:
>> 
>> FAIL: test-nl_langinfo-mt
>> =========================
>> 
>> thread5 disturbed by threadN!
>> FAIL test-nl_langinfo-mt (exit status: 134)
>> 
>> 
>> This looks like a new failure to me.
> 
> It's a new test, that verifies that the nl_langinfo function is
> multithread-safe. Its failure is harmless for GNU grep, since this
> program does not use multiple threads so far.
> 
> But I cannot reproduce the failure on the 'unstable10x' machine,
> even when passing the argument '30' (= timeout for the test, in seconds)
> to the test program.
> 
> Could you please tell
>  - on which version of Solaris exactly you're experiencing it,
>  - which compiler and options and configure options you specified,
>  - which of the three locales en_US.UTF-8, fr_FR.UTF-8, de_DE.UTF-8
>    your machine has (check "locale -a" output).

Then it is due to some settings, my error report is from unstable10x.
I am running with this rather complex commandline

/usr/bin/env -i HOME="/home/dam" 
PATH="/home/dam/mgar/pkg/.buildsys/v2/gar/bin/sos12-wrappers:/home/dam/mgar/pkg/grep/trunk/work/solaris10-i386/install-isa-pentium_pro/opt/csw/bin:/home/dam/mgar/pkg/grep/trunk/work/solaris10-i386/install-isa-pentium_pro/opt/csw/bin:/home/dam/mgar/pkg/grep/trunk/work/solaris10-i386/install-isa-pentium_pro/opt/csw/sbin:/home/dam/mgar/pkg/grep/trunk/work/solaris10-i386/install-isa-pentium_pro/opt/csw/sbin:/opt/csw/bin:/opt/csw/bin:/opt/csw/sbin:/opt/csw/sbin:/opt/SUNWspro/bin:/home/dam/mgar/pkg/.buildsys/v2/gar/bin:/usr/bin:/usr/sbin:/usr/java/bin:/usr/ccs/bin:/usr/openwin/bin"
 LC_ALL="C" prefix="/opt/csw" exec_prefix="/opt/csw" bindir="/opt/csw/bin" 
sbindir="/opt/csw/sbin" libexecdir="/opt/csw/libexec" datadir="/opt/csw/share" 
sysconfdir="/etc/opt/csw" sharedstatedir="/opt/csw/share" 
localstatedir="/var/opt/csw" libdir="/opt/csw/lib" 
infodir="/opt/csw/share/info" lispdir="/opt/csw/share/emacs/site-lisp" 
includedir="/opt/csw/include" mandir="/opt/csw/share/man" 
docdir="/opt/csw/share/doc" sourcedir="/opt/csw/src" 
CPPFLAGS="-I/opt/csw/include" CFLAGS="-xO3 -m32 -xarch=pentium_pro 
-xchip=pentium_pro" CXXFLAGS="-xO3 -m32 -xarch=pentium_pro -xchip=pentium_pro" 
LDFLAGS="-m32 -xarch=pentium_pro -xchip=pentium_pro -L/opt/csw/lib" 
FFLAGS="-xO3 -m32 -xarch=pentium_pro -xchip=pentium_pro" FCFLAGS="-xO3 -m32 
-xarch=pentium_pro -xchip=pentium_pro" F77="/opt/SUNWspro/bin/f77" 
FC="/opt/SUNWspro/bin/f95" ASFLAGS="" OPTFLAGS="-xO3 -m32 -xarch=pentium_pro 
-xchip=pentium_pro" CC="/opt/SUNWspro/bin/cc" CXX="/opt/SUNWspro/bin/CC" 
CC_HOME="/opt/SUNWspro" CC_VERSION="Sun C 5.9 SunOS_i386 Patch 124868-15 
2010/08/11" CXX_VERSION="Sun C++ 5.9 SunOS_i386 Patch 124864-30 2012/07/11" 
GARCH="i386" GAROSREL="5.10" GARPACKAGE="trunk" 
LD_OPTIONS="-R/opt/csw/lib/\$ISALIST -R/opt/csw/lib -M 
/home/dam/mgar/pkg/.buildsys/v2/gar/lib/map.solaris10 -B direct -z ignore"  
MAKEFLAGS= gmake -I/home/dam/mgar/pkg/.buildsys/v2   -C .  check

However, it is very hard to reproduce (and never with just „gmake check“)
and I hvan’t found a reliable pattern where it fails.
What I have found is now a coredump from time to time:

Abort - core dumped
FAIL: test-nl_langinfo-mt

Here is the stacktrace of the core:

core './work/solaris10-i386/build-isa-pentium_pro/grep-3.4/gnulib-tests/core' 
of 22614: ./test-nl_langinfo-mt
-----------------  lwp# 1 / thread# 1  --------------------
 feefb1a5 ___nanosleep (80457c8, 0) + 15
 fe0c2b27 nanosleep (80457c8, 0, feefdeb5, fee73721) + 1b
 08051286 main     (1, 8045808, 8045810, 80457fc) + 196
 08050d5d _start   (1, 804598c, 0, 80459a2, 80459ab, 80459d0) + 7d
-----------------  lwp# 2 / thread# 2  --------------------
 feeb5328 nl_langinfo (31, fee1f478) + 18
 08050e4a thread1_func (0) + 1a
 feef8aab _thr_setup (fe6f0200) + 4e
 feef8db0 _lwp_start (fe6f0200, 0, 0, fe7feff8, feef8db0, fe6f0200)
-----------------  lwp# 3 / thread# 3  --------------------
 feeb5208 __nl_langinfo_std (8062300, 30) + 38
 feeb5334 nl_langinfo (30, fe6eefb6) + 24
 08050eaa thread2_func (0) + 1a
 feef8aab _thr_setup (fe6f0a00) + 4e
 feef8db0 _lwp_start (fe6f0a00, 0, 0, fe6eeff8, feef8db0, fe6f0a00)
-----------------  lwp# 4 / thread# 4  --------------------
 0805130e rpl_nl_langinfo (0) + 1e
 feef8aab _thr_setup (fe6f1200) + 4e
 feef8db0 _lwp_start (fe6f1200, 0, 0, fe5efff8, feef8db0, fe6f1200)
-----------------  lwp# 5 / thread# 5  --------------------
 feef85b9 pthread_getspecific (3, 80, 0) + 50
 feeb51ef __nl_langinfo_std (8062300, 10) + 1f
 feeb5334 nl_langinfo (10, fe4f0fb4) + 24
 08050f6d thread4_func (0) + 1d
 feef8aab _thr_setup (fe6f1a00) + 4e
 feef8db0 _lwp_start (fe6f1a00, 0, 0, fe4f0ff8, feef8db0, fe6f1a00)
-----------------  lwp# 6 / thread# 6  --------------------
 feefc405 _lwp_kill (6, 6) + 15
 feea366f raise    (6) + 1f
 fee82971 abort    (feef7426, fef6f000, fe3f1fec, feef8aab, 0, 0) + cd
 08051004 thread5_func (0) + 54
 feef8aab _thr_setup (fe6f2200) + 4e
 feef8db0 _lwp_start (fe6f2200, 0, 0, fe3f1ff8, feef8db0, fe6f2200)
-----------------  lwp# 7 / thread# 7  --------------------
 feeb5328 nl_langinfo (27, fe2f2fb5) + 18
 0805102a thread6_func (0) + 1a
 feef8aab _thr_setup (fe6f2a00) + 4e
 feef8db0 _lwp_start (fe6f2a00, 0, 0, fe2f2ff8, feef8db0, fe6f2a00)
-----------------  lwp# 8 / thread# 8  --------------------
 08051303 rpl_nl_langinfo (27) + 13
 080510e0 threadN_func (0) + 70
 feef8aab _thr_setup (fe6f3200) + 4e
 feef8db0 _lwp_start (fe6f3200, 0, 0, fe1deff8, feef8db0, fe6f3200)

There are a lot of locales installed, especially there is
- en_US.UTF-8: installed
- fr_FR.UTF-8: installed
- de_DE.UTF-8: installed



Best regards

  — Dago

-- 
"You don't become great by trying to be great, you become great by wanting to 
do something,
and then doing it so hard that you become great in the process." - xkcd #896