From @ntyni

This is a bug report for perl from Niko Tyni <ntyni@​debian.org>,
generated with the help of perlbug 1.39 running under perl 5.15.1.

We're seeing a non-deterministic failure of test 25 (the last one)
of t/op/threads.t on multiprocessor Debian GNU/kFreeBSD systems.

The failure can be reduced to this​:

./perl -Ilib -Mthreads -e 'threads->create(sub {})->detach; fork'

which crashes here in a KVM virtual machine about 10% of the time.

The test triggering the crash was added during the 5.13 series in commit
2e0cfa1 fixing [perl #78494], but I can reproduce the crash
on 5.12.4 too.

It seems possible that the bug is in libc / kernel land, but FWIW I
haven't been able to trigger it on a FreeBSD 8.2-RELEASE system. I've
notified the Debian kFreeBSD porters, and will update this ticket if I
get more information.

If this turns out to be a bug somewhere else, perhaps the test should
be run in a fresh perl instance and skipped on gnukfreebsd?

The issue is tracked on the Debian side as http​://bugs.debian.org/628493

The backtrace is

  Program terminated with signal 11, Segmentation fault.
  #0 0x28491062 in nanosleep () at ../sysdeps/unix/syscall-template.S​:82
  82 ../sysdeps/unix/syscall-template.S​: No such file or directory.
  in ../sysdeps/unix/syscall-template.S
  (gdb) bt
  #0 0x28491062 in nanosleep () at ../sysdeps/unix/syscall-template.S​:82
  #1 0x2848f358 in __pthread_acquire (spinlock=<value optimized out>) at spinlock.c​:728
  #2 0x2848f4ce in wait_node_free (lock=0x83ebec0) at spinlock.c​:307
  #3 __pthread_alt_unlock (lock=0x83ebec0) at spinlock.c​:591
  #4 0x2848c85f in __pthread_mutex_unlock (mutex=0x83ebeb0) at mutex.c​:199
  #5 0x08159a51 in Perl_atfork_unlock () at util.c​:2975
  #6 0x2848cb86 in __pthread_fork (b=0x285c19e0) at ptfork.c​:72
  #7 0x2852dbd1 in __libc_fork () at ../linuxthreads/sysdeps/unix/sysv/linux/fork.c​:39
  #8 0x2848cab4 in __fork () at ptfork.c​:84
  #9 0x08159a8a in Perl_my_fork () at util.c​:2991
  #10 0x0829a080 in Perl_pp_fork (my_perl=0x83ebfb0) at pp_sys.c​:4021
  #11 0x08151247 in Perl_runops_debug (my_perl=0x83ebfb0) at dump.c​:2204
  #12 0x0808ef07 in S_run_body (my_perl=0x83ebfb0, oldscope=1) at perl.c​:2351
  #13 0x0808e208 in perl_run (my_perl=0x83ebfb0) at perl.c​:2269
  #14 0x0805fc9a in main (argc=5, argv=0xbfbfe7b8, env=0xbfbfe7d0) at perlmain.c​:120

Flags​:
  category=core
  severity=low

Site configuration information for perl 5.15.1​:

Configured by niko at Thu Aug 4 14​:19​:19 EEST 2011.

Summary of my perl5 (revision 5 version 15 subversion 1) configuration​:
  Local Commit​: 8a8f86a04d86583490b4ce4a9821ef57289394a9
  Ancestor​: 9cfd094
  Platform​:
  osname=gnukfreebsd, osvers=8.2-1-686-smp, archname=i686-gnukfreebsd-thread-multi
  uname='gnukfreebsd kbsd32 8.2-1-686-smp #0 sun jul 10 20​:35​:30 utc 2011 i686 i386 qemu virtual cpu version 0.14.1 gnukfreebsd '
  config_args='-Dusethreads -Dusedevel -DDEBUGGING -des -Doptimize=-g -O0'
  hint=recommended, useposix=true, d_sigaction=define
  useithreads=define, usemultiplicity=define
  useperlio=define, d_sfio=undef, uselargefiles=define, usesocks=undef
  use64bitint=undef, use64bitall=undef, uselongdouble=undef
  usemymalloc=n, bincompat5005=undef
  Compiler​:
  cc='cc', ccflags ='-D_REENTRANT -D_GNU_SOURCE -DTHREADS_HAVE_PIDS -DDEBUGGING -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64',
  optimize='-g -O0',
  cppflags='-D_REENTRANT -D_GNU_SOURCE -DTHREADS_HAVE_PIDS -DDEBUGGING -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include'
  ccversion='', gccversion='4.6.1', gccosandvers=''
  intsize=4, longsize=4, ptrsize=4, doublesize=8, byteorder=1234
  d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=12
  ivtype='long', ivsize=4, nvtype='double', nvsize=8, Off_t='off_t', lseeksize=8
  alignbytes=4, prototype=define
  Linker and Libraries​:
  ld='cc', ldflags =' -fstack-protector -L/usr/local/lib'
  libpth=/usr/local/lib /lib /usr/lib /usr/lib/i386-kfreebsd-gnu
  libs=-lnsl -lgdbm -ldb -ldl -lm -lcrypt -lutil -lpthread -lc -lgdbm_compat
  perllibs=-lnsl -ldl -lm -lcrypt -lutil -lpthread -lc
  libc=, so=so, useshrplib=false, libperl=libperl.a
  gnulibc_version='2.13'
  Dynamic Linking​:
  dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags='-Wl,-E'
  cccdlflags='-fpic', lddlflags='-shared -L/usr/local/lib -fstack-protector'

Locally applied patches​:
 

@​INC for perl 5.15.1​:
  lib
  /usr/local/lib/perl5/site_perl/5.15.1/i686-gnukfreebsd-thread-multi
  /usr/local/lib/perl5/site_perl/5.15.1
  /usr/local/lib/perl5/5.15.1/i686-gnukfreebsd-thread-multi
  /usr/local/lib/perl5/5.15.1
  .

Environment for perl 5.15.1​:
  HOME=/home/niko
  LANG=en_US.UTF-8
  LANGUAGE=en_US​:en
  LC_CTYPE=fi_FI.UTF-8
  LD_LIBRARY_PATH (unset)
  LOGDIR (unset)
  PATH=/usr/local/bin​:/usr/bin​:/bin​:/usr/local/games​:/usr/games
  PERL_BADLANG (unset)
  SHELL=/bin/bash

From @jkeenan

On Thu Aug 04 05​:19​:21 2011, ntyni@​debian.org wrote​:

This is a bug report for perl from Niko Tyni <ntyni@​debian.org>,
generated with the help of perlbug 1.39 running under perl 5.15.1.

-----------------------------------------------------------------
We're seeing a non-deterministic failure of test 25 (the last one)
of t/op/threads.t on multiprocessor Debian GNU/kFreeBSD systems.

The failure can be reduced to this​:

./perl -Ilib -Mthreads -e 'threads->create(sub {})->detach; fork'

which crashes here in a KVM virtual machine about 10% of the time.

List​: is there anyone testing a threaded perl on Debian GNU/kFreeBSD
who could comment on this problem?

Thank you very much.
Jim Keenan

The RT System itself - Status changed from 'new' to 'open'

From @stevenc99

Hi!

This seems to have been a bug in GNU/kFreeBSD's threads implementation
in eglibc; this was just confirmed as fixed and the Debian bug report
closed (http​://bugs.debian.org/628493) so I'm also marking it as
resolved here too.

Thank you.

From @stevenc99

I don't seem to have permissions to close this bug, but please see my
previous comment and either Niko or an admin here kindly mark this as
resolved. Thanks.

@cpansprout - Status changed from 'open' to 'resolved'