[bug #32436] bugs entering guile from pthreads

[bug #32436] bugs entering guile from pthreads

[Andy Wingo]

URL:
  <http://savannah.gnu.org/bugs/?32436>

                 Summary: bugs entering guile from pthreads
                 Project: Guile
            Submitted by: wingo
            Submitted on: Sat 12 Feb 2011 02:16:29 PM GMT
                Category: None
                Severity: 3 - Normal
              Item Group: None
                  Status: None
                 Privacy: Public
             Assigned to: None
             Open/Closed: Open
         Discussion Lock: Any

    _______________________________________________________

Details:

Cedric Cellier posted an interesting bug report back in September:

  http://thread.gmane.org/gmane.lisp.guile.user/8169

It was about 1.8, but the bugs are different in 1.9.  Specifically, the
program as it is segfaults directly, here with libgc from CVS:


Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7ffff76d3710 (LWP 2983)]
0x00007ffff7a016e3 in GC_is_black_listed (h=0x633000, len=<value optimized
out>) at blacklst.c:233
233	      if (get_pht_entry_from_index(GC_old_normal_bl, index)
(gdb) thr apply all bt

Thread 2 (Thread 0x7ffff76d3710 (LWP 2983)):
#0  0x00007ffff7a016e3 in GC_is_black_listed (h=0x633000, len=<value
optimized out>) at blacklst.c:233
#1  0x00007ffff79ff3db in GC_allochblk_nth (sz=512, kind=1, flags=0, n=14,
may_split=<value optimized out>)
    at allchblk.c:694
#2  0x00007ffff79ff7dc in GC_allochblk (sz=512, kind=1, flags=0) at
allchblk.c:629
#3  0x00007ffff7a0c2fb in GC_new_hblk (gran=32, kind=1) at new_hblk.c:191
#4  0x00007ffff7a01593 in GC_allocobj (gran=32, kind=1) at alloc.c:1312
#5  0x00007ffff7a0658a in GC_generic_malloc_inner (lb=488, k=1) at
malloc.c:126
#6  0x00007ffff7a06638 in GC_generic_malloc (lb=488, k=1) at malloc.c:166
#7  0x00007ffff7a06924 in GC_core_malloc (lb=488) at malloc.c:289
#8  0x00007ffff7d48c0a in guilify_self_1 (base=0x7ffff76d2da8) at
threads.c:379
#9  0x00007ffff7cf762c in scm_i_init_guile (base=<value optimized out>) at
init.c:408
#10 0x00007ffff7d4a6f8 in scm_i_init_thread_for_guile (base=0x7ffff76d2da8,
parent=0x0) at threads.c:649
#11 0x00007ffff7d4a9a6 in scm_i_with_guile_and_parent (func=0x400704
<thread_with_guile>, data=0x0, 
    parent=<value optimized out>) at threads.c:786
#12 0x000000000040072e in the_thread ()
#13 0x00007ffff7a103c5 in GC_inner_start_routine (sb=<value optimized out>,
arg=<value optimized out>) at pthread_start.c:59
#14 0x00007ffff7a0a525 in GC_call_with_stack_base (fn=<value optimized out>,
arg=<value optimized out>) at misc.c:1491
#15 0x0000003cc8c07761 in start_thread (arg=0x7ffff76d3710) at
pthread_create.c:301
#16 0x0000003cc84e14fd in clone () at
../sysdeps/unix/sysv/linux/x86_64/clone.S:115

Thread 1 (Thread 0x7ffff76d5700 (LWP 2980)):
#0  __lll_lock_wait () at
../nptl/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:136
#1  0x0000003cc8c092c5 in _L_lock_868 () from /lib64/libpthread.so.0
#2  0x0000003cc8c09197 in __pthread_mutex_lock (mutex=0x7ffff7c2b740) at
pthread_mutex_lock.c:61
#3  0x00007ffff7a11629 in GC_pthread_create (new_thread=0x7fffffffe058,
attr=0x0, start_routine=0x400713 <the_thread>, 
    arg=0x0) at pthread_support.c:1508
#4  0x0000000000400758 in main ()



But if you do a scm_init_guile in the main thread first, instead we get,
eventually:


Program received signal SIGABRT, Aborted.
[Switching to Thread 0x7ffff741c710 (LWP 2845)]
0x0000003cc84329a5 in raise (sig=6) at
../nptl/sysdeps/unix/sysv/linux/raise.c:64
64	  return INLINE_SYSCALL (tgkill, 3, pid, selftid, sig);
(gdb) bt
#0  0x0000003cc84329a5 in raise (sig=6) at
../nptl/sysdeps/unix/sysv/linux/raise.c:64
#1  0x0000003cc8434185 in abort () at abort.c:92
#2  0x00007ffff7d48d5d in guilify_self_1 (base=0x7ffff741bda8) at
threads.c:422
#3  0x00007ffff7d4a6de in scm_i_init_thread_for_guile (base=0x7ffff741bda8,
parent=0x92e730) at threads.c:658
#4  0x00007ffff7d4a9a6 in scm_i_with_guile_and_parent (func=0x400754
<thread_with_guile>, data=0x0, 
    parent=<value optimized out>) at threads.c:786
#5  0x000000000040077e in the_thread ()
#6  0x00007ffff7a103c5 in GC_inner_start_routine (sb=<value optimized out>,
arg=<value optimized out>) at pthread_start.c:59
#7  0x00007ffff7a0a525 in GC_call_with_stack_base (fn=<value optimized out>,
arg=<value optimized out>) at misc.c:1491
#8  0x0000003cc8c07761 in start_thread (arg=0x7ffff741c710) at
pthread_create.c:301
#9  0x0000003cc84e14fd in clone () at
../sysdeps/unix/sysv/linux/x86_64/clone.S:115
(gdb) fr 2
#2  0x00007ffff7d48d5d in guilify_self_1 (base=0x7ffff741bda8) at
threads.c:422
422	    abort ();
(gdb) l
417	
418	  if (pipe (t->sleep_pipe) != 0)
419	    /* FIXME: Error conditions during the initialization phase are
handled
420	       gracelessly since public functions such as `scm_init_guile ()'
421	       currently have type `void'.  */
422	    abort ();
423	
424	  scm_i_pthread_mutex_init (&t->admin_mutex, NULL);
425	  t->current_mark_stack_ptr = NULL;
426	  t->current_mark_stack_limit = NULL;


It seems that the sleep pipe is never being closed, which would seem to
indicate that the threads's cleanup function is not being called.




    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?32436>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/

[bug #32436] bugs entering guile from pthreads

[Andy Wingo]

Additional Item Attachment, bug #32436 (project guile):

File name: many_threads.c                 Size:0 KB
File name: Makefile                       Size:0 KB


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?32436>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/

[bug #32436] bugs entering guile from pthreads

[Andy Wingo]

Follow-up Comment #1, bug #32436 (project guile):

The is_blacklisted bug is tricky, and is discussed here:

    http://thread.gmane.org/gmane.lisp.guile.bugs/5340


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?32436>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/

[bug #32436] bugs entering guile from pthreads

[Andy Wingo]

Follow-up Comment #2, bug #32436 (project guile):

If I scm_init_guile, where it used to not clean up, now it fails fast (and
incorrectly).

http://thread.gmane.org/gmane.lisp.guile.bugs/5342

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?32436>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/