Re: [Gc] Calling ‘GC_pthread_create’ from a pthread key destructor

[Ivan Maidanski]

Hi Ludo,

> pthread_setspecific (key, GC_STRDUP ("hello, world"));

Just one note. Are you aware that this code is equivalent (since TLS is not 
traced by GC) to:

(void)GC_STRDUP ("hello, world");
pthread_setspecific (key, <some_number>);

Regards.

Wed, 20 Apr 2011 16:20:08 +0200 address@hidden (Ludovic Courtès):

> Hello,
> 
> The program below quickly segfaults on GNU/Linux:
> 
> --8<---------------cut here---------------start------------->8---
> #define GC_THREADS 1
> #define GC_NO_THREAD_REDIRECTS 1
> 
> #include <gc/gc.h>
> #include <pthread.h>
> #include <stdlib.h>
> #include <stdlib.h>
> #include <unistd.h>
> 
> static pthread_key_t key;
> static pthread_once_t key_once = PTHREAD_ONCE_INIT;
> 
> static void *
> entry (void *arg)
> {
> pthread_setspecific (key, GC_STRDUP ("hello, world"));
> return arg;
> }
> 
> static void
> on_thread_exit (void *v)
> {
> pthread_t t;
> GC_pthread_create (&t, NULL, entry, NULL);
> }
> 
> static void
> make_key ()
> {
> pthread_key_create (&key, on_thread_exit);
> }
> 
> int
> main (int argc, char *argv[])
> {
> GC_INIT ();
> 
> pthread_once (&key_once, make_key);
> 
> while (1)
> {
> pthread_t t;
> GC_pthread_create (&t, NULL, entry, NULL);
> }
> 
> return EXIT_SUCCESS;
> }
> --8<---------------cut here---------------end--------------->8---
> 
> The backtrace varies, but it’s typically one of:
> 
> --8<---------------cut here---------------start------------->8---
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0x7fff7d722700 (LWP 1497)]
> GC_generic_malloc_inner (lb=56, k=1) at ../malloc.c:127
> 127             obj_link(op) = 0;
> (gdb) bt
> #0  GC_generic_malloc_inner (lb=56, k=1) at ../malloc.c:127
> #1  0x00007ffff7b916ef in GC_pthread_create (new_thread=0x7fff7d721e98,
> attr=0x0, start_routine=0x400904 <entry>, arg=0x0) at
> ../pthread_support.c:1474
> #2  0x000000000040095a in on_thread_exit (v=0x662d90) at t.c:24
> #3  0x00007ffff75f0b29 in __nptl_deallocate_tsd () from
> /nix/store/vxycd107wjbhcj720hzkw2px7s7kr724-glibc-2.12.2/lib/libpthread.so.0
> #4  0x00007ffff75f0cfa in start_thread () from
> /nix/store/vxycd107wjbhcj720hzkw2px7s7kr724-glibc-2.12.2/lib/libpthread.so.0
> #5  0x00007ffff78d81ed in clone () from
> /nix/store/vxycd107wjbhcj720hzkw2px7s7kr724-glibc-2.12.2/lib/libc.so.6
> --8<---------------cut here---------------end--------------->8---
> 
> or:
> 
> --8<---------------cut here---------------start------------->8---
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0x7fff9274c700 (LWP 1823)]
> 0x0000000000652fc0 in ?? ()
> (gdb) bt
> #0  0x0000000000652fc0 in ?? ()
> #1  0x00007ffff7b90791 in GC_inner_start_routine (sb=<value optimized out>,
> arg=<value optimized out>) at ../pthread_start.c:61
> #2  0x00007ffff7b8b6b5 in GC_call_with_stack_base (fn=<value optimized out>,
> arg=<value optimized out>) at ../misc.c:1505
> #3  0x00007ffff75f0cec in start_thread () from
> /nix/store/vxycd107wjbhcj720hzkw2px7s7kr724-glibc-2.12.2/lib/libpthread.so.0
> #4  0x00007ffff78d81ed in clone () from
> /nix/store/vxycd107wjbhcj720hzkw2px7s7kr724-glibc-2.12.2/lib/libc.so.6
> --8<---------------cut here---------------end--------------->8---
> 
> Replacing ‘GC_pthread_create’ by ‘pthread_create’ in ‘on_thread_exit’
> suffices to avoid the segfault.
> 
> Any ideas?  Is it something we can reasonably expect to work?
> 
> Thanks,
> Ludo’.
> 
> PS: For the record, Guile has a similar problem with ‘scm_spawn_thread’.
> 
> _______________________________________________
> Gc mailing list
> address@hidden
> http://www.hpl.hp.com/hosted/linux/mail-archives/gc/