[question] what can trigger BUG at fork.c:251?

[Jan Hudec <bulb@ucw.cz>]

Hello,

I am writing a module and have this strange problem.
I start a kernel thread. When it runs, first ps ax passes ok, but
the second one thiggers this BUG. It does so between printing
kswapd and bdflush. The bug is triggered, when &init->mm count
gets to zero, which points to I am miscounting it somewhere.
But I dont touch it directly. Please suggest to me, what operations
could mater in triggering this problem.

Here is what I do when I start the thread:

i declare global variable loop_pid of type pid_t.
Then I start the kenrel thread with:
    loop_pid = kernel_thread(dfs_main_loop_thread, NULL,
			    CLONE_FS|CLONE_FILES|CLONE_VM);
(After initializing some of my data).
In dfs_mail_loop_thread followin code exists.

static int dfs_main_loop_thread(void *arg)
{
    int rv;
    DEBUG("kernel", 7, ("Initializing dfs main loop thread."));
    dsem_down_ni(&dfs_event_loop_sem); /* FIXME kontrola "nemozneho" stavu
                                          "ten semafor je ale dole!" */

    DEBUG("kernel", 7, ("daemonize"));
    daemonize(); /* magicka funkce co se musi volat v kernel-threadu */
    /* FIXME - pokud opravdu nebudeme handlovat signaly, tak se tady musi
     * zablokovat (viz reiserfs_journal_commit_thread) */
    sprintf(current->comm, "kdfsd");
    DEBUG("kernel", 7, ("daemonized"));

    DEBUG("kernel", 7, ("dnet_init"));
    rv = dnet_init();
    DEBUG("kernel", 7, ("dnet_init %i",rv));

    DEBUG("kernel", 7, ("dfs_identity_net_init"));
    rv = dfs_identity_net_init();
    DEBUG("kernel", 7, ("dfs_identity_net_init %i", rv));

    DEBUG("kernel", 7, ("event_loop"));
    dnet_event_loop();
    DEBUG("kernel", 7, ("event_looped"));

    DEBUG("kernel", 7, ("dnet_done"));
    rv = dnet_done();
    DEBUG("kernel", 7, ("dnet_done %i",rv));

    DEBUG("kernel", 7, ("Shuting down dfs main loop thread."));
    dsem_up(&dfs_event_loop_sem);
    return 0; /* Nikdo neceka na ukonceni (init, ale tomu je to jedno) */
}

The debug stuff are (bit insane) macros to print log - they do kmallocs/kfrees
and printks. The dnet_init adds current to a wait-queue if that could matter.
The dnet_event_loop() returns only when removind is reuqested (and cleanup
waits for it on the semoaphore).

				    Thanks

								Jan Hudec

--------------------------------------------------------------------------------
                  				- Jan Hudec `Bulb' <bulb@ucw.cz>
--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive:       http://mail.nl.linux.org/kernelnewbies/
IRC Channel:   irc.openprojects.net / #kernelnewbies
Web Page:      http://www.kernelnewbies.org/