[2.6.22.1] spurious (?) message "BUG: lock held at task exit time"

[Tilman Schmidt]

When testing the serial Gigaset driver with kernel release 2.6.22.1
I get a kernel BUG message:

=====================================
[ BUG: lock held at task exit time! ]
-------------------------------------
gigasetm101d/5970 is exiting with locks still held!
1 lock held by gigasetm101d/5970:
 #0:  (&cs->hw.ser->dead_mutex){--..}, at: [<c02d8bc7>] mutex_lock+0x1c/0x1f

which is all well and true except the process that has exited is the
grandfather process of a daemon() call, and the daemonized grandchild
process is still running happily, keeping the serial device open and
ensuring the line discipline stays loaded, and therefore legally
holding the dead_mutex:

ts@xenon:~> ps lw 5992
F   UID   PID  PPID PRI  NI    VSZ   RSS WCHAN  STAT TTY        TIME COMMAND
1     0  5992     1  16   0   1432   168 pause  S    ?          0:00 ./gigaset/driver/daemon/gigasetm101d /dev/ttyS0

The structure of gigasetm101d is basically:

    if ((tty_fd = open(dev, O_RDWR|O_NOCTTY)) < 0) {
        fprintf(stderr, "%s: cannot open %s: %s\n", prog, dev, strerror(errno));
        exit(EXIT_FAILURE);
    }
    if (ioctl(tty_fd, TIOCSETD, &ldisc) < 0)
    {
        fprintf(stderr, "%s: cannot set line discipline: %s\n",
                prog, strerror(errno));
        exit(EXIT_FAILURE);
    }
    if (daemon (0, 0) != 0)
    {
        fprintf(stderr, "%s: cannot daemonize: %s\n",
                prog, strerror(errno));
        exit(EXIT_FAILURE);
    }
    pause();
    exit(EXIT_SUCCESS);

Who is at fault here?

Thanks,
Tilman

-- 
Tilman Schmidt                    E-Mail: tilman@xxxxxxx
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)

Attachment: signature.asc
Description: OpenPGP digital signature