gm_arch_signal_sleep: not interrupted...

Hi all,
   There's a GM_PANIC in drivers/linux/gm/gm_arch.c that's caused me some
trouble a few times.  Here's the code snippet from gm_arch_signal_sleep():

	if (atomic_read(&s->wake_cnt) > 0)
                atomic_dec(&s->wake_cnt);
        else if (signal_pending(gm_current))
                ret = GM_SLEEP_INTERRUPTED;
        else
                GM_PANIC (("gm_arch_signal_sleep: not interrupted, not woken and I am here ?!?!\n"));

This particular GM_PANIC has crashed my busiest machine 3 times in the last 6
months.  I'd like to know how the code gets to the GM_PANIC, and if we can
change the behaviour to something more corrective.

(btw, thanks to the coder that thought to make a syslog entry here or I'd have
no idea what was going on.)