> > It's a beauty wart, it's been there forever, and we know that machines
> > actually tend to survive it fine. So far I have _one_ person who has
>
> 1.2 survives mostly, 2.0 survives mostly, 2.1.x doesnt survive
hm, i've just reviewed the timer code from this angle again, and there is
a suspicious piece of code:
} else if (expires < timer_jiffies) {
/* can happen if you add a timer with expires == jiffies,
* or you set a timer to go off in the past
*/
insert_timer(timer, tv1.vec, tv1.index);
=====> } else if (idx < 0xffffffffUL) {
int i = (expires >> (TVR_BITS + 3 * TVN_BITS)) & TVN_MASK;
insert_timer(timer, tv5.vec, i);
} else {
/* Can only get here on architectures with 64-bit jiffies */
timer->next = timer->prev = timer;
}
shouldnt the (idx < 0xffffffffUL) condition be: (idx <= 0xffffffffUL)?
eg. if we have 'timer_jiffies == 0x0UL' (just wrapped around), and
'expires == 0xffffffffUL', then we have 'idx == 0xffffffffUL', but
(expires < timer_jiffies) is not true on a 32 bit box. Thus we get into
the last branch, which creates a timer with an infinit timeout ...
but this shouldnt happen too often i think when we just randomly wrap
jiffies (without heavy load that delays the processing of timer bhs), so i
guess there is something else going on.
-- mingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/