http://defect.opensolaris.org/bz/show_bug.cgi?id=11784
amaguire <alan.maguire at sun.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |alan.maguire at sun.com
--- Comment #2 from amaguire <alan.maguire at sun.com> 2009-10-07 14:05:10 UTC
---
(In reply to comment #1)
> No SIGALRM leads me to believe that the pending alarm is cancelled (the
> pending
> alarm is cancelled by calling alarm(0)). When we need to set the alarm, we
> look at the first event in the timed_event_queue and set the alarm to the time
> remaining for that event to occur. What if the event is to occur "now", i.e.,
> 0 seconds are remaining?
good catch - according to the manpage:
If seconds is 0, a pending alarm request, if any, is cancelled.
so if we are enqueuing a timed event exactly at the time that the pending event
at the head of the queue is due, we effectively cancel the timer:
529 /*
530 * This event may be the first event in the queue. (Re)set the
531 * alarm to the remaining time of first event in the queue.
532 */
533 e = uu_list_first(timed_event_queue);
534 (void) alarm(e->event_time - now);
If (e->event_time - now) == 0, we cancel the alarm. It might make sense to
ensure a minimal alarm time of 1sec, i.e.
nextalarm = e->event_time - now;
if (nextalarm > 0)
(void) alarm(nextalarm)
else
(void) alarm(1);
This would ensure that we do trigger a SIGALARM to catch the elapsed events.
--
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
