Commit 84999b8b authored by Thomas Gleixner's avatar Thomas Gleixner
Browse files

posix-timers: Clarify posix_timer_fn() comments 



Make the issues vs. SIG_IGN understandable and remove the 15 years old
promise that a proper solution is already on the horizon.

Signed-off-by: default avatarThomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/874jnrdmrq.ffs@tglx
parent 02972d79

kernel/time/posix-timers.c

+32 −30
Original line number Diff line number Diff line
@@ -326,11 +326,11 @@ int posix_timer_event(struct k_itimer *timr, int si_private) 
}



/*

 * This function gets called when a POSIX.1b interval timer expires.  It

 * is used as a callback from the kernel internal timer.  The

 * run_timer_list code ALWAYS calls with interrupts on.



 * This code is for CLOCK_REALTIME* and CLOCK_MONOTONIC* timers.

 * This function gets called when a POSIX.1b interval timer expires from

 * the HRTIMER interrupt (soft interrupt on RT kernels).

 *

 * Handles CLOCK_REALTIME, CLOCK_MONOTONIC, CLOCK_BOOTTIME and CLOCK_TAI

 * based timers.

 */

static enum hrtimer_restart posix_timer_fn(struct hrtimer *timer)

{
@@ -348,9 +348,10 @@ static enum hrtimer_restart posix_timer_fn(struct hrtimer *timer) 


	if (posix_timer_event(timr, si_private)) {

		/*

		 * signal was not sent because of sig_ignor

		 * we will not get a call back to restart it AND

		 * it should be restarted.

		 * The signal was not queued due to SIG_IGN. As a

		 * consequence the timer is not going to be rearmed from

		 * the signal delivery path. But as a real signal handler

		 * can be installed later the timer must be rearmed here.

		 */

		if (timr->it_interval != 0) {

			ktime_t now = hrtimer_cb_get_time(timer);
@@ -359,34 +360,35 @@ static enum hrtimer_restart posix_timer_fn(struct hrtimer *timer) 
			 * FIXME: What we really want, is to stop this

			 * timer completely and restart it in case the

			 * SIG_IGN is removed. This is a non trivial

			 * change which involves sighand locking

			 * (sigh !), which we don't want to do late in

			 * the release cycle.

			 * change to the signal handling code.

			 *

			 * For now let timers with an interval less than a

			 * jiffie expire every jiffie and recheck for a

			 * valid signal handler.

			 *

			 * This avoids interrupt starvation in case of a

			 * very small interval, which would expire the

			 * timer immediately again.

			 *

			 * Moving now ahead of time by one jiffie tricks

			 * hrtimer_forward() to expire the timer later,

			 * while it still maintains the overrun accuracy

			 * for the price of a slight inconsistency in the

			 * timer_gettime() case. This is at least better

			 * than a timer storm.

			 *

			 * For now we just let timers with an interval

			 * less than a jiffie expire every jiffie to

			 * avoid softirq starvation in case of SIG_IGN

			 * and a very small interval, which would put

			 * the timer right back on the softirq pending

			 * list. By moving now ahead of time we trick

			 * hrtimer_forward() to expire the timer

			 * later, while we still maintain the overrun

			 * accuracy, but have some inconsistency in

			 * the timer_gettime() case. This is at least

			 * better than a starved softirq. A more

			 * complex fix which solves also another related

			 * inconsistency is already in the pipeline.

			 * Only required when high resolution timers are

			 * enabled as the periodic tick based timers are

			 * automatically aligned to the next tick.

			 */

#ifdef CONFIG_HIGH_RES_TIMERS

			{

				ktime_t kj = NSEC_PER_SEC / HZ;

			if (IS_ENABLED(CONFIG_HIGHRES_TIMERS)) {

				ktime_t kj = TICK_NSEC;



				if (timr->it_interval < kj)

					now = ktime_add(now, kj);

			}

#endif

			timr->it_overrun += hrtimer_forward(timer, now,

							    timr->it_interval);



			timr->it_overrun += hrtimer_forward(timer, now, timr->it_interval);

			ret = HRTIMER_RESTART;

			++timr->it_requeue_pending;

			timr->it_active = 1;