Piteball
/
android_kernel_google_msm
mirror of https://github.com/followmsi/android_kernel_google_msm.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

tick: Upstream fixes 

* Addresses the issue of timers being scheduled on offline CPUs.

tick: Don't invoke tick_nohz_stop_sched_tick() if the cpu is offline

commit 5b39939a4 (nohz: Move ts->idle_calls incrementation into strict
idle logic) moved code out of tick_nohz_stop_sched_tick() and missed
to bail out when the cpu is offline. That's causing subsequent
failures as an offline CPU is supposed to die and not to fiddle with
nohz magic.

Return false in can_stop_idle_tick() if the cpu is offline.

Reported-and-tested-by: Jiri Kosina <jkosina@suse.cz>
Reported-and-tested-by: Prarit Bhargava <prarit@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Tony Luck <tony.luck@intel.com>
Cc: x86@kernel.org
Link: http://lkml.kernel.org/r/alpine.LFD.2.02.1305132138160.2863@ionos
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

timers: Consolidate base->next_timer update

Another bunch of mindlessly copied code. All callers of
internal_add_timer() except the recascading code updates
base->next_timer.

Move this into internal_add_timer() and let the cascading code call
__internal_add_timer().

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Gilad Ben-Yossef <gilad@benyossef.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20120525214819.189946224@linutronix.de

timers: Create detach_if_pending() and use it

Most callers of detach_timer() have the same pattern around
them. Check whether the timer is pending and eventually updating
base->next_timer.

Create detach_if_pending() and replace the duplicated code.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Gilad Ben-Yossef <gilad@benyossef.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20120525214819.131246037@linutronix.de

timers: Add accounting of non deferrable timers

The code in get_next_timer_interrupt() is suboptimal as it has to run
through the cascade to find the next expiring timer. On a completely
idle core we should only do that when there is an active timer
enqueued and base->next_timer does not give us a fast answer.

Add accounting of the active timers to the now consolidated
attach/detach code. I deliberately avoided sanity checks because the
code is fully symetric and any fiddling with timers w/o using the API
functions will lead to cute explosions anyway. ulong is big enough
even on 32bit and if we really run into the situation to have more
than 1<<32 timers enqueued there, then we are definitely not in a
state to go idle and run through that code.

This allows us to fix another shortcoming of get_next_timer_interrupt().

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Gilad Ben-Yossef <gilad@benyossef.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20120525214819.236377028@linutronix.de

timers: Improve get_next_timer_interrupt()

Gilad reported at

 http://lkml.kernel.org/r/1336056962-10465-2-git-send-email-gilad@benyossef.com

"Current timer code fails to correctly return a value meaning that
 there is no future timer event, with the result that the timer keeps
 getting re-armed in HZ one shot mode even when we could turn it off,
 generating unneeded interrupts.

 What is happening is that when __next_timer_interrupt() wishes
 to return a value that signifies "there is no future timer
 event", it returns (base->timer_jiffies + NEXT_TIMER_MAX_DELTA).

 However, the code in tick_nohz_stop_sched_tick(), which called
 __next_timer_interrupt() via get_next_timer_interrupt(),
 compares the return value to (last_jiffies + NEXT_TIMER_MAX_DELTA)
 to see if the timer needs to be re-armed.

 base->timer_jiffies != last_jiffies and so tick_nohz_stop_sched_tick()
 interperts the return value as indication that there is a distant
 future event 12 days from now and programs the timer to fire next
 after KTIME_MAX nsecs instead of avoiding to arm it. This ends up
 causing a needless interrupt once every KTIME_MAX nsecs."

Fix this by using the new active timer accounting. This avoids scans
when no active timer is enqueued completely, so we don't have to rely
on base->timer_next and base->timer_jiffies anymore.

Change-Id: I874ee5e5f837a228cebf5e6a084baf520d33cd0f
Reported-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20120525214819.317535385@linutronix.de
This commit is contained in:
Thomas Gleixner 2013-05-13 21:40:27 +02:00 committed by Artem Borisov
parent 6529714b12
commit 0cf6cbf5c2
2 changed files with 61 additions and 48 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
kernel/time/tick-sched.c
View File

@ -301,6 +301,7 @@ static void tick_nohz_stop_sched_tick(struct tick_sched *ts)
if (unlikely(!cpu_online(cpu))) {
if (cpu == tick_do_timer_cpu)
tick_do_timer_cpu = TICK_DO_TIMER_NONE;
return;
}
if (unlikely(ts->nohz_mode == NOHZ_MODE_INACTIVE))

108
kernel/timer.c
View File

@ -78,6 +78,7 @@ struct tvec_base {
struct timer_list *running_timer;
unsigned long timer_jiffies;
unsigned long next_timer;
unsigned long active_timers;
struct tvec_root tv1;
struct tvec tv2;
struct tvec tv3;
@ -333,7 +334,8 @@ void set_timer_slack(struct timer_list *timer, int slack_hz)
}
EXPORT_SYMBOL_GPL(set_timer_slack);
static void internal_add_timer(struct tvec_base *base, struct timer_list *timer)
static void
__internal_add_timer(struct tvec_base *base, struct timer_list *timer)
{
unsigned long expires = timer->expires;
unsigned long idx = expires - base->timer_jiffies;
@ -376,6 +378,19 @@ static void internal_add_timer(struct tvec_base *base, struct timer_list *timer)
list_add_tail(&timer->entry, vec);
}
static void internal_add_timer(struct tvec_base *base, struct timer_list *timer)
{
__internal_add_timer(base, timer);
/*
* Update base->active_timers and base->next_timer
*/
if (!tbase_get_deferrable(timer->base)) {
if (time_before(timer->expires, base->next_timer))
base->next_timer = timer->expires;
base->active_timers++;
}
}
#ifdef CONFIG_DEBUG_OBJECTS_TIMERS
static struct debug_obj_descr timer_debug_descr;
@ -626,8 +641,7 @@ void init_timer_deferrable_key(struct timer_list *timer,
}
EXPORT_SYMBOL(init_timer_deferrable_key);
static inline void detach_timer(struct timer_list *timer,
int clear_pending)
static inline void detach_timer(struct timer_list *timer, bool clear_pending)
{
struct list_head *entry = &timer->entry;
@ -639,6 +653,29 @@ static inline void detach_timer(struct timer_list *timer,
entry->prev = LIST_POISON2;
}
static inline void
detach_expired_timer(struct timer_list *timer, struct tvec_base *base)
{
detach_timer(timer, true);
if (!tbase_get_deferrable(timer->base))
timer->base->active_timers--;
}
static int detach_if_pending(struct timer_list *timer, struct tvec_base *base,
bool clear_pending)
{
if (!timer_pending(timer))
return 0;
detach_timer(timer, clear_pending);
if (!tbase_get_deferrable(timer->base)) {
timer->base->active_timers--;
if (timer->expires == base->next_timer)
base->next_timer = base->timer_jiffies;
}
return 1;
}
/*
* We are using hashed locking: holding per_cpu(tvec_bases).lock
* means that all timers which are tied to this base via timer->base are
@ -683,16 +720,9 @@ __mod_timer(struct timer_list *timer, unsigned long expires,
base = lock_timer_base(timer, &flags);
if (timer_pending(timer)) {
detach_timer(timer, 0);
if (timer->expires == base->next_timer &&
!tbase_get_deferrable(timer->base))
base->next_timer = base->timer_jiffies;
ret = 1;
} else {
if (pending_only)
goto out_unlock;
}
ret = detach_if_pending(timer, base, false);
if (!ret && pending_only)
goto out_unlock;
debug_activate(timer, expires);
@ -723,9 +753,6 @@ __mod_timer(struct timer_list *timer, unsigned long expires,
}
timer->expires = expires;
if (time_before(timer->expires, base->next_timer) &&
!tbase_get_deferrable(timer->base))
base->next_timer = timer->expires;
internal_add_timer(base, timer);
out_unlock:
@ -890,9 +917,6 @@ void add_timer_on(struct timer_list *timer, int cpu)
spin_lock_irqsave(&base->lock, flags);
timer_set_base(timer, base);
debug_activate(timer, timer->expires);
if (time_before(timer->expires, base->next_timer) &&
!tbase_get_deferrable(timer->base))
base->next_timer = timer->expires;
internal_add_timer(base, timer);
/*
* Check whether the other CPU is idle and needs to be
@ -928,13 +952,7 @@ int del_timer(struct timer_list *timer)
if (timer_pending(timer)) {
base = lock_timer_base(timer, &flags);
if (timer_pending(timer)) {
detach_timer(timer, 1);
if (timer->expires == base->next_timer &&
!tbase_get_deferrable(timer->base))
base->next_timer = base->timer_jiffies;
ret = 1;
}
ret = detach_if_pending(timer, base, true);
spin_unlock_irqrestore(&base->lock, flags);
}
@ -959,18 +977,9 @@ int try_to_del_timer_sync(struct timer_list *timer)
base = lock_timer_base(timer, &flags);
if (base->running_timer == timer)
goto out;
ret = 0;
if (timer_pending(timer)) {
detach_timer(timer, 1);
if (timer->expires == base->next_timer &&
!tbase_get_deferrable(timer->base))
base->next_timer = base->timer_jiffies;
ret = 1;
if (base->running_timer != timer) {
ret = detach_if_pending(timer, base, true);
}
out:
spin_unlock_irqrestore(&base->lock, flags);
return ret;
@ -1057,7 +1066,8 @@ static int cascade(struct tvec_base *base, struct tvec *tv, int index)
*/
list_for_each_entry_safe(timer, tmp, &tv_list, entry) {
BUG_ON(tbase_get_base(timer->base) != base);
internal_add_timer(base, timer);
/* No accounting, while moving them */
__internal_add_timer(base, timer);
}
return index;
@ -1142,7 +1152,7 @@ static inline void __run_timers(struct tvec_base *base)
data = timer->data;
base->running_timer = timer;
detach_timer(timer, 1);
detach_expired_timer(timer, base);
spin_unlock_irq(&base->lock);
call_timer_fn(timer, fn, data);
@ -1280,18 +1290,21 @@ static unsigned long cmp_next_hrtimer_event(unsigned long now,
unsigned long get_next_timer_interrupt(unsigned long now)
{
struct tvec_base *base = __this_cpu_read(tvec_bases);
unsigned long expires;
unsigned long expires = now + NEXT_TIMER_MAX_DELTA;
/*
* Pretend that there is no timer pending if the cpu is offline.
* Possible pending timers will be migrated later to an active cpu.
*/
if (cpu_is_offline(smp_processor_id()))
return now + NEXT_TIMER_MAX_DELTA;
return expires;
spin_lock(&base->lock);
if (time_before_eq(base->next_timer, base->timer_jiffies))
base->next_timer = __next_timer_interrupt(base);
expires = base->next_timer;
if (base->active_timers) {
if (time_before_eq(base->next_timer, base->timer_jiffies))
base->next_timer = __next_timer_interrupt(base);
expires = base->next_timer;
}
spin_unlock(&base->lock);
if (time_before_eq(expires, now))
@ -1668,6 +1681,7 @@ static int __cpuinit init_timers_cpu(int cpu)
base->timer_jiffies = jiffies;
base->next_timer = base->timer_jiffies;
base->active_timers = 0;
return 0;
}
@ -1678,11 +1692,9 @@ static void migrate_timer_list(struct tvec_base *new_base, struct list_head *hea
while (!list_empty(head)) {
timer = list_first_entry(head, struct timer_list, entry);
detach_timer(timer, 0);
/* We ignore the accounting on the dying cpu */
detach_timer(timer, false);
timer_set_base(timer, new_base);
if (time_before(timer->expires, new_base->next_timer) &&
!tbase_get_deferrable(timer->base))
new_base->next_timer = timer->expires;
internal_add_timer(new_base, timer);
}
}