* [PATCH 2/2] nohz: don't fetch per_cpu variable before we need it
2014-04-25 18:57 [PATCH 1/2] nohz: move NOHZ code bits out of io_schedule{,_timeout} into a helper Denys Vlasenko
@ 2014-04-25 18:57 ` Denys Vlasenko
2014-04-29 15:13 ` [PATCH 1/2] nohz: move NOHZ code bits out of io_schedule{,_timeout} into a helper Frederic Weisbecker
1 sibling, 0 replies; 3+ messages in thread
From: Denys Vlasenko @ 2014-04-25 18:57 UTC (permalink / raw)
To: linux-kernel
Cc: Denys Vlasenko, Frederic Weisbecker, Hidetoshi Seto,
Fernando Luis Vazquez Cao, Tetsuo Handa, Thomas Gleixner,
Ingo Molnar, Peter Zijlstra, Andrew Morton, Arjan van de Ven,
Oleg Nesterov
In get_cpu_{idle,iowait}_time_us, we were fetching
per_cpu(tick_cpu_sched, cpu) upfront. But them we do
if (!tick_nohz_active)
return -1;
which may make that fetch pointless.
Moving it further down, closer to the usage point.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Cc: Fernando Luis Vazquez Cao <fernando_b1@lab.ntt.co.jp>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Oleg Nesterov <oleg@redhat.com>
---
kernel/time/tick-sched.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
index d78c942..4ffccd0 100644
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -476,7 +476,7 @@ void tick_nohz_iowait_to_idle(int cpu)
*/
u64 get_cpu_idle_time_us(int cpu, u64 *last_update_time)
{
- struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);
+ struct tick_sched *ts;
ktime_t now, idle;
unsigned int seq;
@@ -487,6 +487,8 @@ u64 get_cpu_idle_time_us(int cpu, u64 *last_update_time)
if (last_update_time)
*last_update_time = ktime_to_us(now);
+ ts = &per_cpu(tick_cpu_sched, cpu);
+
do {
ktime_t start, delta, iowait_exit;
@@ -531,7 +533,7 @@ EXPORT_SYMBOL_GPL(get_cpu_idle_time_us);
*/
u64 get_cpu_iowait_time_us(int cpu, u64 *last_update_time)
{
- struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);
+ struct tick_sched *ts;
ktime_t now, iowait;
unsigned int seq;
@@ -542,6 +544,8 @@ u64 get_cpu_iowait_time_us(int cpu, u64 *last_update_time)
if (last_update_time)
*last_update_time = ktime_to_us(now);
+ ts = &per_cpu(tick_cpu_sched, cpu);
+
do {
ktime_t delta, end;
--
1.8.1.4
^ permalink raw reply related [flat|nested] 3+ messages in thread* Re: [PATCH 1/2] nohz: move NOHZ code bits out of io_schedule{,_timeout} into a helper
2014-04-25 18:57 [PATCH 1/2] nohz: move NOHZ code bits out of io_schedule{,_timeout} into a helper Denys Vlasenko
2014-04-25 18:57 ` [PATCH 2/2] nohz: don't fetch per_cpu variable before we need it Denys Vlasenko
@ 2014-04-29 15:13 ` Frederic Weisbecker
1 sibling, 0 replies; 3+ messages in thread
From: Frederic Weisbecker @ 2014-04-29 15:13 UTC (permalink / raw)
To: Denys Vlasenko, Peter Zijlstra
Cc: linux-kernel, Hidetoshi Seto, Fernando Luis Vazquez Cao,
Tetsuo Handa, Thomas Gleixner, Ingo Molnar, Andrew Morton,
Arjan van de Ven, Oleg Nesterov
On Fri, Apr 25, 2014 at 08:57:29PM +0200, Denys Vlasenko wrote:
> Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
> Cc: Frederic Weisbecker <fweisbec@gmail.com>
> Cc: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
> Cc: Fernando Luis Vazquez Cao <fernando_b1@lab.ntt.co.jp>
> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Ingo Molnar <mingo@kernel.org>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Arjan van de Ven <arjan@linux.intel.com>
> Cc: Oleg Nesterov <oleg@redhat.com>
> ---
> kernel/sched/core.c | 33 +++++++++++++++++----------------
> 1 file changed, 17 insertions(+), 16 deletions(-)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index ffea757..3137980 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -4208,6 +4208,21 @@ EXPORT_SYMBOL_GPL(yield_to);
> * This task is about to go to sleep on IO. Increment rq->nr_iowait so
> * that process accounting knows that this is a task in IO wait state.
> */
> +#ifdef CONFIG_NO_HZ_COMMON
> +static __sched void io_wait_end(struct rq *rq)
> +{
> + if (atomic_dec_and_test(&rq->nr_iowait)) {
> + if (raw_smp_processor_id() != cpu_of(rq))
> + tick_nohz_iowait_to_idle(cpu_of(rq));
> + }
> +}
> +#else
> +static inline void io_wait_end(struct rq *rq)
> +{
> + atomic_dec(&rq->nr_iowait);
> +}
> +#endif
> +
> void __sched io_schedule(void)
> {
> struct rq *rq = raw_rq();
> @@ -4218,14 +4233,7 @@ void __sched io_schedule(void)
> current->in_iowait = 1;
> schedule();
> current->in_iowait = 0;
> -#ifdef CONFIG_NO_HZ_COMMON
> - if (atomic_dec_and_test(&rq->nr_iowait)) {
> - if (raw_smp_processor_id() != cpu_of(rq))
> - tick_nohz_iowait_to_idle(cpu_of(rq));
> - }
> -#else
> - atomic_dec(&rq->nr_iowait);
> -#endif
> + io_wait_end(rq);
> delayacct_blkio_end();
There is much more to unify that the iowait accounting between all
the io_schedule() declensions.
Peterz I think you had a patch to unify that a few month ago?
^ permalink raw reply [flat|nested] 3+ messages in thread