Re: [PATCH 1/2] Tick: Introduce tick_nohz_check() to check nohz enable status

[Lan Tianyu <tianyu.lan@intel.com>]

On 2013年10月29日 18:29, Lan Tianyu wrote:
> On 10/29/2013 05:51 PM, Paul E. McKenney wrote:
>> On Tue, Oct 29, 2013 at 04:48:56PM +0800, Lan Tianyu wrote:
>>> In some cases, nohz enable status needs to be checked. E.G, in RCU
>>> and cpufreq
>>> ondemand governor. So add tick_nohz_check() to return
>>> tick_nohz_enabled value
>>> And use tick_nohz_check() instead of referencing tick_nohz_enabled in
>>> the rcutree_plugin.h.
>>>
>>> Signed-off-by: Lan Tianyu <tianyu.lan@intel.com>
>>
>> NACK on the rcutree change unless you put the ACCESS_ONCE() in.
>>
>> Or is there some reason that ACCESS_ONCE() is not needed?  If so, what
>> is that reason?
> 
> Hi Paul:
> 
> Thanks for review. When I change this code, I find the tick_nohz_enabled
> isn't changed dynamically. It's only changed during parsing kernel
> params when "nohz=off/on" is set. Except this, it will not be changed.
> So I ignored ACCESS_ONCE(). If necessary, I can add it back.

Hi Paul:
	Does this reason make sense to you? Or you still prefer to add
ACCESS_ONCE() in the new tick_nohz_check()?

> 
>>
>>                             Thanx, Paul
>>
>>> ---
>>>   include/linux/tick.h     | 2 ++
>>>   kernel/rcutree_plugin.h  | 4 +---
>>>   kernel/time/tick-sched.c | 8 +++++++-
>>>   3 files changed, 10 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/include/linux/tick.h b/include/linux/tick.h
>>> index 5128d33..a9c5374 100644
>>> --- a/include/linux/tick.h
>>> +++ b/include/linux/tick.h
>>> @@ -136,6 +136,7 @@ static inline int tick_nohz_tick_stopped(void)
>>>   extern void tick_nohz_idle_enter(void);
>>>   extern void tick_nohz_idle_exit(void);
>>>   extern void tick_nohz_irq_exit(void);
>>> +extern int tick_nohz_check(void);
>>>   extern ktime_t tick_nohz_get_sleep_length(void);
>>>   extern u64 get_cpu_idle_time_us(int cpu, u64 *last_update_time);
>>>   extern u64 get_cpu_iowait_time_us(int cpu, u64 *last_update_time);
>>> @@ -155,6 +156,7 @@ static inline ktime_t
>>> tick_nohz_get_sleep_length(void)
>>>
>>>       return len;
>>>   }
>>> +static inline int tick_nohz_check(void) { return 0; }
>>>   static inline u64 get_cpu_idle_time_us(int cpu, u64 *unused) {
>>> return -1; }
>>>   static inline u64 get_cpu_iowait_time_us(int cpu, u64 *unused) {
>>> return -1; }
>>>   # endif /* !CONFIG_NO_HZ_COMMON */
>>> diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h
>>> index 130c97b..af167ec 100644
>>> --- a/kernel/rcutree_plugin.h
>>> +++ b/kernel/rcutree_plugin.h
>>> @@ -1627,8 +1627,6 @@ module_param(rcu_idle_gp_delay, int, 0644);
>>>   static int rcu_idle_lazy_gp_delay = RCU_IDLE_LAZY_GP_DELAY;
>>>   module_param(rcu_idle_lazy_gp_delay, int, 0644);
>>>
>>> -extern int tick_nohz_enabled;
>>> -
>>>   /*
>>>    * Try to advance callbacks for all flavors of RCU on the current CPU.
>>>    * Afterwards, if there are any callbacks ready for immediate
>>> invocation,
>>> @@ -1718,7 +1716,7 @@ static void rcu_prepare_for_idle(int cpu)
>>>       int tne;
>>>
>>>       /* Handle nohz enablement switches conservatively. */
>>> -    tne = ACCESS_ONCE(tick_nohz_enabled);
>>> +    tne = tick_nohz_check();
>>>       if (tne != rdtp->tick_nohz_enabled_snap) {
>>>           if (rcu_cpu_has_callbacks(cpu, NULL))
>>>               invoke_rcu_core(); /* force nohz to see update. */
>>> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
>>> index 3612fc7..d381a22 100644
>>> --- a/kernel/time/tick-sched.c
>>> +++ b/kernel/time/tick-sched.c
>>> @@ -361,7 +361,13 @@ void __init tick_nohz_init(void)
>>>   /*
>>>    * NO HZ enabled ?
>>>    */
>>> -int tick_nohz_enabled __read_mostly  = 1;
>>> +static int tick_nohz_enabled __read_mostly  = 1;
>>> +
>>> +int tick_nohz_check(void)
>>> +{
>>> +    return    tick_nohz_enabled;
>>> +}
>>> +EXPORT_SYMBOL_GPL(tick_nohz_check);
>>>
>>>   /*
>>>    * Enable / Disable tickless mode
>>> -- 
>>> 1.8.4.rc0.1.g8f6a3e5.dirty
>>>
>>
> 


-- 
Best regards
Tianyu Lan