Re: [PATCH v2] cpuidle: haltpoll: Add trace points for guest_halt_poll_ns grow/shrink

[Eiichi Tsukata]

Hi Marcelo

> On May 26, 2022, at 2:31, Marcelo Tosatti <mtosatti@xxxxxxxxxx> wrote:
> 
> On Mon, May 23, 2022 at 11:53:32PM +0000, Eiichi Tsukata wrote:
>> Add trace points as are implemented in KVM host halt polling.
>> This helps tune guest halt polling params.
>> 
>> Signed-off-by: Eiichi Tsukata <eiichi.tsukata@xxxxxxxxxxx>
>> ---
>> drivers/cpuidle/governors/haltpoll.c | 15 ++++++++-----
>> include/trace/events/power.h         | 33 ++++++++++++++++++++++++++++
>> 2 files changed, 43 insertions(+), 5 deletions(-)
>> 
>> diff --git a/drivers/cpuidle/governors/haltpoll.c b/drivers/cpuidle/governors/haltpoll.c
>> index cb2a96eafc02..a5b6ad32956c 100644
>> --- a/drivers/cpuidle/governors/haltpoll.c
>> +++ b/drivers/cpuidle/governors/haltpoll.c
>> @@ -19,6 +19,7 @@
>> #include <linux/sched.h>
>> #include <linux/module.h>
>> #include <linux/kvm_para.h>
>> +#include <trace/events/power.h>
>> 
>> static unsigned int guest_halt_poll_ns __read_mostly = 200000;
>> module_param(guest_halt_poll_ns, uint, 0644);
>> @@ -77,13 +78,16 @@ static int haltpoll_select(struct cpuidle_driver *drv,
>> 
>> static void adjust_poll_limit(struct cpuidle_device *dev, u64 block_ns)
>> {
>> -	unsigned int val;
>> +	unsigned int val, old;
>> 
>> -	/* Grow cpu_halt_poll_us if
>> -	 * cpu_halt_poll_us < block_ns < guest_halt_poll_us
>> +	val = dev->poll_limit_ns;
>> +	old = val;
>> +
>> +	/* Grow poll_limit_ns if
>> +	 * poll_limit_ns < block_ns < guest_halt_poll_ns
>> 	 */
>> 	if (block_ns > dev->poll_limit_ns && block_ns <= guest_halt_poll_ns) {
>> -		val = dev->poll_limit_ns * guest_halt_poll_grow;
>> +		val *= guest_halt_poll_grow;
>> 
>> 		if (val < guest_halt_poll_grow_start)
>> 			val = guest_halt_poll_grow_start;
>> @@ -91,16 +95,17 @@ static void adjust_poll_limit(struct cpuidle_device *dev, u64 block_ns)
>> 			val = guest_halt_poll_ns;
>> 
> 
> Can do it before the assignment:
> 
> 		trace_guest_halt_poll_ns_grow(val, dev->poll_limit_ns);


Sure. Will fix it in v4.
Thanks

Eiichi