[PATCH 0/3] cpuidle: small improvements & fixes for menu governor (resend)

[PATCH 0/3] cpuidle: small improvements & fixes for menu governor (resend)

[riel]

While working on a paravirt cpuidle driver for KVM guests, I
noticed a number of small logic errors in the menu governor
code.

These patches should get rid of some artifacts that can break
the logic in the menu governor under certain corner cases, and
make idle state selection work better on CPUs with long C1 exit
latencies.

I have not seen any adverse effects with them in my (quick)
tests. As expected, they do not seem to do much on systems with
many power states and very low C1 exit latencies and target residencies. 

[PATCH 1/3] cpuidle,x86: increase forced cut-off for polling to 20us

[riel]

From: Rik van Riel <riel@redhat.com>

The cpuidle menu governor has a forced cut-off for polling at 5us,
in order to deal with firmware that gives the OS bad information
on cpuidle states, leading to the system spending way too much time
in polling.

However, at least one x86 CPU family (Atom) has chips that have
a 20us break-even point for C1. Forcing the polling cut-off to
less than that wastes performance and power.

Increase the polling cut-off to 20us.

Systems with a lower C1 latency will be found in the states table by
the menu governor, which will pick those states as appropriate.

Signed-off-by: Rik van Riel <riel@redhat.com>
---
 drivers/cpuidle/governors/menu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index 22e4463d1787..ecc242a586c9 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -330,7 +330,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * We want to default to C1 (hlt), not to busy polling
 	 * unless the timer is happening really really soon.
 	 */
-	if (data->next_timer_us > 5 &&
+	if (data->next_timer_us > 20 &&
 	    !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
 		dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
 		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
-- 
2.1.0

Re: [PATCH 1/3] cpuidle,x86: increase forced cut-off for polling to 20us

[Arjan van de Ven]

Acked-by: Arjan van de Ven <arjan@linux.intel.com>

[PATCH 2/3] cpuidle,menu: use interactivity_req to disable polling

[riel]

From: Rik van Riel <riel@redhat.com>

The menu governor carefully figures out how much time we typically
sleep for an estimated sleep interval, or whether there is a repeating
pattern going on, and corrects that estimate for the CPU load.

Then it proceeds to ignore that information when determining whether
or not to consider polling. This is not a big deal on most x86 CPUs,
which have very low C1 latencies, and the patch should not have any
effect on those CPUs.

However, certain CPUs (eg. Atom) have much higher C1 latencies, and
it would be good to not waste performance and power on those CPUs if
we are expecting a very low wakeup latency.

Disable polling based on the estimated interactivity requirement, not
on the time to the next timer interrupt.

Signed-off-by: Rik van Riel <riel@redhat.com>
---
 drivers/cpuidle/governors/menu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index ecc242a586c9..b1a55731f921 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -330,7 +330,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * We want to default to C1 (hlt), not to busy polling
 	 * unless the timer is happening really really soon.
 	 */
-	if (data->next_timer_us > 20 &&
+	if (interactivity_req > 20 &&
 	    !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
 		dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
 		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
-- 
2.1.0

Re: [PATCH 2/3] cpuidle,menu: use interactivity_req to disable polling

[Arjan van de Ven]

On 11/3/2015 2:34 PM, riel@redhat.com wrote:
> From: Rik van Riel <riel@redhat.com>
>
> The menu governor carefully figures out how much time we typically
> sleep for an estimated sleep interval, or whether there is a repeating
> pattern going on, and corrects that estimate for the CPU load.
>
> Then it proceeds to ignore that information when determining whether
> or not to consider polling. This is not a big deal on most x86 CPUs,
> which have very low C1 latencies, and the patch should not have any
> effect on those CPUs.
>
> However, certain CPUs (eg. Atom) have much higher C1 latencies, and
> it would be good to not waste performance and power on those CPUs if
> we are expecting a very low wakeup latency.
>
> Disable polling based on the estimated interactivity requirement, not
> on the time to the next timer interrupt.
>
good catch!

Acked-by: Arjan van de Ven <arjan@linux.intel.com>

Re: [PATCH 2/3] cpuidle,menu: use interactivity_req to disable polling

[Sudeep Holla]

Hi Rik,

This change break idle on ARM64(may be on other ARM?) platform.
Sorry for reporting late, but missed to check cpuidle in -next.

On Tue, Nov 3, 2015 at 10:34 PM,  <riel@redhat.com> wrote:
> From: Rik van Riel <riel@redhat.com>
>
> The menu governor carefully figures out how much time we typically
> sleep for an estimated sleep interval, or whether there is a repeating
> pattern going on, and corrects that estimate for the CPU load.
>
> Then it proceeds to ignore that information when determining whether
> or not to consider polling. This is not a big deal on most x86 CPUs,
> which have very low C1 latencies, and the patch should not have any
> effect on those CPUs.
>
> However, certain CPUs (eg. Atom) have much higher C1 latencies, and
> it would be good to not waste performance and power on those CPUs if
> we are expecting a very low wakeup latency.
>
> Disable polling based on the estimated interactivity requirement, not
> on the time to the next timer interrupt.
>
> Signed-off-by: Rik van Riel <riel@redhat.com>
> ---
>  drivers/cpuidle/governors/menu.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
> index ecc242a586c9..b1a55731f921 100644
> --- a/drivers/cpuidle/governors/menu.c
> +++ b/drivers/cpuidle/governors/menu.c
> @@ -330,7 +330,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
>          * We want to default to C1 (hlt), not to busy polling
>          * unless the timer is happening really really soon.
>          */
> -       if (data->next_timer_us > 20 &&
> +       if (interactivity_req > 20 &&

I found that data->predicted_us is gets overwritten in get_typical_interval
when avg computed = 0 which is the case initially on boot when the past
intervals are not yet accumulated.

I just tried a hack and that seem to work and proved what I anticipated
(i.e. interactivity_req = 0). Let me know if you have any clues on how to
solve it ? I can help you getting the change tested.

Regards,
Sudeep

--->8

diff --git i/drivers/cpuidle/governors/menu.c w/drivers/cpuidle/governors/menu.c
index 7b0971d97cc3..7c7f4059705a 100644
--- i/drivers/cpuidle/governors/menu.c
+++ w/drivers/cpuidle/governors/menu.c
@@ -330,7 +330,8 @@ static int menu_select(struct cpuidle_driver *drv,
struct cpuidle_device *dev)
         * We want to default to C1 (hlt), not to busy polling
         * unless the timer is happening really really soon.
         */
-       if (interactivity_req > 20 &&
+       if (((interactivity_req && interactivity_req > 20) ||
+               data->next_timer_us > 20) &&
            !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
                dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
                data->last_state_idx = CPUIDLE_DRIVER_STATE_START;

Re: [PATCH 2/3] cpuidle,menu: use interactivity_req to disable polling

[Rafael J. Wysocki]

Hi,

On Wed, Jan 13, 2016 at 6:27 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
> Hi Rik,
>
> This change break idle on ARM64(may be on other ARM?) platform.
> Sorry for reporting late, but missed to check cpuidle in -next.

OK, so first of all, how exactly is idle broken on those systems?

Do they crash or does something different happen?  If something
different happens, then what's that?

> On Tue, Nov 3, 2015 at 10:34 PM,  <riel@redhat.com> wrote:
>> From: Rik van Riel <riel@redhat.com>
>>
>> The menu governor carefully figures out how much time we typically
>> sleep for an estimated sleep interval, or whether there is a repeating
>> pattern going on, and corrects that estimate for the CPU load.
>>
>> Then it proceeds to ignore that information when determining whether
>> or not to consider polling. This is not a big deal on most x86 CPUs,
>> which have very low C1 latencies, and the patch should not have any
>> effect on those CPUs.
>>
>> However, certain CPUs (eg. Atom) have much higher C1 latencies, and
>> it would be good to not waste performance and power on those CPUs if
>> we are expecting a very low wakeup latency.
>>
>> Disable polling based on the estimated interactivity requirement, not
>> on the time to the next timer interrupt.
>>
>> Signed-off-by: Rik van Riel <riel@redhat.com>
>> ---
>>  drivers/cpuidle/governors/menu.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
>> index ecc242a586c9..b1a55731f921 100644
>> --- a/drivers/cpuidle/governors/menu.c
>> +++ b/drivers/cpuidle/governors/menu.c
>> @@ -330,7 +330,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
>>          * We want to default to C1 (hlt), not to busy polling
>>          * unless the timer is happening really really soon.
>>          */
>> -       if (data->next_timer_us > 20 &&
>> +       if (interactivity_req > 20 &&
>
> I found that data->predicted_us is gets overwritten in get_typical_interval
> when avg computed = 0 which is the case initially on boot when the past
> intervals are not yet accumulated.
>
> I just tried a hack and that seem to work and proved what I anticipated
> (i.e. interactivity_req = 0). Let me know if you have any clues on how to
> solve it ? I can help you getting the change tested.
>
> Regards,
> Sudeep
>
> --->8
>
> diff --git i/drivers/cpuidle/governors/menu.c w/drivers/cpuidle/governors/menu.c
> index 7b0971d97cc3..7c7f4059705a 100644
> --- i/drivers/cpuidle/governors/menu.c
> +++ w/drivers/cpuidle/governors/menu.c
> @@ -330,7 +330,8 @@ static int menu_select(struct cpuidle_driver *drv,
> struct cpuidle_device *dev)
>          * We want to default to C1 (hlt), not to busy polling
>          * unless the timer is happening really really soon.
>          */
> -       if (interactivity_req > 20 &&
> +       if (((interactivity_req && interactivity_req > 20) ||

Well, if interactivity_req > 20, then it also is different from 0, so
the first check should not be necessary here.

> +               data->next_timer_us > 20) &&

I guess that this simply restores the previous behavior on the
platforms in question.

Now, the reason why it may matter is because
CPUIDLE_DRIVER_STATE_START is 0 and so data->last_state_idx may end up
as -1 on them.  So I think this piece of code only ever makes sense if
CPUIDLE_DRIVER_STATE_START is 1.

>             !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
>                 dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
>                 data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
> --

Thanks,
Rafael

Re: [PATCH 2/3] cpuidle,menu: use interactivity_req to disable polling

[Rafael J. Wysocki]

On Wednesday, January 13, 2016 10:58:13 PM Rafael J. Wysocki wrote:
> Hi,
> 
> On Wed, Jan 13, 2016 at 6:27 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
> > Hi Rik,

[cut]

> >
> > diff --git i/drivers/cpuidle/governors/menu.c w/drivers/cpuidle/governors/menu.c
> > index 7b0971d97cc3..7c7f4059705a 100644
> > --- i/drivers/cpuidle/governors/menu.c
> > +++ w/drivers/cpuidle/governors/menu.c
> > @@ -330,7 +330,8 @@ static int menu_select(struct cpuidle_driver *drv,
> > struct cpuidle_device *dev)
> >          * We want to default to C1 (hlt), not to busy polling
> >          * unless the timer is happening really really soon.
> >          */
> > -       if (interactivity_req > 20 &&
> > +       if (((interactivity_req && interactivity_req > 20) ||
> 
> Well, if interactivity_req > 20, then it also is different from 0, so
> the first check should not be necessary here.
> 
> > +               data->next_timer_us > 20) &&
> 
> I guess that this simply restores the previous behavior on the
> platforms in question.
> 
> Now, the reason why it may matter is because
> CPUIDLE_DRIVER_STATE_START is 0 and so data->last_state_idx may end up
> as -1 on them.  So I think this piece of code only ever makes sense if
> CPUIDLE_DRIVER_STATE_START is 1.
> 
> >             !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
> >                 dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
> >                 data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
> > --

So I'm wondering if the appended patch helps by any chance?

Rafael

---
 drivers/cpuidle/governors/menu.c |   21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)

Index: linux-pm/drivers/cpuidle/governors/menu.c
===================================================================
--- linux-pm.orig/drivers/cpuidle/governors/menu.c
+++ linux-pm/drivers/cpuidle/governors/menu.c
@@ -294,8 +294,6 @@ static int menu_select(struct cpuidle_dr
 		data->needs_update = 0;
 	}
 
-	data->last_state_idx = CPUIDLE_DRIVER_STATE_START - 1;
-
 	/* Special case when user has set very strict latency requirement */
 	if (unlikely(latency_req == 0))
 		return 0;
@@ -326,14 +324,19 @@ static int menu_select(struct cpuidle_dr
 	if (latency_req > interactivity_req)
 		latency_req = interactivity_req;
 
-	/*
-	 * We want to default to C1 (hlt), not to busy polling
-	 * unless the timer is happening really really soon.
-	 */
-	if (interactivity_req > 20 &&
-	    !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
-		dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
+	if (CPUIDLE_DRIVER_STATE_START > 0) {
+		data->last_state_idx = CPUIDLE_DRIVER_STATE_START - 1;
+		/*
+		 * We want to default to C1 (hlt), not to busy polling
+		 * unless the timer is happening really really soon.
+		 */
+		if (interactivity_req > 20 &&
+		    !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
+			dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
+			data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
+	} else {
 		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
+	}
 
 	/*
 	 * Find the idle state with the lowest power while satisfying

Re: [PATCH 2/3] cpuidle,menu: use interactivity_req to disable polling

[Sudeep Holla]

On 13/01/16 22:07, Rafael J. Wysocki wrote:
> On Wednesday, January 13, 2016 10:58:13 PM Rafael J. Wysocki wrote:

[..]

>
> So I'm wondering if the appended patch helps by any chance?
>

Yes it does fix. As you mentioned earlier, CPUIDLE_DRIVER_STATE_START is
0 on ARM platforms and hence data->last_state_idx ended up as -1.

You can add by Tested-by when you push this change. Thanks for the quick
fix.

-- 
Regards,
Sudeep

Re: [PATCH 2/3] cpuidle,menu: use interactivity_req to disable polling

[Rafael J. Wysocki]

On Thursday, January 14, 2016 10:40:44 AM Sudeep Holla wrote:
> 
> On 13/01/16 22:07, Rafael J. Wysocki wrote:
> > On Wednesday, January 13, 2016 10:58:13 PM Rafael J. Wysocki wrote:
> 
> [..]
> 
> >
> > So I'm wondering if the appended patch helps by any chance?
> >
> 
> Yes it does fix. As you mentioned earlier, CPUIDLE_DRIVER_STATE_START is
> 0 on ARM platforms and hence data->last_state_idx ended up as -1.
> 
> You can add by Tested-by when you push this change. Thanks for the quick
> fix.

No problem, thanks for testing!

Below it goes again with a changelog and all.

---
From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Subject: [PATCH] cpuidle: menu: Fix menu_select() for CPUIDLE_DRIVER_STATE_START == 0

Commit a9ceb78bc75c (cpuidle,menu: use interactivity_req to disable
polling) exposed a bug in menu_select() causing it to return -1
on systems with CPUIDLE_DRIVER_STATE_START equal to zero, although
it should have returned 0.  As a result, idle states are not entered
by CPUs on those systems.

Namely, on the systems in question data->last_state_idx is initially
equal to -1 and the above commit modified the condition that would
have caused it to be changed to 0 to be less likely to trigger which
exposed the problem.  However, setting data->last_state_idx initially
to -1 doesn't make sense at all and on the affected systems it should
always be set to CPUIDLE_DRIVER_STATE_START (ie. 0) unconditionally,
so make that happen.

Fixes: a9ceb78bc75c (cpuidle,menu: use interactivity_req to disable polling)
Reported-and-tested-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/cpuidle/governors/menu.c |   21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)

Index: linux-pm/drivers/cpuidle/governors/menu.c
===================================================================
--- linux-pm.orig/drivers/cpuidle/governors/menu.c
+++ linux-pm/drivers/cpuidle/governors/menu.c
@@ -294,8 +294,6 @@ static int menu_select(struct cpuidle_dr
 		data->needs_update = 0;
 	}
 
-	data->last_state_idx = CPUIDLE_DRIVER_STATE_START - 1;
-
 	/* Special case when user has set very strict latency requirement */
 	if (unlikely(latency_req == 0))
 		return 0;
@@ -326,14 +324,19 @@ static int menu_select(struct cpuidle_dr
 	if (latency_req > interactivity_req)
 		latency_req = interactivity_req;
 
-	/*
-	 * We want to default to C1 (hlt), not to busy polling
-	 * unless the timer is happening really really soon.
-	 */
-	if (interactivity_req > 20 &&
-	    !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
-		dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
+	if (CPUIDLE_DRIVER_STATE_START > 0) {
+		data->last_state_idx = CPUIDLE_DRIVER_STATE_START - 1;
+		/*
+		 * We want to default to C1 (hlt), not to busy polling
+		 * unless the timer is happening really really soon.
+		 */
+		if (interactivity_req > 20 &&
+		    !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
+			dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
+			data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
+	} else {
 		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
+	}
 
 	/*
 	 * Find the idle state with the lowest power while satisfying

Re: [PATCH 2/3] cpuidle,menu: use interactivity_req to disable polling

[Sudeep Holla]

On 13/01/16 21:58, Rafael J. Wysocki wrote:
> Hi,
>
> On Wed, Jan 13, 2016 at 6:27 PM, Sudeep Holla <sudeep.holla@arm.com> wrote:
>> Hi Rik,
>>
>> This change break idle on ARM64(may be on other ARM?) platform.
>> Sorry for reporting late, but missed to check cpuidle in -next.
>
> OK, so first of all, how exactly is idle broken on those systems?
>

Sorry for the hasty bug report.

> Do they crash or does something different happen?  If something
> different happens, then what's that?
>

No they just don't enter any idle states(as if CPUIdle was disabled)

[...]

>>
>> diff --git i/drivers/cpuidle/governors/menu.c w/drivers/cpuidle/governors/menu.c
>> index 7b0971d97cc3..7c7f4059705a 100644
>> --- i/drivers/cpuidle/governors/menu.c
>> +++ w/drivers/cpuidle/governors/menu.c
>> @@ -330,7 +330,8 @@ static int menu_select(struct cpuidle_driver *drv,
>> struct cpuidle_device *dev)
>>           * We want to default to C1 (hlt), not to busy polling
>>           * unless the timer is happening really really soon.
>>           */
>> -       if (interactivity_req > 20 &&
>> +       if (((interactivity_req && interactivity_req > 20) ||
>
> Well, if interactivity_req > 20, then it also is different from 0, so
> the first check should not be necessary here.
>

True, I just wanted to avoid using interactivity_req when 0, it was just
a quick hack.

>> +               data->next_timer_us > 20) &&
>
> I guess that this simply restores the previous behavior on the
> platforms in question.
>

Yes.

> Now, the reason why it may matter is because
> CPUIDLE_DRIVER_STATE_START is 0 and so data->last_state_idx may end up
> as -1 on them.  So I think this piece of code only ever makes sense if
> CPUIDLE_DRIVER_STATE_START is 1.
>

That's correct.

-- 
Regards,
Sudeep

[PATCH 3/3] cpuidle,menu: smooth out measured_us calculation

[riel]

From: Rik van Riel <riel@redhat.com>

The cpuidle state tables contain the maximum exit latency for each
cpuidle state. On x86, that is the exit latency for when the entire
package goes into that same idle state.

However, a lot of the time we only go into the core idle state,
not the package idle state. This means we see a much smaller exit
latency.

We have no way to detect whether we went into the core or package
idle state while idle, and that is ok.

However, the current menu_update logic does have the potential to
trip up the repeating pattern detection in get_typical_interval.
If the system is experiencing an exit latency near the idle state's
exit latency, some of the samples will have exit_us subtracted,
while others will not. This turns a repeating pattern into mush,
potentially breaking get_typical_interval.

Furthermore, for smaller sleep intervals, we know the chance that
all the cores in the package went to the same idle state are fairly
small. Dividing the measured_us by two, instead of subtracting the
full exit latency when hitting a small measured_us, will reduce the
error.

Signed-off-by: Rik van Riel <riel@redhat.com>
---
 drivers/cpuidle/governors/menu.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index b1a55731f921..7b0971d97cc3 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -404,8 +404,10 @@ static void menu_update(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	measured_us = cpuidle_get_last_residency(dev);
 
 	/* Deduct exit latency */
-	if (measured_us > target->exit_latency)
+	if (measured_us > 2 * target->exit_latency)
 		measured_us -= target->exit_latency;
+	else
+		measured_us /= 2;
 
 	/* Make sure our coefficients do not exceed unity */
 	if (measured_us > data->next_timer_us)
-- 
2.1.0

Re: [PATCH 3/3] cpuidle,menu: smooth out measured_us calculation

[Arjan van de Ven]

On 11/3/2015 2:34 PM, riel@redhat.com wrote:

> Furthermore, for smaller sleep intervals, we know the chance that
> all the cores in the package went to the same idle state are fairly
> small. Dividing the measured_us by two, instead of subtracting the
> full exit latency when hitting a small measured_us, will reduce the
> error.

there is no perfect answer for this issue; but at least this makes the situation
a lot better, so

Acked-by: Arjan van de Ven <arjan@linux.intel.com>

Re: [PATCH 0/3] cpuidle: small improvements & fixes for menu governor (resend)

[Rafael J. Wysocki]

On Tuesday, November 03, 2015 05:34:16 PM riel@redhat.com wrote:
> While working on a paravirt cpuidle driver for KVM guests, I
> noticed a number of small logic errors in the menu governor
> code.
> 
> These patches should get rid of some artifacts that can break
> the logic in the menu governor under certain corner cases, and
> make idle state selection work better on CPUs with long C1 exit
> latencies.
> 
> I have not seen any adverse effects with them in my (quick)
> tests. As expected, they do not seem to do much on systems with
> many power states and very low C1 exit latencies and target residencies. 

Thanks!

The patches look good to me.

I might apply [1-2/3] right away, but I'm a bit hesitant about the [3/3] (I'd
like it to spend some time in linux-next before it goes to Linus).  Also, we've
lived without these changes for quite some time and I don't want to stretch the
process too much, so I'll queue them up for v4.5 if that's not a problem.

Thanks,
Rafael

Re: [PATCH 0/3] cpuidle: small improvements & fixes for menu governor (resend)

[Rik van Riel]

On 11/05/2015 05:34 PM, Rafael J. Wysocki wrote:
> On Tuesday, November 03, 2015 05:34:16 PM riel@redhat.com wrote:
>> While working on a paravirt cpuidle driver for KVM guests, I
>> noticed a number of small logic errors in the menu governor
>> code.
>>
>> These patches should get rid of some artifacts that can break
>> the logic in the menu governor under certain corner cases, and
>> make idle state selection work better on CPUs with long C1 exit
>> latencies.
>>
>> I have not seen any adverse effects with them in my (quick)
>> tests. As expected, they do not seem to do much on systems with
>> many power states and very low C1 exit latencies and target residencies. 
> 
> Thanks!
> 
> The patches look good to me.
> 
> I might apply [1-2/3] right away, but I'm a bit hesitant about the [3/3] (I'd
> like it to spend some time in linux-next before it goes to Linus).  Also, we've
> lived without these changes for quite some time and I don't want to stretch the
> process too much, so I'll queue them up for v4.5 if that's not a problem.

Not a problem at all. I am all for taking these changes carefully,
and seeing what happens.

I did some basic testing with it, but the permutations of what
can happen with cpuidle management are just too many to predict
in advance everything that could happen.

-- 
All rights reversed

[PATCH 0/3] cpuidle: small improvements & fixes for menu governor

[riel]

While working on a paravirt cpuidle driver for KVM guests, I
noticed a number of small logic errors in the menu governor
code.

These patches should get rid of some artifacts that can break
the logic in the menu governor under certain corner cases, and
make idle state selection work better on CPUs with long C1 exit
latencies.

I have not seen any adverse effects with them in my (quick)
tests. As expected, they do not seem to do much on systems with
many power states and very low C1 exit latencies and target residencies.

[PATCH 1/3] cpuidle,x86: increase forced cut-off for polling to 20us

[riel]

From: Rik van Riel <riel@redhat.com>

The cpuidle menu governor has a forced cut-off for polling at 5us,
in order to deal with firmware that gives the OS bad information
on cpuidle states, leading to the system spending way too much time
in polling.

However, at least one x86 CPU family (Atom) has chips that have
a 20us break-even point for C1. Forcing the polling cut-off to
less than that wastes performance and power.

Increase the polling cut-off to 20us.

Systems with a lower C1 latency will be found in the states table by
the menu governor, which will pick those states as appropriate.

Signed-off-by: Rik van Riel <riel@redhat.com>
---
 drivers/cpuidle/governors/menu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
index 22e4463d1787..ecc242a586c9 100644
--- a/drivers/cpuidle/governors/menu.c
+++ b/drivers/cpuidle/governors/menu.c
@@ -330,7 +330,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 	 * We want to default to C1 (hlt), not to busy polling
 	 * unless the timer is happening really really soon.
 	 */
-	if (data->next_timer_us > 5 &&
+	if (data->next_timer_us > 20 &&
 	    !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
 		dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
 		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
-- 
2.1.0

Re: [PATCH 1/3] cpuidle,x86: increase forced cut-off for polling to 20us

[Daniel Lezcano]

On 10/28/2015 11:46 PM, riel@redhat.com wrote:
> From: Rik van Riel <riel@redhat.com>
>
> The cpuidle menu governor has a forced cut-off for polling at 5us,
> in order to deal with firmware that gives the OS bad information
> on cpuidle states, leading to the system spending way too much time
> in polling.

May be I am misunderstanding your explanation but it is not how I read 
the code.

The default idle state is C1 (hlt) if no other states suits the 
constraint. If a timer is happening really soon, then set the default 
idle state to POLL if no other idle state suits the constraint.

That applies only on x86.

This is not related to break-even but exit latency.

IMO, we should just drop this 5us and the POLL state selection in the 
menu governor as we have since a while hyper fast C1 exit. Except a few 
embedded processors where polling is not adequate.

Furthermore, the number of times the poll state is selected vs the other 
states is negligible.

I already raised this point but Len is opposed to the removal.

Len ? Can you elaborate why you are opposed to this removal ?

> However, at least one x86 CPU family (Atom) has chips that have
> a 20us break-even point for C1. Forcing the polling cut-off to
> less than that wastes performance and power.
>
> Increase the polling cut-off to 20us.
>
> Systems with a lower C1 latency will be found in the states table by
> the menu governor, which will pick those states as appropriate.

With this change, I believe the poll state will be selected more often 
(not too much certainly), hence implying a bigger energy consumption.

And finally it may improve the situation for specific processor but 
deteriorate for other processors.

Does anyone have the rational behind this '5' number ? why not '3' or '7' ?

> Signed-off-by: Rik van Riel <riel@redhat.com>
> ---
>   drivers/cpuidle/governors/menu.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c
> index 22e4463d1787..ecc242a586c9 100644
> --- a/drivers/cpuidle/governors/menu.c
> +++ b/drivers/cpuidle/governors/menu.c
> @@ -330,7 +330,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
>   	 * We want to default to C1 (hlt), not to busy polling
>   	 * unless the timer is happening really really soon.
>   	 */
> -	if (data->next_timer_us > 5 &&
> +	if (data->next_timer_us > 20 &&
>   	    !drv->states[CPUIDLE_DRIVER_STATE_START].disabled &&
>   		dev->states_usage[CPUIDLE_DRIVER_STATE_START].disable == 0)
>   		data->last_state_idx = CPUIDLE_DRIVER_STATE_START;
>


-- 
  <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

Re: [PATCH 1/3] cpuidle,x86: increase forced cut-off for polling to 20us

[Rik van Riel]

On 10/29/2015 06:17 AM, Daniel Lezcano wrote:
> On 10/28/2015 11:46 PM, riel@redhat.com wrote:
>> From: Rik van Riel <riel@redhat.com>
>>
>> The cpuidle menu governor has a forced cut-off for polling at 5us,
>> in order to deal with firmware that gives the OS bad information
>> on cpuidle states, leading to the system spending way too much time
>> in polling.
> 
> May be I am misunderstanding your explanation but it is not how I read
> the code.
> 
> The default idle state is C1 (hlt) if no other states suits the
> constraint. If a timer is happening really soon, then set the default
> idle state to POLL if no other idle state suits the constraint.
> 
> That applies only on x86.

With the current code, the default idle state is C1 (hlt) even if
C1 does not suit the constraint.

> This is not related to break-even but exit latency.

Why would we not care about break-even for C1?

On systems where going into C1 for too-short periods wastes
power, why would we waste the power when we expect a very
short sleep?

> IMO, we should just drop this 5us and the POLL state selection in the
> menu governor as we have since a while hyper fast C1 exit. Except a few
> embedded processors where polling is not adequate.

We have hyper fast C1 exit on Nehalem and newer high performance
chips. On those chips, we will pick C1 (or deeper) when we have
an expected sleep time of just a few microseconds.

However, on Atom, and for the paravirt cpuidle driver I am
working on, C1 exit latency and target residence are higher
than the cut-off hardcoded in the menu governor.

> Furthermore, the number of times the poll state is selected vs the other
> states is negligible.

And it will continue to be with this patch, on CPUs with
hyper fast C1 exit.

Which makes me confused about what your are objecting to,
since the system should continue to be have the way you want,
with the patch applied.

-- 
All rights reversed

Re: [PATCH 1/3] cpuidle,x86: increase forced cut-off for polling to 20us

[Daniel Lezcano]

On 10/29/2015 12:54 PM, Rik van Riel wrote:
> On 10/29/2015 06:17 AM, Daniel Lezcano wrote:
>> On 10/28/2015 11:46 PM, riel@redhat.com wrote:
>>> From: Rik van Riel <riel@redhat.com>
>>>
>>> The cpuidle menu governor has a forced cut-off for polling at 5us,
>>> in order to deal with firmware that gives the OS bad information
>>> on cpuidle states, leading to the system spending way too much time
>>> in polling.
>>
>> May be I am misunderstanding your explanation but it is not how I read
>> the code.
>>
>> The default idle state is C1 (hlt) if no other states suits the
>> constraint. If a timer is happening really soon, then set the default
>> idle state to POLL if no other idle state suits the constraint.
>>
>> That applies only on x86.
>
> With the current code, the default idle state is C1 (hlt) even if
> C1 does not suit the constraint.
>
>> This is not related to break-even but exit latency.
>
> Why would we not care about break-even for C1?
>
> On systems where going into C1 for too-short periods wastes
> power, why would we waste the power when we expect a very
> short sleep?
>
>> IMO, we should just drop this 5us and the POLL state selection in the
>> menu governor as we have since a while hyper fast C1 exit. Except a few
>> embedded processors where polling is not adequate.
>
> We have hyper fast C1 exit on Nehalem and newer high performance
> chips. On those chips, we will pick C1 (or deeper) when we have
> an expected sleep time of just a few microseconds.
>
> However, on Atom, and for the paravirt cpuidle driver I am
> working on, C1 exit latency and target residence are higher
> than the cut-off hardcoded in the menu governor.
>
>> Furthermore, the number of times the poll state is selected vs the other
>> states is negligible.
>
> And it will continue to be with this patch, on CPUs with
> hyper fast C1 exit.
>
> Which makes me confused about what your are objecting to,
> since the system should continue to be have the way you want,
> with the patch applied.

Ok, I don't object the correctness of your patch but the reasoning 
behind this small optimization which bring us a lot of mess in the 
cpuidle code.

As you are touching this part of the code, I take the opportunity to 
raise a discussion about it.

 From my POV, the poll state is *not* an idle state. It is like a 
vehicle burnout [1].

But it is inserted into the idle state tables using a trick with a macro 
CPUIDLE_DRIVER_STATE_START which already led us to some bugs.

So instead of falling back into the poll state under certain 
circumstances, I propose we extract this state from the idle state table 
and we let the menu governor to fail choosing a state (or not).

 From the caller, we decide what to do (poll or C1) if the idle state 
selection fails or we choose to poll *before* like what we already have 
in kernel/sched/idle.c:

in the idle loop:

if (cpu_idle_force_poll || tick_check_broadcast_expired())
	cpu_idle_poll();
else
	cpuidle_idle_call();

By this way, we:

1) factor out the idle state selection with the find_deepest_idle_state
2) remove the CPUIDLE_DRIVER_STATE_START macro
3) concentrate the optimization logic outside of a governor which will 
benefit to all architectures

Does it make sense ?

   -- Daniel

[1] https://en.wikipedia.org/wiki/Burnout_%28vehicle%29


-- 
  <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog