[RFC PATCH] thermal: power_allocator: Ignore cutoff when integral is less than zero

[RFC PATCH] thermal: power_allocator: Ignore cutoff when integral is less than zero

[Xuewen Yan]

From: Jeson Gao <jeson.gao@unisoc.com>

The cutoff means threshold below which the error is no longer accumulated.
However, in some scenarios, this may cause performance degradation.

For example:
the control-temp is 85, the cutoff is 0 or other small value:

If the current temperature frequently exceeds the set temperature,
the negative integral will continuously accumulate.
Over an extended period, this will result in a significantly
large negative integral value, the positive integral can’t build up
because of the cutoff. This makes the power_range very low,
even if the temperature is already under the control target.

So, if the err_integral is negative, ignore the cutoff to force
add the positive integral.

Co-developed-by: Di Shen <di.shen@unisoc.com>
Signed-off-by: Di Shen <di.shen@unisoc.com>
Signed-off-by: Jeson Gao <jeson.gao@unisoc.com>
Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
---
 drivers/thermal/gov_power_allocator.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/thermal/gov_power_allocator.c b/drivers/thermal/gov_power_allocator.c
index 0d9f636c80f4..404ae1d75612 100644
--- a/drivers/thermal/gov_power_allocator.c
+++ b/drivers/thermal/gov_power_allocator.c
@@ -263,7 +263,8 @@ static u32 pid_controller(struct thermal_zone_device *tz,
         */
        i = mul_frac(tz->tzp->k_i, params->err_integral);

-       if (err < int_to_frac(tz->tzp->integral_cutoff)) {
+       if (err < int_to_frac(tz->tzp->integral_cutoff) ||
+           (err > 0 && params->err_integral < 0)) {
                s64 i_next = i + mul_frac(tz->tzp->k_i, err);

                if (abs(i_next) < max_power_frac) {
--
2.25.1

________________________________
 This email (including its attachments) is intended only for the person or entity to which it is addressed and may contain information that is privileged, confidential or otherwise protected from disclosure. Unauthorized use, dissemination, distribution or copying of this email or the information herein or taking any action in reliance on the contents of this email or the information herein, by anyone other than the intended recipient, or an employee or agent responsible for delivering the message to the intended recipient, is strictly prohibited. If you are not the intended recipient, please do not read, copy, use or disclose any part of this e-mail to others. Please notify the sender immediately and permanently delete this e-mail and any attachments if you received it in error. Internet communications cannot be guaranteed to be timely, secure, error-free or virus-free. The sender does not accept liability for any errors or omissions.
本邮件及其附件具有保密性质，受法律保护不得泄露，仅发送给本邮件所指特定收件人。严禁非经授权使用、宣传、发布或复制本邮件或其内容。若非该特定收件人，请勿阅读、复制、 使用或披露本邮件的任何内容。若误收本邮件，请从系统中永久性删除本邮件及所有附件，并以回复邮件的方式即刻告知发件人。无法保证互联网通信及时、安全、无误或防毒。发件人对任何错漏均不承担责任。

Re: [RFC PATCH] thermal: power_allocator: Ignore cutoff when integral is less than zero

[Xuewen Yan]

On Mon, Jan 12, 2026 at 1:31 PM Xuewen Yan <xuewen.yan@unisoc.com> wrote:
>
> From: Jeson Gao <jeson.gao@unisoc.com>
>
> The cutoff means threshold below which the error is no longer accumulated.
> However, in some scenarios, this may cause performance degradation.
>
> For example:
> the control-temp is 85, the cutoff is 0 or other small value:
>
> If the current temperature frequently exceeds the set temperature,
> the negative integral will continuously accumulate.
> Over an extended period, this will result in a significantly
> large negative integral value, the positive integral can’t build up
> because of the cutoff. This makes the power_range very low,
> even if the temperature is already under the control target.
>
> So, if the err_integral is negative, ignore the cutoff to force
> add the positive integral.
>
> Co-developed-by: Di Shen <di.shen@unisoc.com>
> Signed-off-by: Di Shen <di.shen@unisoc.com>
> Signed-off-by: Jeson Gao <jeson.gao@unisoc.com>
> Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
> ---
>  drivers/thermal/gov_power_allocator.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/thermal/gov_power_allocator.c b/drivers/thermal/gov_power_allocator.c
> index 0d9f636c80f4..404ae1d75612 100644
> --- a/drivers/thermal/gov_power_allocator.c
> +++ b/drivers/thermal/gov_power_allocator.c
> @@ -263,7 +263,8 @@ static u32 pid_controller(struct thermal_zone_device *tz,
>          */
>         i = mul_frac(tz->tzp->k_i, params->err_integral);
>
> -       if (err < int_to_frac(tz->tzp->integral_cutoff)) {
> +       if (err < int_to_frac(tz->tzp->integral_cutoff) ||
> +           (err > 0 && params->err_integral < 0)) {
>                 s64 i_next = i + mul_frac(tz->tzp->k_i, err);
>
>                 if (abs(i_next) < max_power_frac) {
> --
> 2.25.1
>
> ________________________________
>  This email (including its attachments) is intended only for the person or entity to which it is addressed and may contain information that is privileged, confidential or otherwise protected from disclosure. Unauthorized use, dissemination, distribution or copying of this email or the information herein or taking any action in reliance on the contents of this email or the information herein, by anyone other than the intended recipient, or an employee or agent responsible for delivering the message to the intended recipient, is strictly prohibited. If you are not the intended recipient, please do not read, copy, use or disclose any part of this e-mail to others. Please notify the sender immediately and permanently delete this e-mail and any attachments if you received it in error. Internet communications cannot be guaranteed to be timely, secure, error-free or virus-free. The sender does not accept liability for any errors or omissions.
> 本邮件及其附件具有保密性质，受法律保护不得泄露，仅发送给本邮件所指特定收件人。严禁非经授权使用、宣传、发布或复制本邮件或其内容。若非该特定收件人，请勿阅读、复制、 使用或披露本邮件的任何内容。若误收本邮件，请从系统中永久性删除本邮件及所有附件，并以回复邮件的方式即刻告知发件人。无法保证互联网通信及时、安全、无误或防毒。发件人对任何错漏均不承担责任。


My apologies, I missed the disclaimer in the email. I will coordinate
with our IT department to have it removed and will re-send the
message.

BR

Re: [RFC PATCH] thermal: power_allocator: Ignore cutoff when integral is less than zero

[Xuewen Yan]

On Mon, Feb 9, 2026 at 8:35 PM Lukasz Luba <lukasz.luba@arm.com> wrote:
>
> Hi Xuewen,
>
> On 1/26/26 03:24, Xuewen Yan wrote:
> > Hi lukasz,
> >
> > Do you have any comments on this patch?
>
> I missed that patch, my apologies.
>
> >
> > Thanks!
> >
> > On Mon, Jan 19, 2026 at 10:00 AM Xuewen Yan <xuewen.yan@unisoc.com> wrote:
> >>
> >> From: Jeson Gao <jeson.gao@unisoc.com>
> >>
> >> The cutoff means threshold below which the error is no longer accumulated.
> >> However, in some scenarios, this may cause performance degradation.
> >>
> >> For example:
> >> the control-temp is 85, the cutoff is 0 or other small value:
> >>
> >> If the current temperature frequently exceeds the set temperature,
> >> the negative integral will continuously accumulate.
> >> Over an extended period, this will result in a significantly
> >> large negative integral value, the positive integral can’t build up
> >> because of the cutoff. This makes the power_range very low,
> >> even if the temperature is already under the control target.
> >>
> >> So, if the err_integral is negative, ignore the cutoff to force
> >> add the positive integral.
>
> The accumulated Integral in the PID works like that, so no surprise
> here (the math is correct). The Integral part tires to 'compensate'
> the previous 'mistakes', e.g. slow down the rapid change which
> has been seen and caused the overshoot.
>
> >>
> >> Co-developed-by: Di Shen <di.shen@unisoc.com>
> >> Signed-off-by: Di Shen <di.shen@unisoc.com>
> >> Signed-off-by: Jeson Gao <jeson.gao@unisoc.com>
> >> Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
> >> ---
> >>   drivers/thermal/gov_power_allocator.c | 3 ++-
> >>   1 file changed, 2 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/drivers/thermal/gov_power_allocator.c b/drivers/thermal/gov_power_allocator.c
> >> index 0d9f636c80f4..404ae1d75612 100644
> >> --- a/drivers/thermal/gov_power_allocator.c
> >> +++ b/drivers/thermal/gov_power_allocator.c
> >> @@ -263,7 +263,8 @@ static u32 pid_controller(struct thermal_zone_device *tz,
> >>           */
> >>          i = mul_frac(tz->tzp->k_i, params->err_integral);
> >>
> >> -       if (err < int_to_frac(tz->tzp->integral_cutoff)) {
> >> +       if (err < int_to_frac(tz->tzp->integral_cutoff) ||
> >> +           (err > 0 && params->err_integral < 0)) {
> >>                  s64 i_next = i + mul_frac(tz->tzp->k_i, err);
> >>
> >>                  if (abs(i_next) < max_power_frac) {
> >> --
> >> 2.25.1
> >>
>
> Although, I might probably know where you're coming from and what you
> are trying to achieve (what is your platform's situation). Let me do
> some simulations based on what you've described. I will look into
> those signals in those conditions.
>
> At the moment, I'm a bit afraid about your proposed simple change
> and possible impact to other platforms.
> Let me do my research and come back.

Thank you very much for your reply! We would also be happy to
participate in testing your follow-up research. Thanks again!

BR

---
xuewen

>
> Thank you for reporting this issue!
>
> Regards,
> Lukasz

Re: [RFC PATCH] thermal: power_allocator: Ignore cutoff when integral is less than zero

[Lukasz Luba]

Hi Xuewen,

On 1/26/26 03:24, Xuewen Yan wrote:
> Hi lukasz,
> 
> Do you have any comments on this patch?

I missed that patch, my apologies.

> 
> Thanks!
> 
> On Mon, Jan 19, 2026 at 10:00 AM Xuewen Yan <xuewen.yan@unisoc.com> wrote:
>>
>> From: Jeson Gao <jeson.gao@unisoc.com>
>>
>> The cutoff means threshold below which the error is no longer accumulated.
>> However, in some scenarios, this may cause performance degradation.
>>
>> For example:
>> the control-temp is 85, the cutoff is 0 or other small value:
>>
>> If the current temperature frequently exceeds the set temperature,
>> the negative integral will continuously accumulate.
>> Over an extended period, this will result in a significantly
>> large negative integral value, the positive integral can’t build up
>> because of the cutoff. This makes the power_range very low,
>> even if the temperature is already under the control target.
>>
>> So, if the err_integral is negative, ignore the cutoff to force
>> add the positive integral.

The accumulated Integral in the PID works like that, so no surprise
here (the math is correct). The Integral part tires to 'compensate'
the previous 'mistakes', e.g. slow down the rapid change which
has been seen and caused the overshoot.

>>
>> Co-developed-by: Di Shen <di.shen@unisoc.com>
>> Signed-off-by: Di Shen <di.shen@unisoc.com>
>> Signed-off-by: Jeson Gao <jeson.gao@unisoc.com>
>> Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
>> ---
>>   drivers/thermal/gov_power_allocator.c | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/thermal/gov_power_allocator.c b/drivers/thermal/gov_power_allocator.c
>> index 0d9f636c80f4..404ae1d75612 100644
>> --- a/drivers/thermal/gov_power_allocator.c
>> +++ b/drivers/thermal/gov_power_allocator.c
>> @@ -263,7 +263,8 @@ static u32 pid_controller(struct thermal_zone_device *tz,
>>           */
>>          i = mul_frac(tz->tzp->k_i, params->err_integral);
>>
>> -       if (err < int_to_frac(tz->tzp->integral_cutoff)) {
>> +       if (err < int_to_frac(tz->tzp->integral_cutoff) ||
>> +           (err > 0 && params->err_integral < 0)) {
>>                  s64 i_next = i + mul_frac(tz->tzp->k_i, err);
>>
>>                  if (abs(i_next) < max_power_frac) {
>> --
>> 2.25.1
>>

Although, I might probably know where you're coming from and what you 
are trying to achieve (what is your platform's situation). Let me do
some simulations based on what you've described. I will look into
those signals in those conditions.

At the moment, I'm a bit afraid about your proposed simple change
and possible impact to other platforms.
Let me do my research and come back.

Thank you for reporting this issue!

Regards,
Lukasz

Re: [RFC PATCH] thermal: power_allocator: Ignore cutoff when integral is less than zero

[Xuewen Yan]

Hi lukasz,

Do you have any comments on this patch?

Thanks!

On Mon, Jan 19, 2026 at 10:00 AM Xuewen Yan <xuewen.yan@unisoc.com> wrote:
>
> From: Jeson Gao <jeson.gao@unisoc.com>
>
> The cutoff means threshold below which the error is no longer accumulated.
> However, in some scenarios, this may cause performance degradation.
>
> For example:
> the control-temp is 85, the cutoff is 0 or other small value:
>
> If the current temperature frequently exceeds the set temperature,
> the negative integral will continuously accumulate.
> Over an extended period, this will result in a significantly
> large negative integral value, the positive integral can’t build up
> because of the cutoff. This makes the power_range very low,
> even if the temperature is already under the control target.
>
> So, if the err_integral is negative, ignore the cutoff to force
> add the positive integral.
>
> Co-developed-by: Di Shen <di.shen@unisoc.com>
> Signed-off-by: Di Shen <di.shen@unisoc.com>
> Signed-off-by: Jeson Gao <jeson.gao@unisoc.com>
> Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
> ---
>  drivers/thermal/gov_power_allocator.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/thermal/gov_power_allocator.c b/drivers/thermal/gov_power_allocator.c
> index 0d9f636c80f4..404ae1d75612 100644
> --- a/drivers/thermal/gov_power_allocator.c
> +++ b/drivers/thermal/gov_power_allocator.c
> @@ -263,7 +263,8 @@ static u32 pid_controller(struct thermal_zone_device *tz,
>          */
>         i = mul_frac(tz->tzp->k_i, params->err_integral);
>
> -       if (err < int_to_frac(tz->tzp->integral_cutoff)) {
> +       if (err < int_to_frac(tz->tzp->integral_cutoff) ||
> +           (err > 0 && params->err_integral < 0)) {
>                 s64 i_next = i + mul_frac(tz->tzp->k_i, err);
>
>                 if (abs(i_next) < max_power_frac) {
> --
> 2.25.1
>

[RFC PATCH] thermal: power_allocator: Ignore cutoff when integral is less than zero

[Xuewen Yan]

From: Jeson Gao <jeson.gao@unisoc.com>

The cutoff means threshold below which the error is no longer accumulated.
However, in some scenarios, this may cause performance degradation.

For example:
the control-temp is 85, the cutoff is 0 or other small value:

If the current temperature frequently exceeds the set temperature,
the negative integral will continuously accumulate.
Over an extended period, this will result in a significantly
large negative integral value, the positive integral can’t build up
because of the cutoff. This makes the power_range very low,
even if the temperature is already under the control target.

So, if the err_integral is negative, ignore the cutoff to force
add the positive integral.

Co-developed-by: Di Shen <di.shen@unisoc.com>
Signed-off-by: Di Shen <di.shen@unisoc.com>
Signed-off-by: Jeson Gao <jeson.gao@unisoc.com>
Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
---
 drivers/thermal/gov_power_allocator.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/thermal/gov_power_allocator.c b/drivers/thermal/gov_power_allocator.c
index 0d9f636c80f4..404ae1d75612 100644
--- a/drivers/thermal/gov_power_allocator.c
+++ b/drivers/thermal/gov_power_allocator.c
@@ -263,7 +263,8 @@ static u32 pid_controller(struct thermal_zone_device *tz,
 	 */
 	i = mul_frac(tz->tzp->k_i, params->err_integral);
 
-	if (err < int_to_frac(tz->tzp->integral_cutoff)) {
+	if (err < int_to_frac(tz->tzp->integral_cutoff) ||
+	    (err > 0 && params->err_integral < 0)) {
 		s64 i_next = i + mul_frac(tz->tzp->k_i, err);
 
 		if (abs(i_next) < max_power_frac) {
-- 
2.25.1