From: Daniel Lezcano <daniel.lezcano@linaro.org>
To: Lukasz Luba <lukasz.luba@arm.com>,
linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org,
dri-devel@lists.freedesktop.org
Cc: rui.zhang@intel.com, amit.kucheria@verdurent.com,
orjan.eide@arm.com, robh@kernel.org,
alyssa.rosenzweig@collabora.com, steven.price@arm.com,
airlied@linux.ie, daniel@ffwll.ch, ionela.voinescu@arm.com
Subject: Re: [PATCH v2 2/5] thermal: devfreq_cooling: get a copy of device status
Date: Thu, 3 Dec 2020 14:09:36 +0100 [thread overview]
Message-ID: <5d4743b9-5b2f-8494-8d10-6a5fd2c0fdfd@linaro.org> (raw)
In-Reply-To: <20201118120358.17150-3-lukasz.luba@arm.com>
On 18/11/2020 13:03, Lukasz Luba wrote:
> Devfreq cooling needs to now the correct status of the device in order
> to operate. Do not rely on Devfreq last_status which might be a stale data
> and get more up-to-date values of the load.
>
> Devfreq framework can change the device status in the background. To
> mitigate this situation make a copy of the status structure and use it
> for internal calculations.
>
> In addition this patch adds normalization function, which also makes sure
> that whatever data comes from the device, it is in a sane range.
>
> Signed-off-by: Lukasz Luba <lukasz.luba@arm.com>
> ---
> drivers/thermal/devfreq_cooling.c | 52 +++++++++++++++++++++++++------
> 1 file changed, 43 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/thermal/devfreq_cooling.c b/drivers/thermal/devfreq_cooling.c
> index 659c0143c9f0..925523694462 100644
> --- a/drivers/thermal/devfreq_cooling.c
> +++ b/drivers/thermal/devfreq_cooling.c
> @@ -227,20 +227,46 @@ static inline unsigned long get_total_power(struct devfreq_cooling_device *dfc,
> voltage);
> }
>
> +static void _normalize_load(struct devfreq_dev_status *status)
> +{
> + /* Make some space if needed */
> + if (status->busy_time > 0xffff) {
> + status->busy_time >>= 10;
> + status->total_time >>= 10;
> + }
> +
> + if (status->busy_time > status->total_time)
> + status->busy_time = status->total_time;
How the condition above is possible?
> + status->busy_time *= 100;
> + status->busy_time /= status->total_time ? : 1;
> +
> + /* Avoid division by 0 */
> + status->busy_time = status->busy_time ? : 1;
> + status->total_time = 100;
Why not base the normalization on 1024? and use an intermediate u64.
For example:
static u32 _normalize_load(struct devfreq_dev_status *status)
{
u64 load = 0;
/* Prevent divison by zero */
if (!status->busy_time)
return 0;
/*
* Assuming status->total_time is always greater or equal
* to status->busy_time, it can not be equal to zero because
* of the test above
*/
load = status->busy_time * 1024;
load /= status->total_time;
/*
* load is always [1..1024[, so it can not be truncated by a
* u64 -> u32 coercive cast
*/
return (u32)load;
}
> +}
>
> static int devfreq_cooling_get_requested_power(struct thermal_cooling_device *cdev,
> u32 *power)
> {
> struct devfreq_cooling_device *dfc = cdev->devdata;
> struct devfreq *df = dfc->devfreq;
> - struct devfreq_dev_status *status = &df->last_status;
> + struct devfreq_dev_status status;
> unsigned long state;
> - unsigned long freq = status->current_frequency;
> + unsigned long freq;
> unsigned long voltage;
> u32 dyn_power = 0;
> u32 static_power = 0;
> int res;
>
> + mutex_lock(&df->lock);
> + res = df->profile->get_dev_status(df->dev.parent, &status);
> + mutex_unlock(&df->lock);
> + if (res)
> + return res;
> +
> + freq = status.current_frequency;
> +
> state = freq_get_state(dfc, freq);
> if (state == THERMAL_CSTATE_INVALID) {
> res = -EAGAIN;
> @@ -268,16 +294,18 @@ static int devfreq_cooling_get_requested_power(struct thermal_cooling_device *cd
> } else {
> dyn_power = dfc->power_table[state];
>
> + _normalize_load(&status);
load = _normalize_load(&status);
> +
> /* Scale dynamic power for utilization */
> - dyn_power *= status->busy_time;
> - dyn_power /= status->total_time;
> + dyn_power *= status.busy_time;
> + dyn_power /= status.total_time;
/*
* May be change dyn_power to a u64 to prevent overflow
* when multiplied by 'load'
*/
dyn_power = (dyn_power * load) / 1024;
> /* Get static power */
> static_power = get_static_power(dfc, freq);
>
> *power = dyn_power + static_power;
> }
>
> - trace_thermal_power_devfreq_get_power(cdev, status, freq, *power);
> + trace_thermal_power_devfreq_get_power(cdev, &status, freq, *power);
>
> return 0;
> fail:
> @@ -309,14 +337,20 @@ static int devfreq_cooling_power2state(struct thermal_cooling_device *cdev,
> {
> struct devfreq_cooling_device *dfc = cdev->devdata;
> struct devfreq *df = dfc->devfreq;
> - struct devfreq_dev_status *status = &df->last_status;
> - unsigned long freq = status->current_frequency;
> + struct devfreq_dev_status status;
> unsigned long busy_time;
> + unsigned long freq;
> s32 dyn_power;
> u32 static_power;
> s32 est_power;
> int i;
>
> + mutex_lock(&df->lock);
> + status = df->last_status;
> + mutex_unlock(&df->lock);
> +
> + freq = status.current_frequency;
> +
> if (dfc->power_ops->get_real_power) {
> /* Scale for resource utilization */
> est_power = power * dfc->res_util;
> @@ -328,8 +362,8 @@ static int devfreq_cooling_power2state(struct thermal_cooling_device *cdev,
> dyn_power = dyn_power > 0 ? dyn_power : 0;
>
> /* Scale dynamic power for utilization */
> - busy_time = status->busy_time ?: 1;
> - est_power = (dyn_power * status->total_time) / busy_time;
> + busy_time = status.busy_time ?: 1;
> + est_power = (dyn_power * status.total_time) / busy_time;
> }
>
> /*
>
--
<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
next prev parent reply other threads:[~2020-12-03 13:10 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-11-18 12:03 [PATCH v2 0/5] Thermal devfreq cooling improvements with Energy Model Lukasz Luba
2020-11-18 12:03 ` [PATCH v2 1/5] thermal: devfreq_cooling: change tracing function and arguments Lukasz Luba
2020-12-02 10:23 ` Ionela Voinescu
2020-11-18 12:03 ` [PATCH v2 2/5] thermal: devfreq_cooling: get a copy of device status Lukasz Luba
2020-12-02 10:23 ` Ionela Voinescu
2020-12-03 13:09 ` Daniel Lezcano [this message]
2020-12-03 15:38 ` Lukasz Luba
2020-12-03 16:09 ` Daniel Lezcano
2020-12-07 12:41 ` Lukasz Luba
2020-12-08 14:20 ` Lukasz Luba
2020-11-18 12:03 ` [PATCH v2 3/5] thermal: devfreq_cooling: add new registration functions with Energy Model Lukasz Luba
2020-12-02 10:24 ` Ionela Voinescu
2020-12-02 11:14 ` Lukasz Luba
2020-12-02 11:49 ` Ionela Voinescu
2020-12-02 11:54 ` Lukasz Luba
2020-12-03 15:40 ` Daniel Lezcano
2020-12-07 9:46 ` Lukasz Luba
2020-11-18 12:03 ` [PATCH v2 4/5] thermal: devfreq_cooling: remove old power model and use EM Lukasz Luba
2020-12-02 10:26 ` Ionela Voinescu
2020-11-18 12:03 ` [PATCH v2 5/5] drm/panfrost: Register devfreq cooling and attempt to add Energy Model Lukasz Luba
2020-12-02 15:01 ` [PATCH v2 0/5] Thermal devfreq cooling improvements with " Daniel Lezcano
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5d4743b9-5b2f-8494-8d10-6a5fd2c0fdfd@linaro.org \
--to=daniel.lezcano@linaro.org \
--cc=airlied@linux.ie \
--cc=alyssa.rosenzweig@collabora.com \
--cc=amit.kucheria@verdurent.com \
--cc=daniel@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=ionela.voinescu@arm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pm@vger.kernel.org \
--cc=lukasz.luba@arm.com \
--cc=orjan.eide@arm.com \
--cc=robh@kernel.org \
--cc=rui.zhang@intel.com \
--cc=steven.price@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®