In this patch, the cpu_cooling code checks for the usability of cpufreq
layer before proceeding with the CPU cooling device registration. The
main reason is: CPU cooling device is not usable if cpufreq cannot
switch frequencies.
Similar checks are spread in thermal drivers. Thus, the advantage now
is to have the check in a single place: cpu cooling device registration.
For this reason, this patch also updates the existing drivers that
depend on CPU cooling to simply propagate the error code of the cpu
cooling registration call. Therefore, in case cpufreq is not ready, the
thermal drivers will still return -EPROBE_DEFER, in an attempt to try
again when cpufreq layer gets ready.
Cc: [email protected]
Cc: Grant Likely <[email protected]>
Cc: Kukjin Kim <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: Naveen Krishna Chatradhi <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Zhang Rui <[email protected]>
Signed-off-by: Eduardo Valentin <[email protected]>
---
drivers/thermal/cpu_cooling.c | 5 +++++
drivers/thermal/db8500_cpufreq_cooling.c | 5 -----
drivers/thermal/imx_thermal.c | 5 -----
drivers/thermal/samsung/exynos_thermal_common.c | 2 +-
drivers/thermal/ti-soc-thermal/ti-thermal-common.c | 6 ------
5 files changed, 6 insertions(+), 17 deletions(-)
---
This is attempt to organize the cpu cooling vs. cpufreq boot sequencing.
The main change in this patch, as in the commit log, is to have the check
for the cpufreq layer in the cpu cooling device registration, instead of
in thermal drivers. This way, drivers don't need to bother about it, they
just need to propagate the error value.
This change was tested on top of:
(0) - Viresh's change in cpufreq layer and cpufreq-dt (up to patch 4):
https://patchwork.kernel.org/patch/5384141/
https://patchwork.kernel.org/patch/5384151/
https://patchwork.kernel.org/patch/5384161/
https://patchwork.kernel.org/patch/5384171/
(1) - fix of thermal core:
https://patchwork.kernel.org/patch/5326991/
After Viresh's changes, cpufreq-dt is properly sequenced with cpu cooling
registration. Non-of based drivers also should take advantage if these
changes, as now they do not need to check for cpufreq layer. The check is
where it belongs, in cpu cooling device registration.
BR, Eduardo Valentin
diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
index 1ab0018..9e6945b 100644
--- a/drivers/thermal/cpu_cooling.c
+++ b/drivers/thermal/cpu_cooling.c
@@ -440,6 +440,11 @@ __cpufreq_cooling_register(struct device_node *np,
int ret = 0, i;
struct cpufreq_policy policy;
+ if (!cpufreq_get_current_driver() || !cpufreq_frequency_get_table(0)) {
+ pr_err("cpu_cooling: cpufreq layer not ready! Deferring.\n");
+ return ERR_PTR(-EPROBE_DEFER);
+ }
+
/* Verify that all the clip cpus have same freq_min, freq_max limit */
for_each_cpu(i, clip_cpus) {
/* continue if cpufreq policy not found and not return error */
diff --git a/drivers/thermal/db8500_cpufreq_cooling.c b/drivers/thermal/db8500_cpufreq_cooling.c
index 786d192..1ac7ec6 100644
--- a/drivers/thermal/db8500_cpufreq_cooling.c
+++ b/drivers/thermal/db8500_cpufreq_cooling.c
@@ -18,7 +18,6 @@
*/
#include <linux/cpu_cooling.h>
-#include <linux/cpufreq.h>
#include <linux/err.h>
#include <linux/module.h>
#include <linux/of.h>
@@ -30,10 +29,6 @@ static int db8500_cpufreq_cooling_probe(struct platform_device *pdev)
struct thermal_cooling_device *cdev;
struct cpumask mask_val;
- /* make sure cpufreq driver has been initialized */
- if (!cpufreq_frequency_get_table(0))
- return -EPROBE_DEFER;
-
cpumask_set_cpu(0, &mask_val);
cdev = cpufreq_cooling_register(&mask_val);
diff --git a/drivers/thermal/imx_thermal.c b/drivers/thermal/imx_thermal.c
index 5a1f107..16405b4 100644
--- a/drivers/thermal/imx_thermal.c
+++ b/drivers/thermal/imx_thermal.c
@@ -9,7 +9,6 @@
#include <linux/clk.h>
#include <linux/cpu_cooling.h>
-#include <linux/cpufreq.h>
#include <linux/delay.h>
#include <linux/device.h>
#include <linux/init.h>
@@ -459,10 +458,6 @@ static int imx_thermal_probe(struct platform_device *pdev)
int measure_freq;
int ret;
- if (!cpufreq_get_current_driver()) {
- dev_dbg(&pdev->dev, "no cpufreq driver!");
- return -EPROBE_DEFER;
- }
data = devm_kzalloc(&pdev->dev, sizeof(*data), GFP_KERNEL);
if (!data)
return -ENOMEM;
diff --git a/drivers/thermal/samsung/exynos_thermal_common.c b/drivers/thermal/samsung/exynos_thermal_common.c
index 3f5ad25..f84975e 100644
--- a/drivers/thermal/samsung/exynos_thermal_common.c
+++ b/drivers/thermal/samsung/exynos_thermal_common.c
@@ -373,7 +373,7 @@ int exynos_register_thermal(struct thermal_sensor_conf *sensor_conf)
if (IS_ERR(th_zone->cool_dev[th_zone->cool_dev_size])) {
dev_err(sensor_conf->dev,
"Failed to register cpufreq cooling device\n");
- ret = -EINVAL;
+ ret = PTR_ERR(th_zone->cool_dev[th_zone->cool_dev_size]);
goto err_unregister;
}
th_zone->cool_dev_size++;
diff --git a/drivers/thermal/ti-soc-thermal/ti-thermal-common.c b/drivers/thermal/ti-soc-thermal/ti-thermal-common.c
index 5fd0386..cf88585 100644
--- a/drivers/thermal/ti-soc-thermal/ti-thermal-common.c
+++ b/drivers/thermal/ti-soc-thermal/ti-thermal-common.c
@@ -28,7 +28,6 @@
#include <linux/kernel.h>
#include <linux/workqueue.h>
#include <linux/thermal.h>
-#include <linux/cpufreq.h>
#include <linux/cpumask.h>
#include <linux/cpu_cooling.h>
#include <linux/of.h>
@@ -407,11 +406,6 @@ int ti_thermal_register_cpu_cooling(struct ti_bandgap *bgp, int id)
if (!data)
return -EINVAL;
- if (!cpufreq_get_current_driver()) {
- dev_dbg(bgp->dev, "no cpufreq driver yet\n");
- return -EPROBE_DEFER;
- }
-
/* Register cooling device */
data->cool_dev = cpufreq_cooling_register(cpu_present_mask);
if (IS_ERR(data->cool_dev)) {
--
2.1.3
Few nits..
On 26 November 2014 at 23:20, Eduardo Valentin <[email protected]> wrote:
> Signed-off-by: Eduardo Valentin <[email protected]>
> ---
The normal practice is to write the non-commitable part here ...
> drivers/thermal/cpu_cooling.c | 5 +++++
> drivers/thermal/db8500_cpufreq_cooling.c | 5 -----
> drivers/thermal/imx_thermal.c | 5 -----
> drivers/thermal/samsung/exynos_thermal_common.c | 2 +-
> drivers/thermal/ti-soc-thermal/ti-thermal-common.c | 6 ------
> 5 files changed, 6 insertions(+), 17 deletions(-)
> ---
But this works as well :)
> This is attempt to organize the cpu cooling vs. cpufreq boot sequencing.
> The main change in this patch, as in the commit log, is to have the check
> for the cpufreq layer in the cpu cooling device registration, instead of
> in thermal drivers. This way, drivers don't need to bother about it, they
> just need to propagate the error value.
>
> This change was tested on top of:
> (0) - Viresh's change in cpufreq layer and cpufreq-dt (up to patch 4):
> https://patchwork.kernel.org/patch/5384141/
> https://patchwork.kernel.org/patch/5384151/
> https://patchwork.kernel.org/patch/5384161/
> https://patchwork.kernel.org/patch/5384171/
> (1) - fix of thermal core:
> https://patchwork.kernel.org/patch/5326991/
>
> After Viresh's changes, cpufreq-dt is properly sequenced with cpu cooling
> registration. Non-of based drivers also should take advantage if these
> changes, as now they do not need to check for cpufreq layer. The check is
> where it belongs, in cpu cooling device registration.
>
> BR, Eduardo Valentin
>
>
> diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
> index 1ab0018..9e6945b 100644
> --- a/drivers/thermal/cpu_cooling.c
> +++ b/drivers/thermal/cpu_cooling.c
> @@ -440,6 +440,11 @@ __cpufreq_cooling_register(struct device_node *np,
> int ret = 0, i;
> struct cpufreq_policy policy;
>
> + if (!cpufreq_get_current_driver() || !cpufreq_frequency_get_table(0)) {
Only !cpufreq_frequency_get_table(0) is enough here.
For rest:
Acked-by: Viresh Kumar <[email protected]>
Hello Viresh,
On Thu, Nov 27, 2014 at 09:38:39AM +0530, Viresh Kumar wrote:
> Few nits..
>
> On 26 November 2014 at 23:20, Eduardo Valentin <[email protected]> wrote:
>
> > Signed-off-by: Eduardo Valentin <[email protected]>
> > ---
>
> The normal practice is to write the non-commitable part here ...
>
err... normal practice by whom? hehe...
My "normal" practice is to allow people to read the diff stat before my
extra comments :-)
> > drivers/thermal/cpu_cooling.c | 5 +++++
> > drivers/thermal/db8500_cpufreq_cooling.c | 5 -----
> > drivers/thermal/imx_thermal.c | 5 -----
> > drivers/thermal/samsung/exynos_thermal_common.c | 2 +-
> > drivers/thermal/ti-soc-thermal/ti-thermal-common.c | 6 ------
> > 5 files changed, 6 insertions(+), 17 deletions(-)
> > ---
>
> But this works as well :)
>
hehe ok.
> > This is attempt to organize the cpu cooling vs. cpufreq boot sequencing.
> > The main change in this patch, as in the commit log, is to have the check
> > for the cpufreq layer in the cpu cooling device registration, instead of
> > in thermal drivers. This way, drivers don't need to bother about it, they
> > just need to propagate the error value.
> >
> > This change was tested on top of:
> > (0) - Viresh's change in cpufreq layer and cpufreq-dt (up to patch 4):
> > https://patchwork.kernel.org/patch/5384141/
> > https://patchwork.kernel.org/patch/5384151/
> > https://patchwork.kernel.org/patch/5384161/
> > https://patchwork.kernel.org/patch/5384171/
> > (1) - fix of thermal core:
> > https://patchwork.kernel.org/patch/5326991/
> >
> > After Viresh's changes, cpufreq-dt is properly sequenced with cpu cooling
> > registration. Non-of based drivers also should take advantage if these
> > changes, as now they do not need to check for cpufreq layer. The check is
> > where it belongs, in cpu cooling device registration.
> >
> > BR, Eduardo Valentin
> >
> >
> > diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
> > index 1ab0018..9e6945b 100644
> > --- a/drivers/thermal/cpu_cooling.c
> > +++ b/drivers/thermal/cpu_cooling.c
> > @@ -440,6 +440,11 @@ __cpufreq_cooling_register(struct device_node *np,
> > int ret = 0, i;
> > struct cpufreq_policy policy;
> >
> > + if (!cpufreq_get_current_driver() || !cpufreq_frequency_get_table(0)) {
>
> Only !cpufreq_frequency_get_table(0) is enough here.
>
Yeah, I thought of it too. Just combined what people had in their
drivers here. But I agree that latter is enough.
> For rest:
>
> Acked-by: Viresh Kumar <[email protected]>
Ok.
though.. "normal practice" says ack's are oftern used by the maintainer
of the affected code (quoting Documentation/SubmittingPatches) :-)
BR, Eduardo Valenti
On 27 November 2014 at 19:38, Eduardo Valentin <[email protected]> wrote:
>> Acked-by: Viresh Kumar <[email protected]>
>
> Ok.
>
> though.. "normal practice" says ack's are oftern used by the maintainer
> of the affected code (quoting Documentation/SubmittingPatches) :-)
Hehe :)
Another paragraph from the same file:
Acked-by: does not necessarily indicate acknowledgement of the entire patch.
For example, if a patch affects multiple subsystems and has an Acked-by: from
one subsystem maintainer then this usually indicates acknowledgement of just
the part which affects that maintainer's code. Judgement should be used here.
When in doubt people should refer to the original discussion in the mailing
list archives.
So my Acked-by was as cpufreq subsystem Maintainer :)