__cpufreq_cooling_register() might fail if some CPU other than first one in
clip_cpu mask is present earlier e.g. CPU hotplug. Iterate all CPUs in the mask
to handle this case.
Signed-off-by: Pi-Cheng Chen <[email protected]>
---
drivers/thermal/cpu_cooling.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
index 6509c61..d5e4ea8 100644
--- a/drivers/thermal/cpu_cooling.c
+++ b/drivers/thermal/cpu_cooling.c
@@ -776,7 +776,13 @@ __cpufreq_cooling_register(struct device_node *np,
char dev_name[THERMAL_NAME_LENGTH];
struct cpufreq_frequency_table *pos, *table;
unsigned int freq, i, num_cpus;
- int ret;
+ int ret, cpu;
+
+ for_each_cpu(cpu, clip_cpus) {
+ table = cpufreq_frequency_get_table(cpu);
+ if (table)
+ break;
+ }
table = cpufreq_frequency_get_table(cpumask_first(clip_cpus));
if (!table) {
--
1.9.1
Sorry for the mistake I made when cherry-picking the patch. Fix and resend
again.
__cpufreq_cooling_register() might fail if some CPU other than first one in
clip_cpu mask is present earlier e.g. CPU hotplug. Iterate all CPUs in the mask
to handle this case.
Signed-off-by: Pi-Cheng Chen <[email protected]>
---
drivers/thermal/cpu_cooling.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
index 6509c61..5e90eb6 100644
--- a/drivers/thermal/cpu_cooling.c
+++ b/drivers/thermal/cpu_cooling.c
@@ -776,9 +776,14 @@ __cpufreq_cooling_register(struct device_node *np,
char dev_name[THERMAL_NAME_LENGTH];
struct cpufreq_frequency_table *pos, *table;
unsigned int freq, i, num_cpus;
- int ret;
+ int ret, cpu;
+
+ for_each_cpu(cpu, clip_cpus) {
+ table = cpufreq_frequency_get_table(cpu);
+ if (table)
+ break;
+ }
- table = cpufreq_frequency_get_table(cpumask_first(clip_cpus));
if (!table) {
pr_debug("%s: CPUFreq table not found\n", __func__);
return ERR_PTR(-EPROBE_DEFER);
--
1.9.1
On Wed, Jul 01, 2015 at 05:13:14AM +0100, Pi-Cheng Chen wrote:
> Sorry for the mistake I made when cherry-picking the patch. Fix and resend
> again.
>
> __cpufreq_cooling_register() might fail if some CPU other than first one in
> clip_cpu mask is present earlier e.g. CPU hotplug. Iterate all CPUs in the mask
> to handle this case.
>
> Signed-off-by: Pi-Cheng Chen <[email protected]>
Looks good to me. FWIW,
Acked-by: Javi Merino <[email protected]>
> ---
> drivers/thermal/cpu_cooling.c | 9 +++++++--
> 1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
> index 6509c61..5e90eb6 100644
> --- a/drivers/thermal/cpu_cooling.c
> +++ b/drivers/thermal/cpu_cooling.c
> @@ -776,9 +776,14 @@ __cpufreq_cooling_register(struct device_node *np,
> char dev_name[THERMAL_NAME_LENGTH];
> struct cpufreq_frequency_table *pos, *table;
> unsigned int freq, i, num_cpus;
> - int ret;
> + int ret, cpu;
> +
> + for_each_cpu(cpu, clip_cpus) {
> + table = cpufreq_frequency_get_table(cpu);
> + if (table)
> + break;
> + }
>
> - table = cpufreq_frequency_get_table(cpumask_first(clip_cpus));
> if (!table) {
> pr_debug("%s: CPUFreq table not found\n", __func__);
> return ERR_PTR(-EPROBE_DEFER);
> --
> 1.9.1
>
>
> _______________________________________________
> linux-arm-kernel mailing list
> [email protected]
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
>
On 01-07-15, 12:13, Pi-Cheng Chen wrote:
> Sorry for the mistake I made when cherry-picking the patch. Fix and resend
> again.
You really want above to show up in git logs ?
Any comments like this should be present:
- in cover-letter
- OR after the three dashes below ---
- OR must be followed with a scissors line, like this:
--------------8<--------
> __cpufreq_cooling_register() might fail if some CPU other than first one in
> clip_cpu mask is present earlier e.g. CPU hotplug. Iterate all CPUs in the mask
> to handle this case.
>
> Signed-off-by: Pi-Cheng Chen <[email protected]>
> ---
> drivers/thermal/cpu_cooling.c | 9 +++++++--
> 1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
> index 6509c61..5e90eb6 100644
> --- a/drivers/thermal/cpu_cooling.c
> +++ b/drivers/thermal/cpu_cooling.c
> @@ -776,9 +776,14 @@ __cpufreq_cooling_register(struct device_node *np,
> char dev_name[THERMAL_NAME_LENGTH];
> struct cpufreq_frequency_table *pos, *table;
> unsigned int freq, i, num_cpus;
> - int ret;
> + int ret, cpu;
> +
> + for_each_cpu(cpu, clip_cpus) {
> + table = cpufreq_frequency_get_table(cpu);
> + if (table)
> + break;
> + }
>
> - table = cpufreq_frequency_get_table(cpumask_first(clip_cpus));
Nah, that's wrong. I hope that's a hypothetical problem and not a real
one. Would have been better if cpufreq maintainers were cc'd as they
can provide more insight into this :)
cpufreq_frequency_get_table() does: policy->freq_table and so it
doesn't matter if the cpu is online or not.
cpufreq_cpu_data was getting unset earlier on hotplug, but that's not
the case anymore. So nothing to worry about :)
--
viresh
Hi Viresh,
On Tue, Jul 7, 2015 at 6:27 PM, Viresh Kumar <[email protected]> wrote:
> On 01-07-15, 12:13, Pi-Cheng Chen wrote:
>> Sorry for the mistake I made when cherry-picking the patch. Fix and resend
>> again.
>
> You really want above to show up in git logs ?
>
> Any comments like this should be present:
> - in cover-letter
> - OR after the three dashes below ---
> - OR must be followed with a scissors line, like this:
> --------------8<--------
Thanks for the lesson. :)
>
>> __cpufreq_cooling_register() might fail if some CPU other than first one in
>> clip_cpu mask is present earlier e.g. CPU hotplug. Iterate all CPUs in the mask
>> to handle this case.
>>
>> Signed-off-by: Pi-Cheng Chen <[email protected]>
>> ---
>> drivers/thermal/cpu_cooling.c | 9 +++++++--
>> 1 file changed, 7 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c
>> index 6509c61..5e90eb6 100644
>> --- a/drivers/thermal/cpu_cooling.c
>> +++ b/drivers/thermal/cpu_cooling.c
>> @@ -776,9 +776,14 @@ __cpufreq_cooling_register(struct device_node *np,
>> char dev_name[THERMAL_NAME_LENGTH];
>> struct cpufreq_frequency_table *pos, *table;
>> unsigned int freq, i, num_cpus;
>> - int ret;
>> + int ret, cpu;
>> +
>> + for_each_cpu(cpu, clip_cpus) {
>> + table = cpufreq_frequency_get_table(cpu);
>> + if (table)
>> + break;
>> + }
>>
>> - table = cpufreq_frequency_get_table(cpumask_first(clip_cpus));
>
> Nah, that's wrong. I hope that's a hypothetical problem and not a real
> one. Would have been better if cpufreq maintainers were cc'd as they
> can provide more insight into this :)
>
> cpufreq_frequency_get_table() does: policy->freq_table and so it
> doesn't matter if the cpu is online or not.
>
> cpufreq_cpu_data was getting unset earlier on hotplug, but that's not
> the case anymore. So nothing to worry about :)
Yes. This is based on the case before 4.2-rc1 which policy of cpu will
be destroy after all cpus in a cluster are unplugged (right?). Since it
is not the case anymore now, this patch is not needed at all.
Thanks.
Best Regards,
Pi-Cheng
>
> --
> viresh