cpuidle: dt: bail out if the idle-state DT node is not compatible

Message ID 20190131022749.19046-1-josephl@nvidia.com
State New
Headers show
Series
  • cpuidle: dt: bail out if the idle-state DT node is not compatible
Related show

Commit Message

Joseph Lo Jan. 31, 2019, 2:27 a.m.
Currently, the DT of the idle states will be parsed first whether it's
compatible or not. This could cause a warning message that comes from if
the CPU doesn't support identical idle states. E.g. Tegra186 can run
with 2 Cortex-A57 and 2 Denver cores with different idle states on
different types of these cores.

We fix it to check the compatible string before the parsing procedures.
So it can make sure it only goes through the idle states that the CPU
supported.

Signed-off-by: Joseph Lo <josephl@nvidia.com>
---
 drivers/cpuidle/dt_idle_states.c | 5 +++++
 1 file changed, 5 insertions(+)

Comments

Sudeep Holla Jan. 31, 2019, 10:38 a.m. | #1
On Thu, Jan 31, 2019 at 10:27:49AM +0800, Joseph Lo wrote:
> Currently, the DT of the idle states will be parsed first whether it's
> compatible or not. This could cause a warning message that comes from if
> the CPU doesn't support identical idle states. E.g. Tegra186 can run
> with 2 Cortex-A57 and 2 Denver cores with different idle states on
> different types of these cores.
>
> We fix it to check the compatible string before the parsing procedures.
> So it can make sure it only goes through the idle states that the CPU
> supported.
>
> Signed-off-by: Joseph Lo <josephl@nvidia.com>
> ---
>  drivers/cpuidle/dt_idle_states.c | 5 +++++
>  1 file changed, 5 insertions(+)
>
> diff --git a/drivers/cpuidle/dt_idle_states.c b/drivers/cpuidle/dt_idle_states.c
> index 53342b7f1010..4a528908abb3 100644
> --- a/drivers/cpuidle/dt_idle_states.c
> +++ b/drivers/cpuidle/dt_idle_states.c
> @@ -180,6 +180,11 @@ int dt_init_idle_driver(struct cpuidle_driver *drv,
>  		if (!state_node)
>  			break;
>
> +		if (!of_device_is_compatible(state_node, matches->compatible)) {
> +			err = -EINVAL;
> +			break;
> +		}
> +

Does it make sense to use of_match_node to get the matching device id and
just pass that to init_state_node instead of the complete array to find the
match.

--
Regards,
Sudeep
Joseph Lo Feb. 1, 2019, 1:42 a.m. | #2
On 1/31/19 6:38 PM, Sudeep Holla wrote:
> On Thu, Jan 31, 2019 at 10:27:49AM +0800, Joseph Lo wrote:
>> Currently, the DT of the idle states will be parsed first whether it's
>> compatible or not. This could cause a warning message that comes from if
>> the CPU doesn't support identical idle states. E.g. Tegra186 can run
>> with 2 Cortex-A57 and 2 Denver cores with different idle states on
>> different types of these cores.
>>
>> We fix it to check the compatible string before the parsing procedures.
>> So it can make sure it only goes through the idle states that the CPU
>> supported.
>>
>> Signed-off-by: Joseph Lo <josephl@nvidia.com>
>> ---
>>   drivers/cpuidle/dt_idle_states.c | 5 +++++
>>   1 file changed, 5 insertions(+)
>>
>> diff --git a/drivers/cpuidle/dt_idle_states.c b/drivers/cpuidle/dt_idle_states.c
>> index 53342b7f1010..4a528908abb3 100644
>> --- a/drivers/cpuidle/dt_idle_states.c
>> +++ b/drivers/cpuidle/dt_idle_states.c
>> @@ -180,6 +180,11 @@ int dt_init_idle_driver(struct cpuidle_driver *drv,
>>   		if (!state_node)
>>   			break;
>>
>> +		if (!of_device_is_compatible(state_node, matches->compatible)) {
>> +			err = -EINVAL;
>> +			break;
>> +		}
>> +
> 
> Does it make sense to use of_match_node to get the matching device id and
> just pass that to init_state_node instead of the complete array to find the
> match.

Yes, that would be even better.

Thanks,
Joseph

Patch

diff --git a/drivers/cpuidle/dt_idle_states.c b/drivers/cpuidle/dt_idle_states.c
index 53342b7f1010..4a528908abb3 100644
--- a/drivers/cpuidle/dt_idle_states.c
+++ b/drivers/cpuidle/dt_idle_states.c
@@ -180,6 +180,11 @@  int dt_init_idle_driver(struct cpuidle_driver *drv,
 		if (!state_node)
 			break;
 
+		if (!of_device_is_compatible(state_node, matches->compatible)) {
+			err = -EINVAL;
+			break;
+		}
+
 		if (!of_device_is_available(state_node)) {
 			of_node_put(state_node);
 			continue;