In em_create_perf_table(), power is uninitialized and passed the pointer
to active_power() hook, but the hook function may not assign it and
return 0, such as mtk_cpufreq_get_cpu_power(), so the later zero check for
power is not invalid, initialize power to zero to fix it.

Cc: stable@vger.kernel.org
Fixes: 7d9895c7fbfc ("PM / EM: introduce em_dev_register_perf_domain function")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
---
 kernel/power/energy_model.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
index 927cc55ba0b3..866a3e9c05b2 100644
--- a/kernel/power/energy_model.c
+++ b/kernel/power/energy_model.c
@@ -344,7 +344,7 @@ static int em_create_perf_table(struct device *dev, struct em_perf_domain *pd,
 				struct em_data_callback *cb,
 				unsigned long flags)
 {
-	unsigned long power, freq, prev_freq = 0;
+	unsigned long power = 0, freq, prev_freq = 0;
 	int nr_states = pd->nr_perf_states;
 	int i, ret;
 
-- 
2.34.1

Hi,

On Monday 04 Nov 2024 at 17:03:51 (+0800), Jinjie Ruan wrote:
> In em_create_perf_table(), power is uninitialized and passed the pointer
> to active_power() hook, but the hook function may not assign it and
> return 0, such as mtk_cpufreq_get_cpu_power(), so the later zero check for
> power is not invalid, initialize power to zero to fix it.

Sounds to me like a driver bug? We check the return value of
active_power() first, so if the callback failed it should tell us.

mtk_cpufreq_get_cpu_power() should only fail if we couldn't find a
cpufreq policy IIUC, so why can't this return -EINVAL instead?

Thanks,
Quentin

Hi Jinjie,

On 11/4/24 09:03, Jinjie Ruan wrote:
> In em_create_perf_table(), power is uninitialized and passed the pointer
> to active_power() hook, but the hook function may not assign it and
> return 0, such as mtk_cpufreq_get_cpu_power(), so the later zero check for

Please fix the driver. I have checked that function. It must return
-EINVAL when the 'policy' is not found. We cannot progress with power=0.


> power is not invalid, initialize power to zero to fix it.
> 
> Cc: stable@vger.kernel.org
> Fixes: 7d9895c7fbfc ("PM / EM: introduce em_dev_register_perf_domain function")
> Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
> ---
>   kernel/power/energy_model.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
> index 927cc55ba0b3..866a3e9c05b2 100644
> --- a/kernel/power/energy_model.c
> +++ b/kernel/power/energy_model.c
> @@ -344,7 +344,7 @@ static int em_create_perf_table(struct device *dev, struct em_perf_domain *pd,
>   				struct em_data_callback *cb,
>   				unsigned long flags)
>   {
> -	unsigned long power, freq, prev_freq = 0;
> +	unsigned long power = 0, freq, prev_freq = 0;
>   	int nr_states = pd->nr_perf_states;
>   	int i, ret;
>   


This patch proposal is just a workaround.

When you send a patch to that MTK driver, I can review it for you so
please add me on CC.

Regards,
Lukasz

On 2024/11/4 17:14, Lukasz Luba wrote:
> Hi Jinjie,
> 
> On 11/4/24 09:03, Jinjie Ruan wrote:
>> In em_create_perf_table(), power is uninitialized and passed the pointer
>> to active_power() hook, but the hook function may not assign it and
>> return 0, such as mtk_cpufreq_get_cpu_power(), so the later zero check
>> for
> 
> Please fix the driver. I have checked that function. It must return
> -EINVAL when the 'policy' is not found. We cannot progress with power=0.

Thank you very much! It will be also good to fix the driver.

> 
> 
>> power is not invalid, initialize power to zero to fix it.
>>
>> Cc: stable@vger.kernel.org
>> Fixes: 7d9895c7fbfc ("PM / EM: introduce em_dev_register_perf_domain
>> function")
>> Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
>> ---
>>   kernel/power/energy_model.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
>> index 927cc55ba0b3..866a3e9c05b2 100644
>> --- a/kernel/power/energy_model.c
>> +++ b/kernel/power/energy_model.c
>> @@ -344,7 +344,7 @@ static int em_create_perf_table(struct device
>> *dev, struct em_perf_domain *pd,
>>                   struct em_data_callback *cb,
>>                   unsigned long flags)
>>   {
>> -    unsigned long power, freq, prev_freq = 0;
>> +    unsigned long power = 0, freq, prev_freq = 0;
>>       int nr_states = pd->nr_perf_states;
>>       int i, ret;
>>   
> 
> 
> This patch proposal is just a workaround.
> 
> When you send a patch to that MTK driver, I can review it for you so
> please add me on CC.
> 
> Regards,
> Lukasz
> 

On Monday 04 Nov 2024 at 09:14:36 (+0000), Lukasz Luba wrote:
> Hi Jinjie,
> 
> On 11/4/24 09:03, Jinjie Ruan wrote:
> > In em_create_perf_table(), power is uninitialized and passed the pointer
> > to active_power() hook, but the hook function may not assign it and
> > return 0, such as mtk_cpufreq_get_cpu_power(), so the later zero check for
> 
> Please fix the driver. I have checked that function. It must return
> -EINVAL when the 'policy' is not found. We cannot progress with power=0.
> 
> 
> > power is not invalid, initialize power to zero to fix it.
> > 
> > Cc: stable@vger.kernel.org
> > Fixes: 7d9895c7fbfc ("PM / EM: introduce em_dev_register_perf_domain function")
> > Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
> > ---
> >   kernel/power/energy_model.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
> > index 927cc55ba0b3..866a3e9c05b2 100644
> > --- a/kernel/power/energy_model.c
> > +++ b/kernel/power/energy_model.c
> > @@ -344,7 +344,7 @@ static int em_create_perf_table(struct device *dev, struct em_perf_domain *pd,
> >   				struct em_data_callback *cb,
> >   				unsigned long flags)
> >   {
> > -	unsigned long power, freq, prev_freq = 0;
> > +	unsigned long power = 0, freq, prev_freq = 0;
> >   	int nr_states = pd->nr_perf_states;
> >   	int i, ret;
> 
> 
> This patch proposal is just a workaround.
> 
> When you send a patch to that MTK driver, I can review it for you so
> please add me on CC.

We raced, but +1 to the above :-)