Re: [PATCH v2] cpufreq: cppc: Set fie_disabled to FIE_DISABLED if fails to create kworker_fie

["Liao, Chang" <liaochang1@huawei.com>]

在 2023/8/16 18:37, Viresh Kumar 写道:
> On 16-08-23, 09:49, Liao Chang wrote:
>> The function cppc_freq_invariance_init() may failed to create
>> kworker_fie, make it more robust by setting fie_disabled to FIE_DISBALED
>> to prevent an invalid pointer dereference in kthread_destroy_worker(),
>> which called from cppc_freq_invariance_exit().
>>
>> Link: https://lore.kernel.org/all/20230816034630.a4hvsj373q6aslk3@vireshk-i7/
>>
>> Signed-off-by: Liao Chang <liaochang1@huawei.com>
>> ---
>>  drivers/cpufreq/cppc_cpufreq.c | 28 ++++++++++++++++------------
>>  1 file changed, 16 insertions(+), 12 deletions(-)
>>
>> diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
>> index 022e3555407c..bff4cde06083 100644
>> --- a/drivers/cpufreq/cppc_cpufreq.c
>> +++ b/drivers/cpufreq/cppc_cpufreq.c
>> @@ -220,6 +220,15 @@ static void cppc_cpufreq_cpu_fie_exit(struct cpufreq_policy *policy)
>>  	}
>>  }
>>  
>> +static void cppc_freq_invariance_exit(void)
>> +{
>> +	if (fie_disabled)
>> +		return;
>> +
>> +	kthread_destroy_worker(kworker_fie);
>> +	kworker_fie = NULL;
>> +}
>> +
>>  static void __init cppc_freq_invariance_init(void)
>>  {
>>  	struct sched_attr attr = {
>> @@ -249,27 +258,22 @@ static void __init cppc_freq_invariance_init(void)
>>  		return;
>>  
>>  	kworker_fie = kthread_create_worker(0, "cppc_fie");
>> -	if (IS_ERR(kworker_fie))
>> +	if (IS_ERR(kworker_fie)) {
>> +		pr_warn("%s: failed to create kworker_fie: %ld\n", __func__,
>> +			PTR_ERR(kworker_fie));
>> +		fie_disabled = FIE_DISABLED;
>>  		return;
>> +	}
>>  
>>  	ret = sched_setattr_nocheck(kworker_fie->task, &attr);
>>  	if (ret) {
>>  		pr_warn("%s: failed to set SCHED_DEADLINE: %d\n", __func__,
>>  			ret);
>> -		kthread_destroy_worker(kworker_fie);
>> -		return;
>> +		cppc_freq_invariance_exit();
> 
> I don't really like this change, there aren't a lot of things that we
> need to do here on cleanup, but just kthread_destroy_worker(). Calling
> it directly makes more sense I guess.

OK，I will simplify code in next revisoin.

> 
>> +		fie_disabled = FIE_DISABLED;
>>  	}
>>  }
>>  
>> -static void cppc_freq_invariance_exit(void)
>> -{
>> -	if (fie_disabled)
>> -		return;
>> -
>> -	kthread_destroy_worker(kworker_fie);
> 
>> -	kworker_fie = NULL;
> 
> I don't see the point of this line. Probably it can be dropped.

I agree with you, when fie_disbaled is set to FIE_DISABLED and kthred_destroy_worker() is called,
the driver has no chance to access the invalid kworker_fie pointer.

Thanks.

> 
>> -}
>> -
>>  #else
>>  static inline void cppc_cpufreq_cpu_fie_init(struct cpufreq_policy *policy)
>>  {
>> -- 
>> 2.34.1
> 

-- 
BR
Liao, Chang