Re: [RFC PATCH v15 01/11] ARM: cpuidle: Register per cpuidle device

[Lina Iyer <lina.iyer@linaro.org>]

On Tue, Mar 10 2015 at 04:37 -0600, Russell King - ARM Linux wrote:
>On Mon, Mar 09, 2015 at 09:16:36AM -0600, Lina Iyer wrote:
>> @@ -105,18 +109,46 @@ static int __init arm_idle_init(void)
>>  	if (ret <= 0)
>>  		return ret ? : -ENODEV;
>>
>> +
>
>A bit better formatting would be nice - you don't need the extra blank
>line here.
>
>> +	ret = cpuidle_register_driver(drv);
>> +	if (ret) {
>> +		pr_err("Failed to register cpuidle driver\n");
>> +		return ret;
>> +	}
>> +
>>  	/*
>>  	 * Call arch CPU operations in order to initialize
>>  	 * idle states suspend back-end specific data
>>  	 */
>>  	for_each_possible_cpu(cpu) {
>> +
>
>This blank line is not necessary either.
>
>>  		ret = arm_cpuidle_init(cpu);
>
>However, a blank line here would be a good thing.
>
Sure.
>> +		/*
>> +		 * This cpu does not support any idle states
>> +		 */
>
>Also, formatting this as /* This cpu does not support any idle states */ is
>acceptable too, and doesn't waste as many lines.
>
Will fix.
>> +		if (ret == -ENOSYS)
>> +			continue;
>> +
>>  		if (ret) {
>>  			pr_err("CPU %d failed to init idle CPU ops\n", cpu);
>>  			return ret;
>>  		}
>> +
>> +		dev = kzalloc(sizeof(*dev), GFP_KERNEL);
>> +		if (!dev)
>> +			return -ENOMEM;
>> +
>> +		dev->cpu = cpu;
>> +		ret = cpuidle_register_device(dev);
>> +		if (ret) {
>> +			pr_err("Failed to register cpuidle device for CPU %d\n",
>> +			       cpu);
>> +			return ret;
>
>It looks like we leak the 'dev' allocation here.
>

True, I will amend that.

>Also, other error paths, it looks like we leave the previously registered
>cpuidle devices in place.  That may be acceptable if the intention is to
>initialise as many CPUs as possible - but we then miss the
>cpuidle_register() call below, which seems to make the registered devices
>useless.  It's a little inconsistent.
>
>Also, it's useful to report why something fails - printing the error code
>can help debugging if it isn't already printed elsewhere.
>
Fair point.
>> +		}
>> +
>> +		per_cpu(cpuidle_arm_dev, cpu) = dev;
>>  	}
>>
>> -	return cpuidle_register(drv, NULL);
>> +	return 0;
>>  }
>>  device_initcall(arm_idle_init);
>
>-- 
>FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
>according to speedtest.net.