Re: [PATCH v3] cpufreq: stats: Add 'load_table' debugfs file to show accumulated data of CPUs

[Chanwoo Choi <cw00.choi@samsung.com>]

On 06/27/2013 07:23 PM, Viresh Kumar wrote:
> On 27 June 2013 15:44, Chanwoo Choi <cw00.choi@samsung.com> wrote:
>> On 06/26/2013 05:04 PM, Viresh Kumar wrote:
>>> On 24 June 2013 14:32, Chanwoo Choi <cw00.choi@samsung.com> wrote:
> 
>>> Maybe 10 to 1000.. Lets give others a chance to see long logs :)
>>>
>>
>> OK, I'll extend the maximum value from 100 to 1000.
> 
> 10-1000 please.

OK.

>>>> +       freq.old = freq.new = policy->cur;
>>>
>>> No need to set freq.new here.
>>
>> If cpufreq governor don't change cpu frequency on specific situation,
>> cpufreq SoC driver won't send CPUFREQ_POSTCHANGE. In case of this situation,
>> I store current cpu frequency to freq.new field.
> 
> You are doing this at the time of LOADCHECK notification :)
> 
>                stat->load_table[last_idx].new = freq->old;
> 

OK, I'll fix it.

>>>> +static ssize_t load_table_read(struct file *file, char __user *user_buf,
>>>> +                                       size_t count, loff_t *ppos)
>>>> +{
>>>> +       struct cpufreq_policy *policy = file->private_data;
>>>> +       struct cpufreq_stats *stat = per_cpu(cpufreq_stats_table, policy->cpu);
>>>> +       struct cpufreq_freqs *load_table = stat->load_table;
>>>> +       ssize_t len = 0;
>>>> +       char *buf;
>>>> +       int i, cpu, ret;
>>>> +
>>>> +       buf = kzalloc(MAX_LINE_SIZE * stat->load_max_index, GFP_KERNEL);
>>>> +       if (!buf)
>>>> +               return 0;
>>>
>>> Above use of stat->load_max_index must be inside locks I guess. Otherwise
>>> you may allocate memory for 10 lines and by the time lock is taken, we
>>> already have 12 entries. And so, below loop will go beyond array limits.
>>>
>>
>> I store CONFIG_NR_CPU_LOAD_STORAGE to stat->load_max_index in cpufreq_stats_create_debugfs()
>> So, stat->load_max_index value isn't always 10.
>>
>> If I misunderstood for your comment, I'd like you to explain more detailed about this comment.
> 
> No you didn't but looking second time at the code, i couldn't find a
> problem with it.
> 
> You allocate memory for max entries and so shouldn't be a problem.
> 

OK,

I'll send v4 patch. Thanks.

Best Regards,
Chanwoo Choi