Re: [RFC/PATCH] tracing: show total NMI time in hwlat tracer

[Namhyung Kim]

On Fri, Dec 16, 2016 at 12:32 AM, Namhyung Kim <namhyung@xxxxxxxxxx> wrote:
> On Wed, Dec 14, 2016 at 11:17:16AM -0500, Steven Rostedt wrote:
>> @@ -203,8 +212,13 @@ static int get_sample(void)
>>                       }
>>                       if (diff > outer_sample)
>>                               outer_sample = diff;
>> +
>> +                     diff = nmi_t1 - last_nmi;
>
> It these are time_type, it should use time_to_us() instead.

Oh, I meant "time_to_us(time_sub(nmi_t1, last_nmi))".

Thanks,
Namhyung


>
>> +                     if (diff > nmi_sample)
>> +                             nmi_sample = diff;
>>               }
>>               last_t2 = t2;
>> +             last_nmi = nmi_t2;
>>
>>               total = time_to_us(time_sub(t2, start)); /* sample width */
>>
>> @@ -227,6 +241,10 @@ static int get_sample(void)
>>               if (diff > sample)
>>                       sample = diff; /* only want highest value */
>>
>> +             diff = nmi_t2 - nmi_t1;
>> +             if (diff > nmi_sample)
>> +                     nmi_sample = diff;
>> +
>>       } while (total <= hwlat_data.sample_width);
>>
>>       barrier(); /* finish the above in the view for NMIs */