Re: [PATCH v4 3/3] tracing: format non-nanosec times from tsc clockwithout a decimal point.

[David Sharp]

On Tue, Sep 25, 2012 at 4:36 PM, Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> On Tue, 2012-09-25 at 15:29 -0700, David Sharp wrote:
>
>
>> >> +             ret = trace_seq_printf(
>> >> +                             s, "[%08llx] %ld.%03ldms (+%ld.%03ldms): ",
>> >> +                             ns2usecs(iter->ts),
>> >> +                             abs_msec, abs_usec,
>> >> +                             rel_msec, rel_usec);
>> >> +     } else if (verbose && !in_ns) {
>> >> +             ret = trace_seq_printf(
>> >> +                             s, "[%016llx] %lld (+%lld): ",
>> >> +                             iter->ts, abs_ts, rel_ts);
>> >> +     } else { /* !verbose */
>> >> +             ret = trace_seq_printf(
>> >> +                             s, " %4lld%s%c: ",
>> >> +                             abs_ts,
>> >> +                             in_ns ? "us" : "",
>> >> +                             rel_ts > mark_thresh ? '!' :
>> >> +                               rel_ts > 1 ? '+' : ' ');
>>
>> I just noticed something about this: with x86-tsc clock, this will
>> always print a '+'. Does it matter? Also, is the 200k cycle threshold
>> for '!' okay? I guess the counter clock will always end up with rel_ts
>> == 1, so marks should never appear.
>>
>
> Actually, I'm thinking that counters should not add those annotations.
> As it just doesn't make sense.

Right. But they won't appear anyway, since the delta will always be 1.

wait, by "counters" are you including TSC? Surely that makes sense,
since it is a measurement of time.

Eh... sorry I brought it up. I don't really want to change it. I never
use the latency tracer, so I mostly just don't want to break it.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/