lkml.org 
[lkml]   [2008]   [Oct]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] ftrace: Add a C-state tracer to help power optimization

On Sat, 4 Oct 2008, Arjan van de Ven wrote:

> On Fri, 3 Oct 2008 20:32:51 -0400 (EDT)
> Steven Rostedt <rostedt@goodmis.org> wrote:
> > > +struct trace_cstate {
> > > + struct trace_entry ent;
> > > + struct cstate_trace state_data;
> > > +};
> > > +
> >
> > Can you please register this in the trace_assign_type macro.
> >
> > IF_ASSIGN(var, ent, struct trace_cstate, TRACE_BOOT); \
>
> I don't see this anywhere in the kernel ;(
>
> > And here, do not typecast. Use the trace_assign_type macro below.
>
> nor that guy

The two should be in linux-tip. If you are based against mainline, then
no they will not be.

If you are already using linux-tip, try refreshing.


>
>
> >
> > > + struct cstate_trace *it = &field->state_data;
> > > + struct trace_seq *s = &iter->seq;
> > > + struct timespec stamp = ktime_to_timespec(it->stamp);
> >
> > Also note that iter->ts holds a timestamp counter from bootup in
> > nanosecs. It currently uses the sched_clock to record, but may change
> > later to something a bit better.
>
> given that some of these clocks stop during idle... I really do care
> about which timestamp is used, tracing idle with a clock that stops
> during idle won't work too well.
>

OK, I just wanted to let you know about it.

-- Steve



\
 
 \ /
  Last update: 2008-10-04 19:17    [W:0.060 / U:0.184 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site