Re: [for-next][PATCH 10/15] tracing: Allow perf to read synthetic events

[Namhyung Kim <namhyung@kernel.org>]

Hello,

On Fri, May 22, 2026 at 11:41:26AM -0400, Steven Rostedt wrote:
> On Fri, 22 May 2026 08:19:36 -0700
> Ian Rogers <irogers@google.com> wrote:
> 
> > On Fri, May 22, 2026 at 7:35 AM Steven Rostedt <rostedt@kernel.org> wrote:
> > >
> > > From: Steven Rostedt <rostedt@goodmis.org>
> > >
> > > Currently, perf can not enable synthetic events. When it does, it either
> > > causes a warning in the kernel or errors with "no such device".
> > >
> > > Add the necessary code to allow perf to also attach to synthetic events.  
> > 
> > Thanks so much for doing this work! I hoped to spend some time testing
> > it, but there's never enough time. For the synthetic event I was
> 
> I totally understand not having time. Heck, I'm currently unemployed and
> I'm still having a hard time finding time!

:)

> 
> > wondering what values end up in the perf sample like 'period'? I can
> 
> You mean to poll on the event?
> 
> > imagine this being the case and that, if you are doing something like
> > a futex wait duration, having to interpret the raw data rather than
> > just getting aggregation for free in `perf report` via the period.
> > That's not wrong but I was just curious.
> 
> I guess I'm not quite sure what you mean by "the period". But then again, I
> don't use perf that much.

Oh, it's about the value of the event.

IIUC, the period of any tracepoints is the number of times it hits.
Using a field of tracepoint instead needs interpretation of the event.

Thanks,
Namhyung