Re: [PATCH 1/2] tracing: Fix synth event test to avoid using smp_processor_id()

[Masami Hiramatsu]

On Thu, 20 Feb 2020 17:48:01 -0500
Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:

> On Mon, 17 Feb 2020 18:52:29 +0900
> Masami Hiramatsu <mhiramat@xxxxxxxxxx> wrote:
> 
> > Since smp_processor_id() requires irq-disabled or preempt-disabled,
> > synth event generation test module made some warnings. To prevent
> > that, use get_cpu()/put_cpu() instead of smp_processor_id().
> > 
> > Signed-off-by: Masami Hiramatsu <mhiramat@xxxxxxxxxx>
> > ---
> >  kernel/trace/synth_event_gen_test.c |   23 +++++++++++++++++------
> >  1 file changed, 17 insertions(+), 6 deletions(-)
> >
> 
> I just noticed this patch, after applying my version that just uses the
> raw_smp_processor_id(). We don't really care what CPU it is do we?
> 
> I didn't want a test to muck with preemption disabling and all that fun.

OK, I confirmed that the ring_buffer_nest_start() ensures the preempt
disabled. So just using raw_smp_processor_id() is good to me.

Thank you,

-- 
Masami Hiramatsu <mhiramat@xxxxxxxxxx>