[PATCH RFC] powerpc/ftrace: add powerpc timebase as a trace clock source

[Michael Ellerman]

On Wed, 2015-04-22 at 10:48 +0530, Naveen N. Rao wrote:
> On 2015/04/21 09:25AM, Steven Rostedt wrote:
> > On Tue, 21 Apr 2015 16:33:36 +0530
> > "Naveen N. Rao" <naveen.n.rao at linux.vnet.ibm.com> wrote:
> > 
> > > Add a new powerpc-specific trace clock using the timebase register,
> > > similar to x86-tsc. This gives us a fast, monotonic, cross-cpu clock
> > > for trace entries and can be used to correlate events across cpus as
> > > well as across hypervisor and guest (assuming it is not a migrated guest
> > > with a non-zero tb_offset).
> > > 
> > > Signed-off-by: Naveen N. Rao <naveen.n.rao at linux.vnet.ibm.com>
> > > ---
> > > I have followed the approach used by x86-tsc here, but we could get rid of
> > > trace_clock.c if we directly use get_tb() with perhaps the notrace annotation.
> > > Would that be preferable?
> > > 
> > 
> > Probably. But all clocks used by tracing should be marked by notrace.
> > Don't just wrap it with a notrace. But looking at the code, it seems
> > that get_tb() is a static inline, which wont work as a pointer. Seems
> > you still need the indirect function call.
> > 
> > Note, all "inline" functions are notrace by default, so you do not need
> > to add any notrace annotation to an inlined function.
> 
> Steve,
> Thanks for the clarification - the current approach is better in that 
> case.
> 
> Paul, Mike,
> Can you please let me know your thoughts on this?

What is the value in adding a powerpc specific clock, which requires educating
people to use it, vs just using the global clock?

cheers