Re: [PATCH] PM: trace events for suspend/resume

["Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com>]

On 05/23/2014 04:14 AM, Steven Rostedt wrote:
> On Fri, 2014-05-23 at 03:46 +0530, Srivatsa S. Bhat wrote:
>> Hi Todd,
>>
>> This might be a silly question, but the suspend_enter() function invokes
>> ftrace_stop() before suspending the machine, and restarts ftrace after
>> resume. And your patch seems to instrument code further down (deeper)
>> in the suspend/resume path (such as disable_nonboot_cpus() for example).
>> Doesn't the ftrace stop/start pose any problem for this patch?
> 
> ftrace_start/stop() only affects the function tracer and not the
> tracepoints. Those will still be recorded.
> 

Ah, thank you for the clarification.

> Note, we have an invested interest in getting rid of
> ftrace_stop/start(). I rather mark all the suspend code with notrace or
> find another method to protect suspend from function tracing. The
> ftrace_stop/start() was more of a big hammer approach and I had plans
> for getting rid of it in the future. That future is approaching fast :-)
>

Ok.. :-)

 
Regards,
Srivatsa S. Bhat