Re: [PATCH 3/3] Stop tracing on a schedule bug

[Frederic Weisbecker <fweisbec@gmail.com>]

On Fri, Apr 16, 2010 at 12:46:53PM -0400, Steven Rostedt wrote:
> On Thu, 2010-04-15 at 21:01 -0700, Chase Douglas wrote:
> 
> > > 2) tracing off can be done via filters on functions and/or events
> > >   already - so I doubt that the tracing_off_event(level) is necessary
> > >   at all.
> > >
> > >   schedule_bug() definitely deserves a separate trace_schedule_bug()
> > >   event which can be used to stop the tracer by already existing
> > >   functionality.
> > 
> > Steven said he would be fine with a separate TRACE_EVENT_<blah> macro
> > for the schedule bug if needed, but I'm not sure we need to go that
> > far. If it's configurable through debugfs at run time then it serves
> > my purpose. Unless you feel we should have finer grained control
> > specifically for scheduling while atomic bugs, I'll just leave it as
> > TRACE_EVENT_WARN.
> 
> I actually like Thomas's idea better. I need to add the "stop trace on
> event" functionality, and we can insert trace events for bugs, and not
> have this whole "stop tracing here" functions. Instead we could just add
> tracepoints and have a way to pick and choose where to stop tracing.
> 
> add a:
> 
> include/trace/events/errors.h
> 
> #define TRACE_SYSTEM errors
> 
> TRACE_EVENT(sched_bug, ....)
> 
> etc,
> 


Looks good.


 
> When I get back home, I'll add this functionality to stop tracing on
> events. Perhaps I'll even add "TRACE_SUB_SYSTEM" so in the events
> directory, we can have sub layers:
> 
> events/errors/BUG/...
> events/errors/WARNING/...


I agree this could be nice. I'm thinking about the printk all around
the kernel that can be represented as trace events with sub-sub-systems
that could be file paths of the kernel.

I just hope we can find a way to do this that won't break tools.