Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1619013

Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for trace_*_rcuidle() trace events

From Steven Rostedt <rostedt@goodmis.org>
Newsgroups linux.kernel
Subject Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for trace_*_rcuidle() trace events
Date 2017-04-07 20:00 +0200
Message-ID <ttGkW-28n-13@gated-at.bofh.it> (permalink)
References <ttCKl-8a9-13@gated-at.bofh.it> <ttFyx-1Kj-3@gated-at.bofh.it> <ttFIe-1OF-15@gated-at.bofh.it> <ttFRT-1SP-1@gated-at.bofh.it> <ttGbg-21h-19@gated-at.bofh.it>
Organization linux.* mail to news gateway

Show all headers | View raw


On Fri, 7 Apr 2017 17:49:17 +0000 (UTC)
Mathieu Desnoyers <mathieu.desnoyers@efficios.com> wrote:

> > Welcome to MACRO MAGIC!

Somebody is not wizardly happy.

> >   
> >> 
> >> as one argument to the __DO_TRACE() macro. To me it's a bit unexpected
> >> coding-style wise. Am I the only one not comfortable with the proposed
> >> syntax ?  
> > 
> > The entire TRACE_EVENT()/__DO_TRACE() is special.
> > 
> > I thought about add yet another parameter, but as it doesn't change
> > much, I figured this was good enough. We could beak it up if you like:
> > 
> > #define RCU_IRQ_ENTER_CHECK \
> >	if (WARN_ON_ONCE(rcu_irq_enter_disabled()) 	\
> >		return;					\
> >	rcu_irq_enter_irqson();
> > 
> > [..]
> >			__DO_TRACE(&__tracepoint_##name,		\
> >				TP_PROTO(data_proto),			\
> >				TP_ARGS(data_args),			\
> >				TP_CONDITION(cond),			\
> >				PARAMS(RCU_IRQ_ENTER_CHECK),		\
> >				rcu_irq_exit_irqson());			\
> > 
> > 
> > Would that make you feel more comfortable?  
> 
> No, it's almost worse and adds still adds a return that apply within __DO_TRACE(),
> but which is passed as an argument (code as macro argument), which I find really
> unsettling.

/me finds it strangely enjoyable to make Mathieu unsettled.

> 
> I would prefer to add a new argument to __DO_TRACE, which we can call
> "checkrcu", e.g.:
> 
> #define __DO_TRACE(tp, proto, args, cond, checkrcu, prercu, postrcu)    \

Grumble. I was trying to avoid making the patch more intrusive. But I
do understand your concern.

>         do {                                                            \
>                 struct tracepoint_func *it_func_ptr;                    \
>                 void *it_func;                                          \
>                 void *__data;                                           \
>                                                                         \
>                 if (!((cond) && (checkrcu)))                            \
>                         return;                                         \
>                 prercu;                                                 \
>                 rcu_read_lock_sched_notrace();                          \
>                 it_func_ptr = rcu_dereference_sched((tp)->funcs);       \
>                 if (it_func_ptr) {                                      \
>                         do {                                            \
>                                 it_func = (it_func_ptr)->func;          \
>                                 __data = (it_func_ptr)->data;           \
>                                 ((void(*)(proto))(it_func))(args);      \
>                         } while ((++it_func_ptr)->func);                \
>                 }                                                       \
>                 rcu_read_unlock_sched_notrace();                        \
>                 postrcu;                                                \
>         } while (0)
> 
> And use it like this:
> 
> #define __DECLARE_TRACE_RCU(name, proto, args, cond, data_proto, data_args)     \
>         static inline void trace_##name##_rcuidle(proto)                \
>         {                                                               \
>                 if (static_key_false(&__tracepoint_##name.key))         \
>                         __DO_TRACE(&__tracepoint_##name,                \
>                                 TP_PROTO(data_proto),                   \
>                                 TP_ARGS(data_args),                     \
>                                 TP_CONDITION(cond),                     \
>                                 !WARN_ON_ONCE(rcu_irq_enter_disabled()),\
>                                 rcu_irq_enter_irqson(),                 \
>                                 rcu_irq_exit_irqson());                 \
>         }
> 
> This way we only pass evaluated expression (not code with "return" that
> changes the flow) as arguments to __DO_TRACE, which makes it behave more
> like a "sub-function", which is what we usually expect.

I understand what you are getting at, and I will concede your point.
OK, I'll do it your way, but I still think you take all the fun out of
it. ;-)

-- Steve

Back to linux.kernel | Previous | NextPrevious in thread | Next in thread | Find similar | Unroll thread


Thread

[PATCH 0/5 v2] tracing: Add usecase of synchronize_rcu_tasks() and stack_tracer_disable() Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 16:10 +0200
  [PATCH 1/5 v2] ftrace: Add use of synchronize_rcu_tasks() with dynamic trampolines Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 16:10 +0200
  [PATCH 3/5 v2] tracing: Add stack_tracer_disable/enable() functions Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 16:10 +0200
    [PATCH 3/5 v2.1] tracing: Add stack_tracer_disable/enable()  functions Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 16:30 +0200
    Re: [PATCH 3/5 v2] tracing: Add stack_tracer_disable/enable()  functions Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 16:30 +0200
  Re: [PATCH 0/5 v2] tracing: Add usecase of synchronize_rcu_tasks()  and stack_tracer_disable() "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-04-07 16:50 +0200
    Re: [PATCH 0/5 v2] tracing: Add usecase of synchronize_rcu_tasks()  and stack_tracer_disable() Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 17:10 +0200
      Re: [PATCH 0/5 v2] tracing: Add usecase of synchronize_rcu_tasks()  and stack_tracer_disable() "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-04-07 17:20 +0200
        Re: [PATCH 0/5 v2] tracing: Add usecase of synchronize_rcu_tasks()  and stack_tracer_disable() Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 17:30 +0200
  [PATCH 6/5]rcu/tracing: Add rcu_disabled to denote when  rcu_irq_enter() will not work Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 18:40 +0200
    Re: [PATCH 6/5]rcu/tracing: Add rcu_disabled to denote when  rcu_irq_enter() will not work "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-04-07 18:50 +0200
      Re: [PATCH 6/5]rcu/tracing: Add rcu_disabled to denote when  rcu_irq_enter() will not work Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 18:50 +0200
        Re: [PATCH 6/5]rcu/tracing: Add rcu_disabled to denote when  rcu_irq_enter() will not work "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-04-07 19:00 +0200
      [PATCH 6/5 v2] rcu/tracing: Add rcu_disabled to denote when  rcu_irq_enter() will not work Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 19:10 +0200
        Re: [PATCH 6/5 v2] rcu/tracing: Add rcu_disabled to denote when  rcu_irq_enter() will not work "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-04-07 19:20 +0200
  [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 19:10 +0200
    Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-04-07 19:20 +0200
    Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Mathieu Desnoyers <mathieu.desnoyers@efficios.com> - 2017-04-07 19:20 +0200
      Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 19:30 +0200
        Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 19:40 +0200
        Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Mathieu Desnoyers <mathieu.desnoyers@efficios.com> - 2017-04-07 19:50 +0200
          Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 20:00 +0200
          [PATCH 7/5 v3] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 20:20 +0200
            Re: [PATCH 7/5 v3] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Mathieu Desnoyers <mathieu.desnoyers@efficios.com> - 2017-04-07 20:20 +0200
              Re: [PATCH 7/5 v4] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 21:50 +0200
              [PATCH 7/5 v4] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 21:50 +0200
      Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 19:30 +0200
      [PATCH 7/5 v2] tracing: Make sure rcu_irq_enter() can work for  trace_*_rcuidle() trace events Steven Rostedt <rostedt@goodmis.org> - 2017-04-07 19:50 +0200

csiph-web