Messages in this thread | | | From | Joel Fernandes <> | Date | Sun, 10 Sep 2017 15:21:23 -0700 | Subject | Re: [PATCH RFC v4 1/3] tracing/irqsoff: Prepare to add preempt and irq trace events |
| |
Hi Steven,
On Sun, Sep 10, 2017 at 2:06 AM, Steven Rostedt <rostedt@goodmis.org> wrote: > On Wed, 6 Sep 2017 17:17:50 -0700 > Joel Fernandes <joelaf@google.com> wrote: > > >> +#else /* IRQSOFF_TRACER || PREEMPTOFF_TRACER */ >> + >> +#define start_critical_timings_tracer() do { } while (0) >> +#define stop_critical_timings_tracer() do { } while (0) >> + >> +#endif >> + >> +#ifndef CONFIG_IRQSOFF_TRACER >> +#define tracer_hardirqs_on() do { } while (0) >> +#define tracer_hardirqs_off() do { } while (0) >> +#define tracer_hardirqs_on_caller(x) do { } while (0) >> +#define tracer_hardirqs_off_caller(x) do { } while (0) >> +#endif >> + >> +#ifndef CONFIG_PREEMPT_TRACER >> +#define tracer_preempt_on(x, y) do { } while (0) >> +#define tracer_preempt_off(x, y) do { } while (0) >> +#endif > > Is there a reason the above is #define and not static inline? The > preferred method is static inline as that allows the compiler to check > types. #define is used if one of the parameters has a struct or > something that is not defined.
Sure, I'll make use of static inline in the next version. thanks!
-Joel
| |