Re: [PATCH v4] trace,hardirq: No moar _rcuidle() tracing

[Masami Hiramatsu (Google) <mhiramat@kernel.org>]

On Tue, 17 Jan 2023 10:17:16 +0100
Ingo Molnar <mingo@kernel.org> wrote:

> 
> * Peter Zijlstra <peterz@infradead.org> wrote:
> 
> > On Tue, Jan 17, 2023 at 01:24:46PM +0900, Masami Hiramatsu wrote:
> > > Hi Peter,
> > > 
> > > On Thu, 12 Jan 2023 20:43:49 +0100
> > > Peter Zijlstra <peterz@infradead.org> wrote:
> > > 
> > > > Robot reported that trace_hardirqs_{on,off}() tickle the forbidden
> > > > _rcuidle() tracepoint through local_irq_{en,dis}able().
> > > > 
> > > > For 'sane' configs, these calls will only happen with RCU enabled and
> > > > as such can use the regular tracepoint. This also means it's possible
> > > > to trace them from NMI context again.
> > > > 
> > > > Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> > > 
> > > The code looks good to me. I just have a question about comment.
> > > 
> > > > ---
> > > >  kernel/trace/trace_preemptirq.c |   21 +++++++++++++--------
> > > >  1 file changed, 13 insertions(+), 8 deletions(-)
> > > > 
> > > > --- a/kernel/trace/trace_preemptirq.c
> > > > +++ b/kernel/trace/trace_preemptirq.c
> > > > @@ -20,6 +20,15 @@
> > > >  static DEFINE_PER_CPU(int, tracing_irq_cpu);
> > > >  
> > > >  /*
> > > > + * ...
> > > 
> > > Is this intended? Wouldn't you leave any comment here?
> > 
> > I indeed forgot to write the comment before posting, my bad :/ Ingo fixed
> > it up when he applied.
> 
> For completeness, I've attached the final commit, which has this comment 
> included:
> 
> +/*
> + * Use regular trace points on architectures that implement noinstr
> + * tooling: these calls will only happen with RCU enabled, which can
> + * use a regular tracepoint.
> + *
> + * On older architectures, use the rcuidle tracing methods (which
> + * aren't NMI-safe - so exclude NMI contexts):
> + */

Thanks! This looks good to me.

Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>

> 
> Thanks,
> 
> 	Ingo
> 
> ================>
> From: Peter Zijlstra <peterz@infradead.org>
> Date: Thu, 12 Jan 2023 20:43:49 +0100
> Subject: [PATCH] tracing, hardirq: No moar _rcuidle() tracing
> 
> Robot reported that trace_hardirqs_{on,off}() tickle the forbidden
> _rcuidle() tracepoint through local_irq_{en,dis}able().
> 
> For 'sane' configs, these calls will only happen with RCU enabled and
> as such can use the regular tracepoint. This also means it's possible
> to trace them from NMI context again.
> 
> Reported-by: kernel test robot <lkp@intel.com>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> Signed-off-by: Ingo Molnar <mingo@kernel.org>
> Link: https://lore.kernel.org/r/20230112195541.477416709@infradead.org
> ---
>  kernel/trace/trace_preemptirq.c | 26 ++++++++++++++++++--------
>  1 file changed, 18 insertions(+), 8 deletions(-)
> 
> diff --git a/kernel/trace/trace_preemptirq.c b/kernel/trace/trace_preemptirq.c
> index 629f2854e12b..f992444a0b1f 100644
> --- a/kernel/trace/trace_preemptirq.c
> +++ b/kernel/trace/trace_preemptirq.c
> @@ -19,6 +19,20 @@
>  /* Per-cpu variable to prevent redundant calls when IRQs already off */
>  static DEFINE_PER_CPU(int, tracing_irq_cpu);
>  
> +/*
> + * Use regular trace points on architectures that implement noinstr
> + * tooling: these calls will only happen with RCU enabled, which can
> + * use a regular tracepoint.
> + *
> + * On older architectures, use the rcuidle tracing methods (which
> + * aren't NMI-safe - so exclude NMI contexts):
> + */
> +#ifdef CONFIG_ARCH_WANTS_NO_INSTR
> +#define trace(point)	trace_##point
> +#else
> +#define trace(point)	if (!in_nmi()) trace_##point##_rcuidle
> +#endif
> +
>  /*
>   * Like trace_hardirqs_on() but without the lockdep invocation. This is
>   * used in the low level entry code where the ordering vs. RCU is important
> @@ -28,8 +42,7 @@ static DEFINE_PER_CPU(int, tracing_irq_cpu);
>  void trace_hardirqs_on_prepare(void)
>  {
>  	if (this_cpu_read(tracing_irq_cpu)) {
> -		if (!in_nmi())
> -			trace_irq_enable(CALLER_ADDR0, CALLER_ADDR1);
> +		trace(irq_enable)(CALLER_ADDR0, CALLER_ADDR1);
>  		tracer_hardirqs_on(CALLER_ADDR0, CALLER_ADDR1);
>  		this_cpu_write(tracing_irq_cpu, 0);
>  	}
> @@ -40,8 +53,7 @@ NOKPROBE_SYMBOL(trace_hardirqs_on_prepare);
>  void trace_hardirqs_on(void)
>  {
>  	if (this_cpu_read(tracing_irq_cpu)) {
> -		if (!in_nmi())
> -			trace_irq_enable_rcuidle(CALLER_ADDR0, CALLER_ADDR1);
> +		trace(irq_enable)(CALLER_ADDR0, CALLER_ADDR1);
>  		tracer_hardirqs_on(CALLER_ADDR0, CALLER_ADDR1);
>  		this_cpu_write(tracing_irq_cpu, 0);
>  	}
> @@ -63,8 +75,7 @@ void trace_hardirqs_off_finish(void)
>  	if (!this_cpu_read(tracing_irq_cpu)) {
>  		this_cpu_write(tracing_irq_cpu, 1);
>  		tracer_hardirqs_off(CALLER_ADDR0, CALLER_ADDR1);
> -		if (!in_nmi())
> -			trace_irq_disable(CALLER_ADDR0, CALLER_ADDR1);
> +		trace(irq_disable)(CALLER_ADDR0, CALLER_ADDR1);
>  	}
>  
>  }
> @@ -78,8 +89,7 @@ void trace_hardirqs_off(void)
>  	if (!this_cpu_read(tracing_irq_cpu)) {
>  		this_cpu_write(tracing_irq_cpu, 1);
>  		tracer_hardirqs_off(CALLER_ADDR0, CALLER_ADDR1);
> -		if (!in_nmi())
> -			trace_irq_disable_rcuidle(CALLER_ADDR0, CALLER_ADDR1);
> +		trace(irq_disable)(CALLER_ADDR0, CALLER_ADDR1);
>  	}
>  }
>  EXPORT_SYMBOL(trace_hardirqs_off);


-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>