Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 14 Apr 2022 11:32:04 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Zqiang <qiang1.zhang@intel.com>
Cc:     akpm@linux-foundation.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] irq_work: Add event-tracing points for irq_work
Message-ID: <YlfqFJLau+7PlPJ3@hirez.programming.kicks-ass.net>
References: <20220414063829.2472251-1-qiang1.zhang@intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220414063829.2472251-1-qiang1.zhang@intel.com>
Precedence: bulk

On Thu, Apr 14, 2022 at 02:38:29PM +0800, Zqiang wrote:
> Add irq_work_queue_on tracepoints allow tracing when and how
> a irq-work is queued, irq_work_execute_start/end tracepoints
> allow know when a irq-work is executed and the executed time.

Yeah, but why?

> diff --git a/kernel/irq_work.c b/kernel/irq_work.c
> index 7afa40fe5cc4..edad992556d0 100644
> --- a/kernel/irq_work.c
> +++ b/kernel/irq_work.c
> @@ -22,6 +22,9 @@
>  #include <asm/processor.h>
>  #include <linux/kasan.h>
>  
> +#define CREATE_TRACE_POINTS
> +#include <trace/events/irq_work.h>
> +
>  static DEFINE_PER_CPU(struct llist_head, raised_list);
>  static DEFINE_PER_CPU(struct llist_head, lazy_list);
>  static DEFINE_PER_CPU(struct task_struct *, irq_workd);
> @@ -81,7 +84,9 @@ static void __irq_work_queue_local(struct irq_work *work)
>  	bool rt_lazy_work = false;
>  	bool lazy_work = false;
>  	int work_flags;
> +	int cpu = smp_processor_id();
>  
> +	trace_irq_work_queue_on(cpu, work);

That's not very nice, you made this function more expensive for no
reason. And you violated coding style while doing it :/

>  	work_flags = atomic_read(&work->node.a_flags);
>  	if (work_flags & IRQ_WORK_LAZY)
>  		lazy_work = true;
> @@ -143,7 +148,7 @@ bool irq_work_queue_on(struct irq_work *work, int cpu)
>  	if (cpu != smp_processor_id()) {
>  		/* Arch remote IPI send/receive backend aren't NMI safe */
>  		WARN_ON_ONCE(in_nmi());
> -
> +		trace_irq_work_queue_on(cpu, work);
>  		/*
>  		 * On PREEMPT_RT the items which are not marked as
>  		 * IRQ_WORK_HARD_IRQ are added to the lazy list and a HARD work
> @@ -208,7 +213,9 @@ void irq_work_single(void *arg)
>  	smp_mb();
>  
>  	lockdep_irq_work_enter(flags);
> +	trace_irq_work_execute_start(work);
>  	work->func(work);
> +	trace_irq_work_execute_end(work);
>  	lockdep_irq_work_exit(flags);
>  
>  	/*