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:   Mon, 3 Apr 2023 19:24:32 -0700
From:   Yury Norov <yury.norov@gmail.com>
To:     Ye Bin <yebin@huaweicloud.com>
Cc:     dennis@kernel.org, tj@kernel.org, cl@linux.com, linux-mm@kvack.org,
        andriy.shevchenko@linux.intel.com, linux@rasmusvillemoes.dk,
        linux-kernel@vger.kernel.org, dchinner@redhat.com,
        yebin10@huawei.com
Subject: Re: [PATCH 1/2] cpu/hotplug: introduce 'num_dying_cpus' to get dying
 CPUs count
Message-ID: <ZCuKYN2ZUJjn752J@yury-laptop>
References: <20230404014206.3752945-1-yebin@huaweicloud.com>
 <20230404014206.3752945-2-yebin@huaweicloud.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230404014206.3752945-2-yebin@huaweicloud.com>
Precedence: bulk

On Tue, Apr 04, 2023 at 09:42:05AM +0800, Ye Bin wrote:
> From: Ye Bin <yebin10@huawei.com>
> 
> Introduce '__num_dying_cpus' variable to cache the number of dying CPUs
> in the core and just return the cached variable.
> 
> Signed-off-by: Ye Bin <yebin10@huawei.com>
> ---
>  include/linux/cpumask.h | 20 ++++++++++++++++----
>  kernel/cpu.c            |  2 ++
>  2 files changed, 18 insertions(+), 4 deletions(-)
> 
> diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h
> index 2a61ddcf8321..8127fd598f51 100644
> --- a/include/linux/cpumask.h
> +++ b/include/linux/cpumask.h
> @@ -135,6 +135,8 @@ extern struct cpumask __cpu_dying_mask;
>  
>  extern atomic_t __num_online_cpus;
>  
> +extern atomic_t __num_dying_cpus;
> +
>  extern cpumask_t cpus_booted_once_mask;
>  
>  static __always_inline void cpu_max_bits_warn(unsigned int cpu, unsigned int bits)
> @@ -1018,10 +1020,14 @@ set_cpu_active(unsigned int cpu, bool active)
>  static __always_inline void
>  set_cpu_dying(unsigned int cpu, bool dying)
>  {
> -	if (dying)
> -		cpumask_set_cpu(cpu, &__cpu_dying_mask);
> -	else
> -		cpumask_clear_cpu(cpu, &__cpu_dying_mask);
> +	if (dying) {
> +		if (!cpumask_test_and_set_cpu(cpu, &__cpu_dying_mask))
> +			atomic_inc(&__num_dying_cpus);
> +	}
> +	else {
> +		if (cpumask_test_and_clear_cpu(cpu, &__cpu_dying_mask))
> +			atomic_dec(&__num_dying_cpus);
> +	}
>  }

Corresponding set_cpu_online() is implemented in C-file probably for a
reason. Are you sure that similar function for dying mask should
reside in a header? If so, can you share your reasoning?

Regardless, now that you added the identical function to
set_cpu_online, I think it's worth to make it a general approach:

void set_cpu_counted(unsigned int cpu, bool set,
                        struct cpumask *mask, atomic_t *cnt);

void __always_inline set_cpu_online(unsigned int cpu, bool online)
{
      set_cpu_counted(cpu, online, &__cpu_online_mask, &__num_online_cpus);  
}

void __always_inline set_cpu_dying(unsigned int cpu, bool dying)
{
      set_cpu_counted(cpu, dying, &__cpu_dying_mask, &__num_dying_cpus);  
}