Date:   Sun, 12 Mar 2023 13:25:24 -0700
From:   Andrew Morton <akpm@linux-foundation.org>
To:     Yair Podemsky <ypodemsk@redhat.com>
Cc:     will@kernel.org, aneesh.kumar@linux.ibm.com, npiggin@gmail.com,
        peterz@infradead.org, arnd@arndb.de, linux-arch@vger.kernel.org,
        linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        mtosatti@redhat.com, ppandit@redhat.com, alougovs@redhat.com,
        David Hildenbrand <david@redhat.com>
Subject: Re: [PATCH] mm/mmu_gather: send tlb_remove_table_smp_sync IPI only
 to MM CPUs
Message-Id: <20230312132524.c37abfffe2c70eeda20f3217@linux-foundation.org>
In-Reply-To: <20230312080945.14171-1-ypodemsk@redhat.com>
References: <20230312080945.14171-1-ypodemsk@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Sun, 12 Mar 2023 10:09:45 +0200 Yair Podemsky <ypodemsk@redhat.com> wrote:

> Currently the tlb_remove_table_smp_sync IPI is sent to all CPUs
> indiscriminately, this causes unnecessary work and delays notable in
> real-time use-cases and isolated cpus, this patch will limit this IPI to
> only be sent to cpus referencing the effected mm and are currently in
> kernel space.
> 
> ...
>
> --- a/mm/mmu_gather.c
> +++ b/mm/mmu_gather.c
> @@ -191,7 +192,15 @@ static void tlb_remove_table_smp_sync(void *arg)
>  	/* Simply deliver the interrupt */
>  }
>  
> -void tlb_remove_table_sync_one(void)
> +static bool cpu_in_kernel(int cpu, void *info)
> +{
> +	struct context_tracking *ct = per_cpu_ptr(&context_tracking, cpu);
> +	int statue = atomic_read(&ct->state);

Strange identifier.  Should be "state"?

> +	//will return true only for cpu's in kernel space

Please use /* */ style comments

And use "cpus" rather than "cpu's" - plural, not possessive.

> +	return !(statue & CT_STATE_MASK);

Using

	return state & CT_STATE_MASK == CONTEXT_KERNEL;

would more clearly express the intent.

> +}

And...  surely this function is racy.  ct->state can change value one
nanosecond after cpu_in_kernel() reads it, so cpu_in_kernel()'s return
value is now wrong?