Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed, 22 Sep 2021 16:23:41 -0700 (PDT)
From:   Hugh Dickins <hughd@google.com>
To:     Eric Dumazet <eric.dumazet@gmail.com>
cc:     Andrew Morton <akpm@linux-foundation.org>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        linux-mm <linux-mm@kvack.org>,
        Eric Dumazet <edumazet@google.com>,
        Hugh Dickins <hughd@google.com>
Subject: Re: [PATCH] mm: do not acquire zone lock in is_free_buddy_page()
In-Reply-To: <20210922152833.4023972-1-eric.dumazet@gmail.com>
Message-ID: <26db20dd-8e18-e68a-adb-adf171d62bf8@google.com>
References: <20210922152833.4023972-1-eric.dumazet@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Precedence: bulk

On Wed, 22 Sep 2021, Eric Dumazet wrote:
> From: Eric Dumazet <edumazet@google.com>
> 
> Grabbing zone lock in is_free_buddy_page() gives a wrong sense of safety,
> and has potential performance implications when zone is experiencing
> lock contention.
> 
> In any case, if a caller needs a stable result, it should grab zone
> lock before calling this function.
> 
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Cc: Hugh Dickins <hughd@google.com>

Yes indeed, and you have already explained it well above: thanks.

Acked-by: Hugh Dickins <hughd@google.com>

> ---
>  mm/page_alloc.c | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index e115e21524739341d409b28379942241ed403060..cd8a72372b047e55c4cde80fe6b7a428d7721027 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -9354,21 +9354,21 @@ void __offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn)
>  }
>  #endif
>  
> +/*
> + * This function returns a stable result only if called under zone lock.
> + */
>  bool is_free_buddy_page(struct page *page)
>  {
> -	struct zone *zone = page_zone(page);
>  	unsigned long pfn = page_to_pfn(page);
> -	unsigned long flags;
>  	unsigned int order;
>  
> -	spin_lock_irqsave(&zone->lock, flags);
>  	for (order = 0; order < MAX_ORDER; order++) {
>  		struct page *page_head = page - (pfn & ((1 << order) - 1));
>  
> -		if (PageBuddy(page_head) && buddy_order(page_head) >= order)
> +		if (PageBuddy(page_head) &&
> +		    buddy_order_unsafe(page_head) >= order)
>  			break;
>  	}
> -	spin_unlock_irqrestore(&zone->lock, flags);
>  
>  	return order < MAX_ORDER;
>  }
> -- 
> 2.33.0.464.g1972c5931b-goog