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;
Message-ID: <578db833-e20a-cec8-c00f-5cd1e8959ccf@gmail.com>
Date:   Thu, 27 Apr 2023 19:19:35 +0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0)
 Gecko/20100101 Thunderbird/102.10.1
Subject: Re: [PATCH 25/34] maple_tree: Clear up index and last setting in
 single entry tree
To:     "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        maple-tree@lists.infradead.org,
        Andrew Morton <akpm@linux-foundation.org>
References: <20230425140955.3834476-1-Liam.Howlett@oracle.com>
 <20230425140955.3834476-26-Liam.Howlett@oracle.com>
From:   Peng Zhang <perlyzhang@gmail.com>
In-Reply-To: <20230425140955.3834476-26-Liam.Howlett@oracle.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk


在 2023/4/25 22:09, Liam R. Howlett 写道:
> When there is a single entry tree (range of 0-0 pointing to an entry),
> then ensure the limit is either 0-0 or 1-oo, depending on where the user
> walks.  Ensure the correct node setting as well; either MAS_ROOT or
> MAS_NONE.
> 
> Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
> ---
>   lib/maple_tree.c | 21 +++++++++++----------
>   1 file changed, 11 insertions(+), 10 deletions(-)
> 
> diff --git a/lib/maple_tree.c b/lib/maple_tree.c
> index 20f0a10dc5608..31cbfd7b44728 100644
> --- a/lib/maple_tree.c
> +++ b/lib/maple_tree.c
> @@ -5099,24 +5099,25 @@ void *mas_walk(struct ma_state *mas)
>   {
>   	void *entry;
>   
> +	if (mas_is_none(mas) || mas_is_paused(mas))
> +		mas->node = MAS_START;
>   retry:
>   	entry = mas_state_walk(mas);
> -	if (mas_is_start(mas))
> +	if (mas_is_start(mas)) {
>   		goto retry;
> -
> -	if (mas_is_ptr(mas)) {
> +	} else if (mas_is_none(mas)) {
> +		mas->index = 0;
> +		mas->last = ULONG_MAX;
> +	} else if (mas_is_ptr(mas)) {
>   		if (!mas->index) {
>   			mas->last = 0;
> -		} else {
> -			mas->index = 1;
> -			mas->last = ULONG_MAX;
> +			return mas_root(mas);
Why we call mas_root() to get the single entry stored in root again?
I think it's not safe. In RCU mode, if someone modify the tree to a 
normal tree(not a single entry tree), mas_root() will return a address.
So, this may cause a race bug. We can return entry directly.
>   		}
> -		return entry;
> -	}
>   
> -	if (mas_is_none(mas)) {
> -		mas->index = 0;
> +		mas->index = 1;
>   		mas->last = ULONG_MAX;
> +		mas->node = MAS_NONE;
> +		return NULL;
>   	}
>   
>   	return entry;