Received-SPF: pass (google.com: domain of linux-kernel+bounces-71295-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
Message-ID: <99e3e994-ef6a-4339-abf2-cda62d24b1ce@linux.dev>
Date: Mon, 19 Feb 2024 20:19:23 +0800
Precedence: bulk
MIME-Version: 1.0
Subject: Re: [PATCH] net/mlx5: fix possible stack overflows
To: Hamdan Agbariya <hamdani@nvidia.com>, Arnd Bergmann <arnd@arndb.de>,
 Arnd Bergmann <arnd@kernel.org>, Saeed Mahameed <saeedm@nvidia.com>,
 Leon Romanovsky <leon@kernel.org>
Cc: "David S . Miller" <davem@davemloft.net>,
 Eric Dumazet <edumazet@google.com>, Jakub Kicinski <kuba@kernel.org>,
 Paolo Abeni <pabeni@redhat.com>, Yevgeny Kliteynik <kliteyn@nvidia.com>,
 Alex Vesker <valex@nvidia.com>, Netdev <netdev@vger.kernel.org>,
 "linux-rdma@vger.kernel.org" <linux-rdma@vger.kernel.org>,
 "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
References: <20240213100848.458819-1-arnd@kernel.org>
 <84874528-daea-424d-af63-b9b86835fae6@linux.dev>
 <2ebe5a36-ce81-4d26-a12b-7affbd65c5e3@app.fastmail.com>
 <11f40993-ec02-48b7-aec5-13ff7cddf665@linux.dev>
 <DM6PR12MB45168A0957212864D8D53B80CE512@DM6PR12MB4516.namprd12.prod.outlook.com>
From: Zhu Yanjun <yanjun.zhu@linux.dev>
In-Reply-To: <DM6PR12MB45168A0957212864D8D53B80CE512@DM6PR12MB4516.namprd12.prod.outlook.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit

在 2024/2/19 17:05, Hamdan Agbariya 写道:
>> 在 2024/2/15 16:03, Arnd Bergmann 写道:
>>> On Thu, Feb 15, 2024, at 01:18, Zhu Yanjun wrote:
>>>> 在 2024/2/13 18:08, Arnd Bergmann 写道:
>>>>>     static int
>>>>> -dr_dump_rule_rx_tx(struct seq_file *file, struct mlx5dr_rule_rx_tx
>>>>> *rule_rx_tx,
>>>>> +dr_dump_rule_rx_tx(struct seq_file *file, char *buff,
>>>>> +              struct mlx5dr_rule_rx_tx *rule_rx_tx,
>>>>>                 bool is_rx, const u64 rule_id, u8 format_ver)
>>>>>     {
>>>>>      struct mlx5dr_ste *ste_arr[DR_RULE_MAX_STES +
>>>>> DR_ACTION_MAX_STES]; @@ -533,7 +532,7 @@
>> dr_dump_rule_rx_tx(struct seq_file *file, struct mlx5dr_rule_rx_tx
>> *rule_rx_tx,
>>>>>              return 0;
>>>>>
>>>>>      while (i--) {
>>>>> -           ret = dr_dump_rule_mem(file, ste_arr[i], is_rx, rule_id,
>>>> Before buff is reused, I am not sure whether buff should be firstly
>>>> zeroed or not.
>>> I don't see why it would, but if you want to zero it, that would be a
>>> separate patch that is already needed on the existing code, which
>>> never zeroes its buffers.
>>
>> Sure. I agree with you. In the existing code, the buffers are not zeroed.
>>
>> But to a buffer which is used for several times, it is good to zero it before it is
>> used again.
>>
>> Can you add a new commit with the following?
>>
>> 1). Zero the buffers in the existing code
>>
> 
> No need to zero the buffers, as it does not have any necessity and it will only affect performance.
> Thanks,

Sorry. I can not get your point. Can you explain why no need to zero the 
buffers? Thanks in advance.

> Hamdan
> 
> 
> 
> 
>> 2). Add the zero functionality to your patch

If a buffer is used for many times, is it necessary to zero it before it 
is used again?

Thanks,
Zhu Yanjun

>>
>>   From my perspective, it is good to the whole commit.
>>
>> Please Jason and Leon comment on this.
>>
>> Thanks,
>>
>> Zhu Yanjun
>>
>>>
>>>       Arnd