Re: [PATCH for-next v4 05/12] RDMA/erdma: Add cmdq implementation

From: Wenpeng Liang <liangwenpeng@huawei.com>
To: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Cheng Xu <chengyou@linux.alibaba.com>, <dledford@redhat.com>,
	<leon@kernel.org>, <linux-rdma@vger.kernel.org>,
	<KaiShen@linux.alibaba.com>, <tonylu@linux.alibaba.com>,
	<BMT@zurich.ibm.com>
Subject: Re: [PATCH for-next v4 05/12] RDMA/erdma: Add cmdq implementation
Date: Sat, 19 Mar 2022 09:26:14 +0800	[thread overview]
Message-ID: <7e72c0db-27fb-45bb-b0a6-a9f6a20a3b77@huawei.com> (raw)
In-Reply-To: <20220318181746.GF64706@ziepe.ca>

On 2022/3/19 2:17, Jason Gunthorpe wrote:
> On Fri, Mar 18, 2022 at 07:16:08PM +0800, Wenpeng Liang wrote:
>> On 2022/3/14 14:47, Cheng Xu wrote:
>> <...>
>>> +int erdma_cmdq_init(struct erdma_dev *dev)
>>> +{
>>> +	int err, i;
>>> +	struct erdma_cmdq *cmdq = &dev->cmdq;
>>> +	u32 status, ctrl;
>>
>> Hi, Jason and Leon
>>
>> Defining and initializing variables in the form of an inverted triangle at
>> the head of the function can make the code clearer. The kernel's coding-style
>> does not specify this behavior, and the various kernel subsystems do not seem
>> to have formed a unified opinion. Does our RDMA subsystem recommend this?
>>
>> struct erdma_cmdq *cmdq = &dev->cmdq;
>> u32 status, ctrl;
>> int err, i;
> 
> This is often called reverse christmas tree style
> 
> It is common in RDMA, but I would not insist on it.

Thanks for your reply and correction.

Thanks,
Wenpeng

> 
> Thanks,
> Jason
> .
> 