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;
Received-SPF: Pass (protection.outlook.com: domain of nvidia.com designates
 216.228.112.32 as permitted sender) receiver=protection.outlook.com;
 client-ip=216.228.112.32; helo=mail.nvidia.com;
Subject: Re: [PATCH rdma-next v1 05/11] RDMA/counter: Add optional counter
 support
To:     Jason Gunthorpe <jgg@nvidia.com>, Leon Romanovsky <leon@kernel.org>
CC:     Doug Ledford <dledford@redhat.com>,
        Aharon Landau <aharonl@nvidia.com>,
        "David S. Miller" <davem@davemloft.net>,
        Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>,
        Gal Pressman <galpress@amazon.com>,
        Jakub Kicinski <kuba@kernel.org>,
        <linux-kernel@vger.kernel.org>, <linux-rdma@vger.kernel.org>,
        Maor Gottlieb <maorg@nvidia.com>,
        Mike Marciniszyn <mike.marciniszyn@cornelisnetworks.com>,
        Mustafa Ismail <mustafa.ismail@intel.com>,
        Naresh Kumar PBS <nareshkumar.pbs@broadcom.com>,
        Neta Ostrovsky <netao@nvidia.com>, <netdev@vger.kernel.org>,
        Potnuri Bharat Teja <bharat@chelsio.com>,
        "Saeed Mahameed" <saeedm@nvidia.com>,
        Selvin Xavier <selvin.xavier@broadcom.com>,
        Shiraz Saleem <shiraz.saleem@intel.com>,
        Yishai Hadas <yishaih@nvidia.com>,
        Zhu Yanjun <zyjzyj2000@gmail.com>
References: <cover.1631660727.git.leonro@nvidia.com>
 <04bd7354c6a375e684712d79915f7eb816efee92.1631660727.git.leonro@nvidia.com>
 <20210927170318.GB1529966@nvidia.com>
From:   Mark Zhang <markzhang@nvidia.com>
Message-ID: <d9cd401a-b5fe-65ea-21c4-6d4c037fd641@nvidia.com>
Date:   Tue, 28 Sep 2021 17:03:24 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Thunderbird/78.14.0
MIME-Version: 1.0
In-Reply-To: <20210927170318.GB1529966@nvidia.com>
Content-Type: text/plain; charset="utf-8"; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 28 Sep 2021 09:03:33.5740
 (UTC)
X-MS-Exchange-CrossTenant-Network-Message-Id: cd7121ad-9d75-4d4f-dbf9-08d9825ed97a
X-MS-Exchange-CrossTenant-Id: 43083d15-7273-40c1-b7db-39efd9ccc17a
X-MS-Exchange-CrossTenant-OriginalAttributedTenantConnectingIp: TenantId=43083d15-7273-40c1-b7db-39efd9ccc17a;Ip=[216.228.112.32];Helo=[mail.nvidia.com]
X-MS-Exchange-CrossTenant-AuthSource: CO1NAM11FT029.eop-nam11.prod.protection.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Anonymous
X-MS-Exchange-CrossTenant-FromEntityHeader: HybridOnPrem
X-MS-Exchange-Transport-CrossTenantHeadersStamped: BYAPR12MB3079
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 9/28/2021 1:03 AM, Jason Gunthorpe wrote:
> On Wed, Sep 15, 2021 at 02:07:24AM +0300, Leon Romanovsky wrote:
>>   
>> +int rdma_counter_modify(struct ib_device *dev, u32 port, int index, bool enable)
>> +{
>> +	struct rdma_hw_stats *stats;
>> +	int ret;
>> +
>> +	if (!dev->ops.modify_hw_stat)
>> +		return -EOPNOTSUPP;
>> +
>> +	stats = ib_get_hw_stats_port(dev, port);
>> +	if (!stats)
>> +		return -EINVAL;
>> +
>> +	mutex_lock(&stats->lock);
>> +	ret = dev->ops.modify_hw_stat(dev, port, index, enable);
>> +	if (!ret)
>> +		enable ? clear_bit(index, stats->is_disabled) :
>> +			set_bit(index, stats->is_disabled);
> 
> This is not a kernel coding style write out the if, use success
> oriented flow
> 
> Also, shouldn't this logic protect the driver from being called on
> non-optional counters?

We leave it to driver, driver would return failure if modify is not 
supported. Is it good?

>>   	for (i = 0; i < data->stats->num_counters; i++) {
>> -		attr = &data->attrs[i];
>> +		if (data->stats->descs[i].flags & IB_STAT_FLAG_OPTIONAL)
>> +			continue;
>> +		attr = &data->attrs[pos];
>>   		sysfs_attr_init(&attr->attr.attr);
>>   		attr->attr.attr.name = data->stats->descs[i].name;
>>   		attr->attr.attr.mode = 0444;
>>   		attr->attr.show = hw_stat_device_show;
>>   		attr->show = show_hw_stats;
>> -		data->group.attrs[i] = &attr->attr.attr;
>> +		data->group.attrs[pos] = &attr->attr.attr;
>> +		pos++;
>>   	}
> 
> This isn't OK, the hw_stat_device_show() computes the stat index like
> this:
> 
> 	return stat_attr->show(ibdev, ibdev->hw_stats_data->stats,
> 			       stat_attr - ibdev->hw_stats_data->attrs, 0, buf);
> 
> Which assumes the stats are packed contiguously. This only works
> because mlx5 is always putting the optional stats at the end.

Yes you are right, thanks. Maybe we can add an "index" field in struct 
hw_stats_device/port_attribute, then set it in setup and use it in show.