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: <ecd8d769-4143-550c-13b5-ae96e6a76ff6@acm.org>
Date:   Tue, 30 May 2023 16:50:53 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.10.0
Subject: Re: [PATCH v2 1/1] scsi: ufs: core: Make UFS_MCQ_NUM_DEV_CMD_QUEUES a
 module parameter
Content-Language: en-US
To:     Po-Wen Kao <powen.kao@mediatek.com>, linux-scsi@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
        linux-mediatek@lists.infradead.org,
        Alim Akhtar <alim.akhtar@samsung.com>,
        Avri Altman <avri.altman@wdc.com>,
        "James E.J. Bottomley" <jejb@linux.ibm.com>,
        "Martin K. Petersen" <martin.petersen@oracle.com>,
        Matthias Brugger <matthias.bgg@gmail.com>,
        AngeloGioacchino Del Regno 
        <angelogioacchino.delregno@collabora.com>
Cc:     wsd_upstream@mediatek.com, peter.wang@mediatek.com,
        stanley.chu@mediatek.com, alice.chao@mediatek.com,
        naomi.chu@mediatek.com, chun-hung.wu@mediatek.com,
        cc.chou@mediatek.com, eddie.huang@mediatek.com
References: <20230530023518.16965-1-powen.kao@mediatek.com>
From:   Bart Van Assche <bvanassche@acm.org>
In-Reply-To: <20230530023518.16965-1-powen.kao@mediatek.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 5/29/23 19:35, Po-Wen Kao wrote:
> A dedicated queue for dev commands is not mandatory, hence let
> UFS_MCQ_NUM_DEV_CMD_QUEUES become module parameter `dev_cmd_queues`
> to allow sharing first hw queue for dev commands.
> 
> When `dev_cmd_queues` is set to 0, the hwq 0 will be shared for I/O
> requests and dev commands. In the same hwq, commands are processed based
> on submission order hence might take longer to dispatch dev command
> under heavy traffic. For the host with dedicated hwq for dev commands
> can benefit in such scenario.

The UFS core has already too many kernel module parameters. I don't 
think that we need more kernel module parameters. Instead of adding a 
new kernel module parameter, I propose to never reserve a hardware queue 
for device commands. Reserving a hardware queue for device commands 
seems wasteful to me.

Thanks,

Bart.