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: <87a78633-5c9e-49c2-4e8a-8b1c6b21f2f0@acm.org>
Date:   Mon, 31 Oct 2022 12:42:15 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.3.0
Subject: Re: [PATCH v3 07/17] ufs: core: mcq: Calculate queue depth
Content-Language: en-US
To:     Asutosh Das <quic_asutoshd@quicinc.com>
Cc:     quic_cang@quicinc.com, martin.petersen@oracle.com,
        linux-scsi@vger.kernel.org, quic_nguyenb@quicinc.com,
        quic_xiaosenh@quicinc.com, stanley.chu@mediatek.com,
        eddie.huang@mediatek.com, daejun7.park@samsung.com,
        avri.altman@wdc.com, mani@kernel.org, beanhuo@micron.com,
        quic_richardp@quicinc.com, linux-arm-msm@vger.kernel.org,
        Alim Akhtar <alim.akhtar@samsung.com>,
        "James E.J. Bottomley" <jejb@linux.ibm.com>,
        Andy Gross <agross@kernel.org>,
        Bjorn Andersson <bjorn.andersson@linaro.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>,
        Jinyoung Choi <j-young.choi@samsung.com>,
        Kiwoong Kim <kwmad.kim@samsung.com>,
        open list <linux-kernel@vger.kernel.org>
References: <cover.1666288432.git.quic_asutoshd@quicinc.com>
 <1987fbada1d33c04c9598614ef712e0a48fe065e.1666288432.git.quic_asutoshd@quicinc.com>
 <0fb3f8ae-5ed7-9057-0d2b-8866f36c2441@acm.org>
 <20221031192435.GD9077@asutoshd-linux1.qualcomm.com>
From:   Bart Van Assche <bvanassche@acm.org>
In-Reply-To: <20221031192435.GD9077@asutoshd-linux1.qualcomm.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

On 10/31/22 12:24, Asutosh Das wrote:
> On Thu, Oct 27 2022 at 14:52 -0700, Bart Van Assche wrote:
>> On 10/20/22 11:03, Asutosh Das wrote:
>>> +u32 ufshcd_mcq_decide_queue_depth(struct ufs_hba *hba)
>>> +{
>>> +    u32 qd, val;
>>> +    int mac;
>>> +
>>> +    mac = ufshcd_mcq_vops_get_hba_mac(hba);
>>> +    if (mac < 0) {
>>> +        val = ufshcd_readl(hba, REG_UFS_MCQ_CFG);
>>> +        mac = FIELD_GET(MCQ_CFG_MAC_MASK, val);
>>> +    }
>>
>> According to the UFSHCI 4.0 specification the MAC value is set by the 
>> host. Can the above code read the MAC value from the host controller 
>> before it has been set by the host? If so, how about leaving out the 
>> code that reads the MAC value from the controller and making it 
>> mandatory to implement the new get_hba_mac vop?
>>
> The reason it is not mandatory to define get_hba_mac vop is UFSHCI 4.0
> specification mentions that the default value of MAC is 32. So even if a 
> vendor
> HC doesn't override the MAC, it'd be 32.
> Hence, the current code first checks for an override, and if there's 
> none uses
> the default value defined in the HC.

Hi Asutosh,

Please ignore the value reported by the controller in the MAC field of 
the MCQConfig register and overwrite the MAC field without reading it 
first. It doesn't seem useful to me to read this field. I think the host 
should decide about the queue depth no matter what the current value of 
the MAC field is.

Thanks,

Bart.