Received-SPF: pass (google.com: domain of linux-kernel+bounces-74374-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Message-ID: <c7635c10-1724-4db5-9568-d554e1c64f72@quicinc.com>
Date: Wed, 21 Feb 2024 17:13:24 +0800
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH] scsi: ufs: core: Fix setup_xfer_req invocation
Content-Language: en-US
To: Bart Van Assche <bvanassche@acm.org>, Rohit Ner <rohitner@google.com>,
        Bean Huo <beanhuo@micron.com>, Stanley Chu <stanley.chu@mediatek.com>,
        "Martin K. Petersen" <martin.petersen@oracle.com>,
        Jaegeuk Kim
	<jaegeuk@kernel.org>
CC: Avri Altman <avri.altman@wdc.com>, <linux-scsi@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>
References: <20240220090805.2886914-1-rohitner@google.com>
 <1920a2f6-e398-47af-a5d7-9dad9c70e03d@acm.org>
From: Can Guo <quic_cang@quicinc.com>
In-Reply-To: <1920a2f6-e398-47af-a5d7-9dad9c70e03d@acm.org>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 8bit

Hi Bart,

On 2/21/2024 1:21 AM, Bart Van Assche wrote:
> On 2/20/24 01:08, Rohit Ner wrote:
>> Allow variant callback to setup transfers without
>> restricting the transfers to use legacy doorbell
>>
>> Signed-off-by: Rohit Ner <rohitner@google.com>
>> ---
>>   drivers/ufs/core/ufshcd.c | 6 +++---
>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
>> index d77b25b79ae3..91e483dd3974 100644
>> --- a/drivers/ufs/core/ufshcd.c
>> +++ b/drivers/ufs/core/ufshcd.c
>> @@ -2280,6 +2280,9 @@ void ufshcd_send_command(struct ufs_hba *hba, 
>> unsigned int task_tag,
>>           ufshcd_clk_scaling_start_busy(hba);
>>       if (unlikely(ufshcd_should_inform_monitor(hba, lrbp)))
>>           ufshcd_start_monitor(hba, lrbp);
>> +    if (hba->vops && hba->vops->setup_xfer_req)
>> +        hba->vops->setup_xfer_req(hba, lrbp->task_tag,
>> +                        !!lrbp->cmd);
>>       if (is_mcq_enabled(hba)) {
>>           int utrd_size = sizeof(struct utp_transfer_req_desc);
>> @@ -2293,9 +2296,6 @@ void ufshcd_send_command(struct ufs_hba *hba, 
>> unsigned int task_tag,
>>           spin_unlock(&hwq->sq_lock);
>>       } else {
>>           spin_lock_irqsave(&hba->outstanding_lock, flags);
>> -        if (hba->vops && hba->vops->setup_xfer_req)
>> -            hba->vops->setup_xfer_req(hba, lrbp->task_tag,
>> -                          !!lrbp->cmd);
>>           __set_bit(lrbp->task_tag, &hba->outstanding_reqs);
>>           ufshcd_writel(hba, 1 << lrbp->task_tag,
>>                     REG_UTP_TRANSFER_REQ_DOOR_BELL);
> 
> UFS controllers that are compliant with the JEDEC UFSHCI specification do
> not need the .setup_xfer_req() callback so I think a better motivation is
> needed to make this change.

I am going to push some BUG fixes for Qualcomm UFSHCI MCQ engine, one of 
which would count on a vops in ufshcd_send_command(). My original plan 
was to add a new vops.mcq_setup_xfer_req() to differentiate from the 
existing one used in legacy mode. But if Rohit moves the existing 
setup_xfer_req() up, I can use it instead of introducing the new one.

Thanks,
Can Guo.

> 
> Thanks,
> 
> Bart.