Re: [PATCH] scsi: ufs: Refactor ufshcd_is_intr_aggr_allowed()

[Adrian Hunter <adrian.hunter@intel.com>]

On 24/06/21 1:44 pm, Keoseong Park wrote:
>> On 24/06/21 9:41 am, Keoseong Park wrote:
>>>> On 21/06/21 11:51 am, Keoseong Park wrote:
>>>>> Change conditional compilation to IS_ENABLED macro,
>>>>> and simplify if else statement to return statement.
>>>>> No functional change.
>>>>>
>>>>> Signed-off-by: Keoseong Park <keosung.park@samsung.com>
>>>>> ---
>>>>>  drivers/scsi/ufs/ufshcd.h | 17 ++++++++---------
>>>>>  1 file changed, 8 insertions(+), 9 deletions(-)
>>>>>
>>>>> diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
>>>>> index c98d540ac044..6d239a855753 100644
>>>>> --- a/drivers/scsi/ufs/ufshcd.h
>>>>> +++ b/drivers/scsi/ufs/ufshcd.h
>>>>> @@ -893,16 +893,15 @@ static inline bool ufshcd_is_rpm_autosuspend_allowed(struct ufs_hba *hba)
>>>>>  
>>>>>  static inline bool ufshcd_is_intr_aggr_allowed(struct ufs_hba *hba)
>>>>>  {
>>>>> -/* DWC UFS Core has the Interrupt aggregation feature but is not detectable*/
>>>>> -#ifndef CONFIG_SCSI_UFS_DWC
>>>>> -	if ((hba->caps & UFSHCD_CAP_INTR_AGGR) &&
>>>>> -	    !(hba->quirks & UFSHCD_QUIRK_BROKEN_INTR_AGGR))
>>>>> +	/*
>>>>> +	 * DWC UFS Core has the Interrupt aggregation feature
>>>>> +	 * but is not detectable.
>>>>> +	 */
>>>>> +	if (IS_ENABLED(CONFIG_SCSI_UFS_DWC))
>>>>
>>>> Why is this needed?  It seems like you could just set UFSHCD_CAP_INTR_AGGR
>>>> and clear UFSHCD_QUIRK_BROKEN_INTR_AGGR instead?
>>>
>>> Hello Adrian,
>>> Sorry for late reply.
>>>
>>> The code that returns true when CONFIG_SCSI_UFS_DWC is set in the original code 
>>> is only changed using the IS_ENABLED macro.
>>> (Linux kernel coding style, 21) Conditional Compilation)
>>>
>>> When CONFIG_SCSI_UFS_DWC is not defined, the code for checking quirk 
>>> and caps has been moved to the newly added return statement below.
>>
>> Looking closer I cannot find CONFIG_SCSI_UFS_DWC at all.  It seems like it
>> never existed.
>>
>> Why should we not remove the code related to CONFIG_SCSI_UFS_DWC entirely?
> 
> You're right. What do you think of deleting the code related to CONFIG_SCSI_UFS_DWC 
> and changing it to the patch below?

Yes, but cc Joao Pinto <jpinto@synopsys.com> who introduced the code

> 
> ---
> diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> index c98d540ac044..c9faca237290 100644
> --- a/drivers/scsi/ufs/ufshcd.h
> +++ b/drivers/scsi/ufs/ufshcd.h
> @@ -893,16 +893,8 @@ static inline bool ufshcd_is_rpm_autosuspend_allowed(struct ufs_hba *hba)
> 
>  static inline bool ufshcd_is_intr_aggr_allowed(struct ufs_hba *hba)
>  {
> -/* DWC UFS Core has the Interrupt aggregation feature but is not detectable*/
> -#ifndef CONFIG_SCSI_UFS_DWC
> -       if ((hba->caps & UFSHCD_CAP_INTR_AGGR) &&
> -           !(hba->quirks & UFSHCD_QUIRK_BROKEN_INTR_AGGR))
> -               return true;
> -       else
> -               return false;
> -#else
> -return true;
> -#endif
> +       return (hba->caps & UFSHCD_CAP_INTR_AGGR) &&
> +               !(hba->quirks & UFSHCD_QUIRK_BROKEN_INTR_AGGR);
>  }
> 
>>
>>
>>>
>>> Thanks,
>>> Keoseong
>>>
>>>>
>>>>>  		return true;
>>>>> -	else
>>>>> -		return false;
>>>>> -#else
>>>>> -return true;
>>>>> -#endif
>>>>> +
>>>>> +	return (hba->caps & UFSHCD_CAP_INTR_AGGR) &&
>>>>> +		!(hba->quirks & UFSHCD_QUIRK_BROKEN_INTR_AGGR);
>>>>>  }
>>>>>  
>>>>>  static inline bool ufshcd_can_aggressive_pc(struct ufs_hba *hba)
>>>>>
>>>>
>>