[EXTERNAL] Re: [EXTERNAL] Re: [EXTERNAL] Re: [PATCH v3 3/6] soc: ti: pruss: Add pruss_cfg_read()/update() API

Roger Quadros rogerq at kernel.org
Sat Mar 11 04:06:48 PST 2023


Hi Danish,

On 10/03/2023 17:36, Md Danish Anwar wrote:
> Hi Roger,
> 
> On 10/03/23 18:53, Roger Quadros wrote:
>> Hi Danish,
>>
>> On 10/03/2023 13:53, Md Danish Anwar wrote:
>>> Hi Roger,
>>>
>>> On 09/03/23 17:00, Md Danish Anwar wrote:
>>>> Hi Roger,
>>>>
>>>> On 08/03/23 17:12, Roger Quadros wrote:
>>>>>
>>>>>
>>>>> On 08/03/2023 13:36, Md Danish Anwar wrote:
>>>>>> Hi Roger,
>>>>>>
>>>>>> On 08/03/23 13:57, Roger Quadros wrote:
>>>>>>> Hi,
>>>>>>>
>>>>>>> On 06/03/2023 13:09, MD Danish Anwar wrote:
>>>>>>>> From: Suman Anna <s-anna at ti.com>
>>>>>>>>
>>>>>>>> Add two new generic API pruss_cfg_read() and pruss_cfg_update() to
>>>>>>>> the PRUSS platform driver to allow other drivers to read and program
>>>>>>>> respectively a register within the PRUSS CFG sub-module represented
>>>>>>>> by a syscon driver. This interface provides a simple way for client
>>>>>>>
>>>>>>> Do you really need these 2 functions to be public?
>>>>>>> I see that later patches (4-6) add APIs for doing specific things
>>>>>>> and that should be sufficient than exposing entire CFG space via
>>>>>>> pruss_cfg_read/update().
>>>>>>>
>>>>>>>
>>>>>>
>>>>>> I think the intention here is to keep this APIs pruss_cfg_read() and
>>>>>> pruss_cfg_update() public so that other drivers can read / modify PRUSS config
>>>>>> when needed.
>>>>>
>>>>> Where are these other drivers? If they don't exist then let's not make provision
>>>>> for it now.
>>>>> We can provide necessary API helpers when needed instead of letting client drivers
>>>>> do what they want as they can be misused and hard to debug.
>>>>>
>>>>
>>>> The ICSSG Ethernet driver uses pruss_cfg_update() API. It is posted upstream in
>>>> the series [1]. The ethernet driver series is dependent on this series. In
>>>> series [1] we are using pruss_cfg_update() in icssg_config.c file,
>>>> icssg_config() API.
>>
>> You can instead add a new API on what exactly you want it to do rather than exposing
>> entire CFG space.
>>
> 
> Sure.
> 
> In icssg_config.c, a call to pruss_cfg_update() is made to enable XFR shift for
> PRU and RTU,
> 
> 	/* enable XFR shift for PRU and RTU */
> 	mask = PRUSS_SPP_XFER_SHIFT_EN | PRUSS_SPP_RTU_XFR_SHIFT_EN;
> 	pruss_cfg_update(prueth->pruss, PRUSS_CFG_SPP, mask, mask);
> 
> I will add the below API as part of Patch 4 of the series. We'll call this API
> and entire CFG space will not be exposed.
> 
> /**
>  * pruss_cfg_xfr_pru_rtu_enable() - Enable/disable XFR shift for PRU and RTU
>  * @pruss: the pruss instance
>  * @enable: enable/disable
>  *
>  * Return: 0 on success, or an error code otherwise
>  */
> static inline int pruss_cfg_xfr_pru_rtu_enable(struct pruss *pruss, bool enable)
> {
> 	u32 mask = PRUSS_SPP_XFER_SHIFT_EN | PRUSS_SPP_RTU_XFR_SHIFT_EN;
> 	u32 set = enable ? mask : 0;
> 
> 	return pruss_cfg_update(pruss, PRUSS_CFG_SPP, mask, set);
> }

I would suggest to make separate APIs for PRU XFR vs RTU XFR.

> 
> To make pruss_cfg_update() and pruss_cfg_read() API internal to pruss.c, I will
> add the below change to pruss.h file and pruss.c file. Let me know if this
> change looks okay to you.
> 
> diff --git a/drivers/soc/ti/pruss.c b/drivers/soc/ti/pruss.c
> 
> index 537a3910ffd8..9f01c8809deb 100644
> 
> --- a/drivers/soc/ti/pruss.c
> 
> +++ b/drivers/soc/ti/pruss.c
> 
> @@ -182,7 +182,6 @@ int pruss_cfg_read(struct pruss *pruss, unsigned int reg,
> unsigned int *val)

Need to declare this as 'static'.

> 
> 
> 
>         return regmap_read(pruss->cfg_regmap, reg, val);
> 
>  }
> 
> -EXPORT_SYMBOL_GPL(pruss_cfg_read);
> 
> 
> 
>  /**
> 
>   * pruss_cfg_update() - configure a PRUSS CFG sub-module register
> 
> @@ -203,7 +202,6 @@ int pruss_cfg_update(struct pruss *pruss, unsigned int reg,

this as well.

> 
> 
> 
>         return regmap_update_bits(pruss->cfg_regmap, reg, mask, val);
> 
>  }
> 
> -EXPORT_SYMBOL_GPL(pruss_cfg_update);
> 
> 
> 
>  static void pruss_of_free_clk_provider(void *data)
> 
>  {
> 
> diff --git a/include/linux/remoteproc/pruss.h b/include/linux/remoteproc/pruss.h
> 
> index d41bec448f06..12ef10b9fe9a 100644
> 
> --- a/include/linux/remoteproc/pruss.h
> 
> +++ b/include/linux/remoteproc/pruss.h
> 
> @@ -165,9 +165,6 @@ int pruss_request_mem_region(struct pruss *pruss, enum
> pruss_mem mem_id,
> 
>                              struct pruss_mem_region *region);
> 
>  int pruss_release_mem_region(struct pruss *pruss,
> 
>                              struct pruss_mem_region *region);
> 
> -int pruss_cfg_read(struct pruss *pruss, unsigned int reg, unsigned int *val);
> 
> -int pruss_cfg_update(struct pruss *pruss, unsigned int reg,
> 
> -                    unsigned int mask, unsigned int val);
> 
> 
> 
>  #else
> 
> 
> 
> @@ -191,18 +188,6 @@ static inline int pruss_release_mem_region(struct pruss
> *pruss,
> 
>         return -EOPNOTSUPP;
> 
>  }
> 
> 
> 
> -static inline int pruss_cfg_read(struct pruss *pruss, unsigned int reg,
> 
> -                                unsigned int *val)
> 
> -{
> 
> -       return -EOPNOTSUPP;
> 
> -}
> 
> -
> 
> -static inline int pruss_cfg_update(struct pruss *pruss, unsigned int reg,
> 
> -                                  unsigned int mask, unsigned int val)
> 
> -{
> 
> -       return -EOPNOTSUPP;
> 
> -}
> 
> -
> 
>  #endif /* CONFIG_TI_PRUSS */
> 
> 
> 
>  #if IS_ENABLED(CONFIG_PRU_REMOTEPROC)
> 
> 
> Please have a look and let me know if above API and code changes looks OK to you.
> 

Rest looks OK.

cheers,
-roger



More information about the linux-arm-kernel mailing list