[RFC,2/5] i2c: core: add new variant to check for a client

For the new 'alias' feature, we need to scan for devices while holding
the lock. We focus on read_byte transactions for now to keep things
simple. Requesting an alias will be rare, so there is not much overhead.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
---
 drivers/i2c/i2c-core-base.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

Hi Wolfram,

Thank you for the patch.

On Tue, Dec 31, 2019 at 05:13:57PM +0100, Wolfram Sang wrote:
> For the new 'alias' feature, we need to scan for devices while holding
> the lock. We focus on read_byte transactions for now to keep things
> simple. Requesting an alias will be rare, so there is not much overhead.
> 
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> ---
>  drivers/i2c/i2c-core-base.c | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
> 
> diff --git a/drivers/i2c/i2c-core-base.c b/drivers/i2c/i2c-core-base.c
> index 20a726dc78db..51bd953ddfb2 100644
> --- a/drivers/i2c/i2c-core-base.c
> +++ b/drivers/i2c/i2c-core-base.c
> @@ -2108,6 +2108,23 @@ static int i2c_default_probe(struct i2c_adapter *adap, unsigned short addr)
>  	return err >= 0;
>  }
>  
> +static int i2c_unlocked_read_byte_probe(struct i2c_adapter *adap, unsigned short addr)
> +{
> +	union i2c_smbus_data dummy;
> +	int err;
> +
> +	if (i2c_check_functionality(adap, I2C_FUNC_SMBUS_READ_BYTE)) {
> +		err = __i2c_smbus_xfer(adap, addr, 0, I2C_SMBUS_READ, 0,
> +				     I2C_SMBUS_BYTE, &dummy);
> +	} else {
> +		dev_warn(&adap->dev, "No suitable probing method supported for address 0x%02X\n",

This should be wrapped after the first argument.

s/02X/02x/

As this function is static and unused in this patch, this may introduce
a new warning when bisection. I see little reason to keep this patch
standalone, you can squash it with 3/5. It otherwise looks good to me.

> +			 addr);
> +		err = -EOPNOTSUPP;
> +	}
> +
> +	return err >= 0;
> +}
> +
>  static int i2c_scan_for_client(struct i2c_adapter *adap, unsigned short addr,
>  			    int (*probe)(struct i2c_adapter *adap, unsigned short addr))
>  {

Hi Wolfram,

On 31/12/2019 16:13, Wolfram Sang wrote:
> For the new 'alias' feature, we need to scan for devices while holding
> the lock. We focus on read_byte transactions for now to keep things
> simple. Requesting an alias will be rare, so there is not much overhead.
> 
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> ---
>  drivers/i2c/i2c-core-base.c | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
> 
> diff --git a/drivers/i2c/i2c-core-base.c b/drivers/i2c/i2c-core-base.c
> index 20a726dc78db..51bd953ddfb2 100644
> --- a/drivers/i2c/i2c-core-base.c
> +++ b/drivers/i2c/i2c-core-base.c
> @@ -2108,6 +2108,23 @@ static int i2c_default_probe(struct i2c_adapter *adap, unsigned short addr)
>  	return err >= 0;
>  }
>  
> +static int i2c_unlocked_read_byte_probe(struct i2c_adapter *adap, unsigned short addr)
> +{
> +	union i2c_smbus_data dummy;
> +	int err;
> +
> +	if (i2c_check_functionality(adap, I2C_FUNC_SMBUS_READ_BYTE)) {
> +		err = __i2c_smbus_xfer(adap, addr, 0, I2C_SMBUS_READ, 0,
> +				     I2C_SMBUS_BYTE, &dummy);
> +	} else {
> +		dev_warn(&adap->dev, "No suitable probing method supported for address 0x%02X\n",

Laurent has already pointed out the format issue here, and that this
function might be squashed into 3/5 anyway,

But otherwise it looks good to me, so if you keep it separate:

Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>

> +			 addr);
> +		err = -EOPNOTSUPP;
> +	}
> +
> +	return err >= 0;
> +}
> +
>  static int i2c_scan_for_client(struct i2c_adapter *adap, unsigned short addr,
>  			    int (*probe)(struct i2c_adapter *adap, unsigned short addr))
>  {
>