[PATCH] i2c: fsi: Create busses for all ports

[Eddie James]

On 6/3/19 12:57 AM, Oliver O'Halloran wrote:
> Currently we only create an I2C bus for the ports listed in the
> device-tree for that master. There's no real reason for this since
> we can discover the number of ports the master supports by looking
> at the port_max field of the status register.
>
> This patch re-works the bus add logic so that we always create buses
> for each port, unless the bus is marked as unavailable in the DT. This
> is useful since it ensures that all the buses provided by the CFAM I2C
> master are accessible to debug tools.
>
> Cc: Eddie James <eajames at linux.vnet.ibm.com>
> Signed-off-by: Oliver O'Halloran <oohall at gmail.com>
> ---
>   drivers/i2c/busses/i2c-fsi.c | 30 +++++++++++++++++++++++++-----
>   1 file changed, 25 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/i2c/busses/i2c-fsi.c b/drivers/i2c/busses/i2c-fsi.c
> index 1e2be2219a60..59a76c6e31ad 100644
> --- a/drivers/i2c/busses/i2c-fsi.c
> +++ b/drivers/i2c/busses/i2c-fsi.c
> @@ -658,13 +658,27 @@ static const struct i2c_algorithm fsi_i2c_algorithm = {
>   	.functionality = fsi_i2c_functionality,
>   };
>
> +static device_node *fsi_i2c_find_port_of_node(struct device_node *master,
> +					      int port)
> +{
> +	struct device_node *np;
> +
> +	for_each_child_of_node(fsi, np) {
> +		rc = of_property_read_u32(np, "reg", &port_no);
> +		if (!rc && port_no == port)
> +			return np;
> +	}
> +
> +	return NULL;
> +}
> +
>   static int fsi_i2c_probe(struct device *dev)
>   {
>   	struct fsi_i2c_master *i2c;
>   	struct fsi_i2c_port *port;
>   	struct device_node *np;
> +	u32 port_no, ports, stat;
>   	int rc;
> -	u32 port_no;
>
>   	i2c = devm_kzalloc(dev, sizeof(*i2c), GFP_KERNEL);
>   	if (!i2c)
> @@ -678,10 +692,16 @@ static int fsi_i2c_probe(struct device *dev)
>   	if (rc)
>   		return rc;
>
> -	/* Add adapter for each i2c port of the master. */
> -	for_each_available_child_of_node(dev->of_node, np) {
> -		rc = of_property_read_u32(np, "reg", &port_no);
> -		if (rc || port_no > USHRT_MAX)
> +	rc = fsi_i2c_read_reg(i2c->fsi, I2C_FSI_STAT, &state);
> +	if (rc)
> +		return rc;
> +
> +	ports = FIELD_GET(I2C_STAT_MAX_PORT, stat);
> +	dev_dbg(dev, "I2C master has %d ports\n", ports);


Thanks for the patch Oliver. This looks great except some older CFAM 
types don't report the max port number, in which case this would not 
probe up any ports. So we probably need a fallback to dts if the max 
ports is 0.


Thanks,

Eddie


> +
> +	for (port_no = 0; port_no < ports; port_no++) {
> +		np = fsi_i2c_find_port_of_node(dev.of_node, port_no);
> +		if (np && !of_device_is_available(np))
>   			continue;
>
>   		port = kzalloc(sizeof(*port), GFP_KERNEL);