Re: [RFC net-next] net: phy: at803x: add cable diagnostics support

[Michael Walle <michael@walle.cc>]

Am 2020-05-05 15:07, schrieb Andrew Lunn:
>> +static int at803x_cable_test_get_status(struct phy_device *phydev,
>> +					bool *finished)
>> +{
>> +	struct at803x_priv *priv = phydev->priv;
>> +	static const int ethtool_pair[] = {
>> +		ETHTOOL_A_CABLE_PAIR_0, ETHTOOL_A_CABLE_PAIR_1,
>> +		ETHTOOL_A_CABLE_PAIR_2, ETHTOOL_A_CABLE_PAIR_3};
> 
> If you put one per line, you will keep the reverse christmas tree, and
> David will be happy.
> 
>> +	int pair, val, ret;
>> +	unsigned int delay_ms;
> 
> Well, David will be happy if you move this as well.

Damn, this should really be a checkpatch.pl check ;) It was "int 
delay_ms;"
before, then it was changed to "unsigned int delay_ms;"..

> 
>> +	*finished = false;
>> +
>> +	if (priv->cdt_start) {
>> +		delay_ms = AT803X_CDT_DELAY_MS;
>> +		delay_ms -= jiffies_delta_to_msecs(jiffies - priv->cdt_start);
>> +		if (delay_ms > 0)
>> +			msleep(delay_ms);
>> +	}
>> +
>> +	for (pair = 0; pair < 4; pair++) {
>> +		ret = at803x_cdt_start(phydev, pair);
>> +		if (ret)
>> +			return ret;
>> +
>> +		ret = at803x_cdt_wait_for_completion(phydev);
>> +		if (ret)
>> +			return ret;
>> +
>> +		val = phy_read(phydev, AT803X_CDT_STATUS);
>> +		if (val < 0)
>> +			return val;
>> +
>> +		ethnl_cable_test_result(phydev, ethtool_pair[pair],
>> +					at803x_cdt_test_result(val));
>> +
>> +		if (at803x_cdt_fault_length_valid(val))
>> +			continue;
> 
> The name is not very intuitive. It return false if it is valid?

Mhh, this is actually wrong, it returns true if the length is
valid. I need to double check that. what about
at803x_cdt_fault_length_is_valid()

> Otherwise, this looks good.

I'll wait for your v3 and then I'll rebase on that.

-michael