Re: [PATCH net-next v11 5/6] net: dsa: mv88e6xxx: rmon: Use RMU for reading RMON data

[Mattias Forsblad <mattias.forsblad@gmail.com>]

On 2022-09-14 16:42, Andrew Lunn wrote:
>> @@ -1234,16 +1234,30 @@ static int mv88e6xxx_stats_get_stats(struct mv88e6xxx_chip *chip, int port,
>>  				     u16 bank1_select, u16 histogram)
>>  {
>>  	struct mv88e6xxx_hw_stat *stat;
>> +	int offset = 0;
>> +	u64 high;
>>  	int i, j;
>>  
>>  	for (i = 0, j = 0; i < ARRAY_SIZE(mv88e6xxx_hw_stats); i++) {
>>  		stat = &mv88e6xxx_hw_stats[i];
>>  		if (stat->type & types) {
>> -			mv88e6xxx_reg_lock(chip);
>> -			data[j] = _mv88e6xxx_get_ethtool_stat(chip, stat, port,
>> -							      bank1_select,
>> -							      histogram);
>> -			mv88e6xxx_reg_unlock(chip);
>> +			if (mv88e6xxx_rmu_available(chip) &&
>> +			    !(stat->type & STATS_TYPE_PORT)) {
> 
> If i understand you comment to a previous version, the problem here is
> STATS_TYPE_PORT. You are falling back to MDIO for those three
> statistics. The data sheet for the 6352 shows these statistics are
> available at offset 129 and above in the RMU reply.
> 
> I looked through the datasheets i have. I don't have a full set, but
> all that i have which include the RMU have these statistics at offset
> 129.
> 
> So i think you can remove the fallback to MDIO. That should then allow
> you to have a fully independent implementation, making it much
> cleaner.
> 
> Or do you see other problems i currently don't?
> 
>     Andrew
> 
I'll look into that, thanks.

/Mattias