Re: [PATCH net-next 5/5] ionic: Add .get_fec_stats ethtool handler

[Eric Joyner <eric.joyner@amd.com>]

On 5/6/2026 3:00 AM, Vadim Fedorenko wrote:
> [You don't often get email from vadim.fedorenko@linux.dev. Learn why this is
> important at https://aka.ms/LearnAboutSenderIdentification ]
> 
> Caution: This message originated from an External Source. Use proper caution
> when opening attachments, clicking links, or responding.
> 
> 
> On 05/05/2026 20:43, Eric Joyner wrote:
>> On 5/5/2026 6:54 AM, Vadim Fedorenko wrote:
>>> [You don't often get email from vadim.fedorenko@linux.dev. Learn why this is
>>> important at https://aka.ms/LearnAboutSenderIdentification ]
>>>
>>> Caution: This message originated from an External Source. Use proper caution
>>> when opening attachments, clicking links, or responding.
>>>
>>>
>>> On 01/05/2026 04:15, Eric Joyner wrote:
>>>> Several FEC error statistics being collected can be reported in a
>>>> dedicated ethtool callback for FEC errors, so implement the handler that
>>>> does so. This includes 802.3ck FEC histogram data that some newer
>>>> hardware collects.
>>>>
>>>> Assisted-by: Claude:claude-4.6-sonnet
>>>> Signed-off-by: Eric Joyner <eric.joyner@amd.com>
>>>> ---
>>>>    .../ethernet/pensando/ionic/ionic_ethtool.c   | 51 +++++++++++++++++++
>>>>    1 file changed, 51 insertions(+)
>>>>
>>>> diff --git a/drivers/net/ethernet/pensando/ionic/ionic_ethtool.c b/drivers/
>>>> net/ethernet/pensando/ionic/ionic_ethtool.c
>>>> index 78a802eb159f..fe1f753b6115 100644
>>>> --- a/drivers/net/ethernet/pensando/ionic/ionic_ethtool.c
>>>> +++ b/drivers/net/ethernet/pensando/ionic/ionic_ethtool.c
>>>> @@ -418,6 +418,56 @@ static int ionic_get_fecparam(struct net_device *netdev,
>>>>        return 0;
>>>>    }
>>>>
>>>> +static const struct ethtool_fec_hist_range ionic_fec_ranges[] = {
>>>> +     { 0, 0},
>>>> +     { 1, 1},
>>>> +     { 2, 2},
>>>> +     { 3, 3},
>>>> +     { 4, 4},
>>>> +     { 5, 5},
>>>> +     { 6, 6},
>>>> +     { 7, 7},
>>>> +     { 8, 8},
>>>> +     { 9, 9},
>>>> +     { 10, 10},
>>>> +     { 11, 11},
>>>> +     { 12, 12},
>>>> +     { 13, 13},
>>>> +     { 14, 14},
>>>> +     { 15, 15},
>>>> +     { 0, 0},
>>>> +};
>>>> +
>>>> +static void
>>>> +ionic_fill_fec_hist(const struct ionic_port_extra_stats *extra_stats,
>>>> +                 struct ethtool_fec_hist *hist)
>>>> +{
>>>> +     int i;
>>>> +
>>>> +     hist->ranges = ionic_fec_ranges;
>>>> +     for (i = 0; i < ETHTOOL_FEC_HIST_MAX - 1; i++)
>>>> +             hist->values[i].sum = extra_stats->fec_codeword_error_bin[i];
>>>> +}
>>>
>>> ETHTOOL_FEC_HIST_MAX = 17, you defined 16 bins, but iterating over 15 of
>>> them. Looks like bin {15, 15} will be lost in stats.
>>>
>>>
>>
>> This looks correct to me -- (ETHTOOL_FEC_HIST_MAX - 1) = 16, so starting with
>> i=0, it'll iterate through the 16 bins and ignore the 17th end marker bin. Bin
>> 15 does get included and gets its sum set.
> 
> Ahh, yeah, you're right, smth was wrong with my math yesterday. btw, how
> does it work for different FEC? RS(528, 514) will not give you 16 bins..
> 
> 

According to the datasheet for the device that does support reporting these
histogram bins, there's no support for "Clause 91 RS(528,514) "KR4" FEC for
100GbE". I only see support for PAM4 and not NRZ.

- Eric