[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <f6941817-6be2-400e-bee9-0be075884aa2@linux.dev>
Date: Wed, 30 Jul 2025 15:39:50 +0100
From: Vadim Fedorenko <vadim.fedorenko@...ux.dev>
To: Jakub Kicinski <kuba@...nel.org>
Cc: Andrew Lunn <andrew@...n.ch>, Michael Chan <michael.chan@...adcom.com>,
Pavan Chebbi <pavan.chebbi@...adcom.com>, Tariq Toukan <tariqt@...dia.com>,
Gal Pressman <gal@...dia.com>, intel-wired-lan@...ts.osuosl.org,
Donald Hunter <donald.hunter@...il.com>, Paolo Abeni <pabeni@...hat.com>,
Simon Horman <horms@...nel.org>, netdev@...r.kernel.org
Subject: Re: [RFC PATCH] ethtool: add FEC bins histogramm report
On 30/07/2025 14:44, Jakub Kicinski wrote:
> On Wed, 30 Jul 2025 10:18:46 +0100 Vadim Fedorenko wrote:
>>> IDK, 0,0 means all symbols were completely correct.
>>> It may be useful for calculating bit error rate?
>>
>> The standard doesn't have this bin, its value can be potentially
>> deducted from all packets counter.
>
> We have a number of counters outside of the standard. Here the
> extension is pretty trivial, so I don't see why we'd deprive
> the user of the information HW collects. The translation between
> bytes and symbols is not exact. Not sure we care about exactness
> but, again, trivial to keep the 0,0 bin.
>
>>> A workaround for having the {-1, -1} sentinel could also be to skip
>>> the first entry:
>>>
>>> if (i && !ranges[i].low && !ranges[i].high)
>>> break;
>>
>> I was thinking of this way, the problem is that in the core we rely on
>> the driver to provide at least 2 bins and we cannot add any compile-time
>> checks because it's all dynamic.
>
> 1 bin is no binning, its not a legit use of the histogram API.
> We have a counter for corrected symbols already, that's the "1 bin".
Got it. Ok, I agree, we can keep bin (0,0) as the very first one, I'll
implement it in the way you suggested above
Powered by blists - more mailing lists