Re: [PATCH net-next] net: bcmgenet: convert RX path to page_pool

[Jacob Keller <jacob.e.keller@intel.com>]

On 6/4/2026 12:39 PM, Nicolai Buchwitz wrote:
> Hi Simon
> 
> On 4.6.2026 19:32, Simon Horman wrote:
>> From: 'Simon Horman' <horms@kernel.org>
>>
>> This is an AI-generated review of your patch. The human sending this
>> email has considered the AI review valid, or at least plausible.
>> Full review at: https://netdev-ai.bots.linux.dev/sashiko/
> 
> Thanks, I will have a look and respond individually per addressed feedback.
> 
> Out of interest, how does this sashiko instance differs from sashiko.dev?
> The later doesn't flag anything in its review [1], so I'm just curious what
> I might can add to my local pre checks.
> 

Your review on sashiko.dev is here:
https://sashiko.dev/#/patchset/20260602094248.4130712-1-nb%40tipi-net.de

And on netdev its at here:

https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260602094248.4130712-1-nb%40tipi-net.de

On sashiko.dev, the review used
gemini/gemini-3.1-pro-preview, but on the netdev instance it used
bedrock/global.anthropic.claude-opus-4-7

So at a minimum it definitely uses different models, but it might also
be using different context files as well, I know the netdev review had
its own context files from when it predated sashiko as a whole, so I
don't know if those are still fed in or differ.

Looking at the raw logs, the system prompts used are slightly different,
so that also likely plays a part in the differing results.

Additionally, repeating runs can sometimes end up with different/new
results because of the nature of LLMs relying on some randomness when
picking next tokens from its weighted outputs. Even using the same model
can give different results if you repeat the run.

> [1] https://sashiko.dev/#/patchset/20260602094248.4130712-1-nb%40tipi-
> net.de
> 
>> [...]
> 
> Regards,
> Nicolai
>