Re: [PATCH net-next 2/2] net: dsa: mv88e6xxx: add support for credit based shaper

[Luke Howard]

> I think some of these were the result of running the kernel review
> prompts through Claude. Other drivers are probably from before LLM
> reviews were introduced.

It seems to me it would be better to validate idleslope and sendslope in cbs_change() or cbs_enable_offload() so that all drivers can benefit from this? Also stylistically perhaps the comparison could be broken into >1 statements.

> I have a different opinion. The driver is in control, if these get out of sync
> you're having bigger problems. Either something else is writing to the device,
> or the hardware is changing state by itself which means it's unstable.

Valid. :)

> The 6020 family has 4 queues, but only 2 queues support CBS. That's why there's
> a seperate queue mask, even if CBS support wasn't added for 6020.

Ah, I wasn’t aware of this. I suppose this is the best solution then even though for the majority case it could be inferred.

Luke