Re: [RFC] kernel/ review rules

[Bradley Morgan]

On 21 August 2026 13:13:08 BST, Theodore Tso <tytso@xxxxxxx> wrote:
>On Fri, Aug 21, 2026 at 08:30:02AM -0500, Bradley Morgan wrote:
>> >Acked-by:			"cool changelog!"
>

well, ik I may not be good enough for your time, but here goes, :)

>As a maintainer, I completely ignore an Acked-by unless it comes from
>another Maintainer, in which case it means that the other Maintainer
>is OK with that commit which might affect their subsystem (either
>directly because it modifies code in their subsystem or indirectly
>because it changes functions that their subsystem depends upon).
>
>As a result I will likely drop an Acked-by and not include it in the
>git commit descritpion if it comes from a non-Maintainer.
>

Fair enough.

>> >Reviewed-by: + nothing		"cool changelog and I scrolled through it"
>> >Reviewed-by: + commentary:	"I reviewed it!"
>
>If there is a reviewed-by, and other problems are discovered by other
>reviewers, the reputation of the person who submitted the Reviewed-by
>will go down, and with enough bogus reviewed-by, I will start ignoring
>that reviewer and drop their Reviewed-by from the git commit
>description.
>

well, this one is, ok.

Me personally, doing reviews, do reviews for a reason, not representing any
company or anything, but I personally do this as a mental release, when
stuff gets bad, I review patches, it's a fix.


If a reviewer does a crappy review, maybe once or twice, then don't ignore
the guy yet, people are wrong, it always happens.


If it's a multiple time thing, or consistent, then we have some talking to
do, because I suspect: AI, or just the fact the reviewer doesn't
understand the code sufficiently enough to provide a review, what I would
do in this case is send a off list email saying something along the lines
of:


"What the hell's going on XXXX? You got X reviews wrong, and you nearly
just got (Bad thing, bug or regression) merged! I want you to be honest
with me, because I can help you, are you using AI for review?"

And if they say yes, I'll probably delegate this to a smarter person, not
as in let him do the brunt, as in for advice, if he doesn't understand the
code, ill help him get a understanding of the code in order to make better
reviews.


>And if there is Reviewed-by commentary which is actively wrong, I may
>start sending patch submitters form letters informing them that they
>should feel free to ignore reviews from Reviewer XXXXX since the
>reviews may be unreliable.
>

I kinda ranted above ^


>So I wouldn't focus on review numbers, but review *quality*.
>

Ack.

>Cheers,
>
>						- Ted
>


I just woke up. So bare with my very low grammar rant, please? :)

Thanks!