Re: [RFC] Mitigating Sashiko-driven iteration noise on the main MPTCP list

[Matthieu Baerts <matttbe@kernel.org>]

Hi Shardul,

Thank you for sharing this feedback!

On 07/09/2026 14:57, Shardul Bankar wrote:
> On 04/09/2026 12:32, Matthieu Baerts wrote:
>> So at the end, it seems like a good thing to have access to various
>> models to cross checks the reviews.
> 
> Agreed, and that is where I ran into something worth passing on. If the
> plan is for people to run reviews locally, and more than one model at
> that, then what a local run costs decides whether anyone actually keeps
> doing it.

Just to avoid confusions, even if running multiple models *should* be
better (and/or having one cross-reviewing the comment from another one),
running one is already good, but only for those who are able to, i.e.
companies already using them. And that's what netdev maintainers expect,
so it somehow applies to MPTCP as well. I'm clearly not going to enforce
it, especially to individuals (like me) or people who don't want to use
for various reasons. The idea is "if your company is using LLM, then use
it to pre-review patches to save cycles".

> I ran Sashiko locally against my Claude subscription, using the Claude
> Code CLI provider rather than an API key. One patch took about 45
> minutes and consumed essentially a whole usage window. That window is
> five hours, so it works out at roughly one patch per window on that
> setup, which was enough to have me scheduling runs around the reset
> rather than just running them when I wanted a review. That timing was in
> August, on a build from before the declarative workflow engine landed,
> so I cannot say the 45 minutes still holds. What has not changed is the
> shape of it: the CLI provider still starts a separate claude process per
> stage, with no session carried between them, so the context is built and
> sent again for each one.
> 
> The part I did not know at the time is that there are two quite
> different things here. "sashiko init" then "sashiko review" runs the
> full staged pipeline, one process per stage, and that is the expensive
> one. The Sashiko repository also ships, in the prompt bundle it vendors,
> a /kreview command that runs the review as a single Claude Code session,
> so the context is read once instead of once per stage.
> 
> /kreview is not mentioned in Sashiko's README, its maintainers guide, or
> anything under docs/. It is documented only in
> third_party/prompts/README.md, inside the vendored copy of the review
> prompts, with the command itself in
> third_party/prompts/kernel/slash-commands/kreview.md. So anyone
> following the setup instructions gets the expensive path and has no
> particular reason to learn the other one exists. That seems worth fixing
> if the goal is more people running checks before they post.

Good idea!

(I didn't know it could take that long + the window time)

> A few of the things that make a local run costlier than it needs to be
> look fixable, and I have sent patches for them to the Sashiko project
> rather than describe them here, since that is where they belong. One is
> worth a line because of where it does and does not apply: Sashiko can
> cache model responses so a repeated request is answered from disk rather
> than sent again, but that setting has no effect on the local path, so
> retrying a failed run pays for everything a second time. The reviews you
> see arriving on this list are unaffected, since the daemon that posts
> them does use the cache.

I do wonder how people use it. Maybe many companies have deployed their
own Sashiko and local reviews are not common? (but I doubt that)

I would also not be surprised that many people don't care about the
cost, and if they use an internal ranking, it is possibly even better
without cache...

> One other thing, unrelated to cost and possibly of wider interest. Of
> the findings Sashiko has published for this list, the ones it marks as
> pre-existing are graded higher than the ones it attributes to the patch
> under review. Roughly 95 percent of the pre-existing ones are High or
> Critical and not one is Low, against roughly half for the findings
> attributed to the patch. I had been treating "this was already broken"
> as a reason to downgrade, and that is not how the tool treats it. The
> reasoning holds up, since a pre-existing bug is live in the shipping
> kernel while one introduced by a patch under review will most likely be
> caught before it merges.
Note that pre-existing issues don't always need to be fixed as part of a
series adding a new feature. A message about that in the commit message,
in the commit note or in reply to Sashiko's email. If the issue is real,
the impact has not changed with the modification, and the fix is not
simple, it is enough to create a new ticket on GitHub and look (or
someone else) at it later.

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.