Re: [ANNOUNCE] Email reviews from Sashiko enabled for kvm@

[Paolo Bonzini <pbonzini@redhat.com>]

On 5/22/26 18:38, Sean Christopherson wrote:
> FYI, we've opted-in kvm@vger.kernel.org for email reviews from Sashiko[*], so
> don't be surprised if/when you start getting review feedback.  Given the broad
> usage of kvm@, Sashiko is configured to email the patch author and Cc kvm@,
> e.g. won't blast an email to the entire Cc of the original patch.
> 
> If Sashiko points out a real flaw, you'll likely get asked to address the issue,
> but unless the maintainer(s) for a (sub)subsystem say otherwise, there are no
> expectations that you respond to sashiko-bot (though it's totally fine and even
> encouraged to do so).  In short, treat Sashiko like any other bot.
> 
> The signal:noise ratio for Sashiko has gotten pretty solid, so we don't expect
> an onslaught of slop, but if there are problems with the volume and/or quality
> of feedback, we'll adjust and tweak as needed (again, as we do for any other bot).

A couple more things that are specific to KVM generic and x86 patches 
(though, generally speaking, none of this is a bad idea):

* if you see a false positive, consider that a human might be confused 
in the same way: adding a comment about something being preexisting or 
not expected to happen will often be enough to silence Sashiko and might 
help people reading the code months or years down the line.

* while you are not expected to respond to the bot, it can be convenient 
to do at least one quick reply somewhere in the thread noting that 
you've read it and whether you are waiting a few days for human reviews 
before sending v(N+1).  This is not mandatory either, it's just kindness 
to the reviewers.

* the only thing you *are* expected to do is to point out, as you would 
for human input, any changes from the previous posting that were 
suggested or done in response to Sashiko reviews.  This is nothing new, 
as it is standard practice to point out all changes, but please do 
mention the source of the suggestion.

Also, right now Sashiko assumes that all patches are based on 
kvm/master, which can result in failure to apply them.  This should be 
fixed soon on either our or their side.

Thanks,

Paolo