sashiko-bot@kernel.org's Cc list on reviews

sashiko-bot@kernel.org's Cc list on reviews

[Sean Christopherson]

Hey Roman!

Sashiko's choice of Cc is... odd.  The original mail has a whole pile of folks
in both the To: and Cc:, but sashiko-bot is only replying to myself and (very
bizarrely) linux-hyperv@vger.kernel.org.

Oooh, is it perhaps because linux-hyperv@vger.kernel.org is the only mailing
list on the original Cc that has opted-in to getting emails?  If that's the case,
we might want to change that, even though it could result in people getting grumpy
about unwanted bot mail, because IMO the alternative is worse.  I've been responding
to some (mostly very good!) good sashiko-bot feedback, but only just now noticed
that the vast majority of folks on the To/Cc of the original patch won't see the
reviews.  That's bound to cause confusion, e.g. if someone reports the same issues
and/or manually looks at sashiko.dev.

----- Forwarded message from sashiko-bot@kernel.org -----

Date: Fri, 15 May 2026 19:55:51 +0000
From: sashiko-bot@kernel.org
To: Sean Christopherson <seanjc@google.com>
Cc: linux-hyperv@vger.kernel.org
Subject: Re: [PATCH v3 36/41] x86/kvmclock: Get local APIC bus frequency from PV CPUID Timing Info

...

----- End forwarded message -----

Re: sashiko-bot@kernel.org's Cc list on reviews

[Roman Gushchin]

On Mon, May 18, 2026 at 2:08 PM Sean Christopherson <seanjc@google.com> wrote:
>
> Hey Roman!
>
> Sashiko's choice of Cc is... odd.  The original mail has a whole pile of folks
> in both the To: and Cc:, but sashiko-bot is only replying to myself and (very
> bizarrely) linux-hyperv@vger.kernel.org.
>
> Oooh, is it perhaps because linux-hyperv@vger.kernel.org is the only mailing
> list on the original Cc that has opted-in to getting emails?  If that's the case,
> we might want to change that, even though it could result in people getting grumpy
> about unwanted bot mail, because IMO the alternative is worse.  I've been responding
> to some (mostly very good!) good sashiko-bot feedback, but only just now noticed
> that the vast majority of folks on the To/Cc of the original patch won't see the
> reviews.  That's bound to cause confusion, e.g. if someone reports the same issues
> and/or manually looks at sashiko.dev.

Hi Sean!

It's expected per configuration here:
https://github.com/sashiko-dev/sashiko/blob/main/sashiko.dev/email_policy.toml#L138
.
It can be configured to reply to all as well, but since there are some
very vocal anti-AI folks in the community, this might create some
friction. In any case, I'm relying on the corresponding maintainers'
decision on how they want to configure Sashiko.

Thanks!

Re: sashiko-bot@kernel.org's Cc list on reviews

[Sean Christopherson]

On Mon, May 18, 2026, Roman Gushchin wrote:
> On Mon, May 18, 2026 at 2:08 PM Sean Christopherson <seanjc@google.com> wrote:
> >
> > Hey Roman!
> >
> > Sashiko's choice of Cc is... odd.  The original mail has a whole pile of folks
> > in both the To: and Cc:, but sashiko-bot is only replying to myself and (very
> > bizarrely) linux-hyperv@vger.kernel.org.
> >
> > Oooh, is it perhaps because linux-hyperv@vger.kernel.org is the only mailing
> > list on the original Cc that has opted-in to getting emails?  If that's the case,
> > we might want to change that, even though it could result in people getting grumpy
> > about unwanted bot mail, because IMO the alternative is worse.  I've been responding
> > to some (mostly very good!) good sashiko-bot feedback, but only just now noticed
> > that the vast majority of folks on the To/Cc of the original patch won't see the
> > reviews.  That's bound to cause confusion, e.g. if someone reports the same issues
> > and/or manually looks at sashiko.dev.
> 
> Hi Sean!
> 
> It's expected per configuration here:
> https://github.com/sashiko-dev/sashiko/blob/main/sashiko.dev/email_policy.toml#L138
> .
> It can be configured to reply to all as well, but since there are some
> very vocal anti-AI folks in the community, this might create some
> friction. 

Heh, "might".

> In any case, I'm relying on the corresponding maintainers' decision on how
> they want to configure Sashiko.

What level of configurability does the parser have?  Specifically, can it key off
the Cc/To in addition to the list?  I ask because I really, really, want to enable
auto-mailing for at least KVM x86, but kvm@vger.kernel.org is wildly overloaded.
I could try and drive consensus across all 30+ maintainers, but if there's a way
to precisely enable it for just KVM x86...

E.g. is there a way to do something like:

[subsystems.kvm_x86]
lists = ["kvm@vger.kernel.org" && <email is To: seanjc@google.com>]
reply_all = true

Re: sashiko-bot@kernel.org's Cc list on reviews

[Roman Gushchin]

On Mon, May 18, 2026 at 3:26 PM Sean Christopherson <seanjc@google.com> wrote:
>
> On Mon, May 18, 2026, Roman Gushchin wrote:
> > On Mon, May 18, 2026 at 2:08 PM Sean Christopherson <seanjc@google.com> wrote:
> > >
> > > Hey Roman!
> > >
> > > Sashiko's choice of Cc is... odd.  The original mail has a whole pile of folks
> > > in both the To: and Cc:, but sashiko-bot is only replying to myself and (very
> > > bizarrely) linux-hyperv@vger.kernel.org.
> > >
> > > Oooh, is it perhaps because linux-hyperv@vger.kernel.org is the only mailing
> > > list on the original Cc that has opted-in to getting emails?  If that's the case,
> > > we might want to change that, even though it could result in people getting grumpy
> > > about unwanted bot mail, because IMO the alternative is worse.  I've been responding
> > > to some (mostly very good!) good sashiko-bot feedback, but only just now noticed
> > > that the vast majority of folks on the To/Cc of the original patch won't see the
> > > reviews.  That's bound to cause confusion, e.g. if someone reports the same issues
> > > and/or manually looks at sashiko.dev.
> >
> > Hi Sean!
> >
> > It's expected per configuration here:
> > https://github.com/sashiko-dev/sashiko/blob/main/sashiko.dev/email_policy.toml#L138
> > .
> > It can be configured to reply to all as well, but since there are some
> > very vocal anti-AI folks in the community, this might create some
> > friction.
>
> Heh, "might".

:)

> > In any case, I'm relying on the corresponding maintainers' decision on how
> > they want to configure Sashiko.
>
> What level of configurability does the parser have?  Specifically, can it key off
> the Cc/To in addition to the list?  I ask because I really, really, want to enable
> auto-mailing for at least KVM x86, but kvm@vger.kernel.org is wildly overloaded.
> I could try and drive consensus across all 30+ maintainers, but if there's a way
> to precisely enable it for just KVM x86...
>
> E.g. is there a way to do something like:
>
> [subsystems.kvm_x86]
> lists = ["kvm@vger.kernel.org" && <email is To: seanjc@google.com>]
> reply_all = true

No, currently it's only per-mailing list. I did not expect that it
would be that contentious.
Maybe it should be based on entries in the maintainers file instead.
But even then,
what if a sub-subsystem wants emails and the top-level subsystem maintainers
are against it? Or if a patchset touches multiple subsystems.

I mean if building a complex configuration system is the only way
forward, ok, it's kinda silly,
but why not.

Re: sashiko-bot@kernel.org's Cc list on reviews

[Sean Christopherson]

On Mon, May 18, 2026, Roman Gushchin wrote:
> On Mon, May 18, 2026 at 3:26 PM Sean Christopherson <seanjc@google.com> wrote:
> > On Mon, May 18, 2026, Roman Gushchin wrote:
> > > On Mon, May 18, 2026 at 2:08 PM Sean Christopherson <seanjc@google.com> wrote:
> > > In any case, I'm relying on the corresponding maintainers' decision on how
> > > they want to configure Sashiko.
> >
> > What level of configurability does the parser have?  Specifically, can it key off
> > the Cc/To in addition to the list?  I ask because I really, really, want to enable
> > auto-mailing for at least KVM x86, but kvm@vger.kernel.org is wildly overloaded.
> > I could try and drive consensus across all 30+ maintainers, but if there's a way
> > to precisely enable it for just KVM x86...
> >
> > E.g. is there a way to do something like:
> >
> > [subsystems.kvm_x86]
> > lists = ["kvm@vger.kernel.org" && <email is To: seanjc@google.com>]
> > reply_all = true
> 
> No, currently it's only per-mailing list. I did not expect that it
> would be that contentious.

Before you go add more configuration goo on your end, let me see what the other
kvm@ maintainers think.  Maybe we'll get lucky and everyone is on-board already.

> Maybe it should be based on entries in the maintainers file instead.
> But even then, what if a sub-subsystem wants emails and the top-level
> subsystem maintainers are against it? Or if a patchset touches multiple
> subsystems.

Hmm, maybe use a page table walker approach, and only send emails if every node of
the tree says yes?  Only getting email some of the time would be pretty darn
annoying though :-/

> I mean if building a complex configuration system is the only way forward,
> ok, it's kinda silly, but why not.

s/kinda silly/flat out stupid, given how easy it is to filter/ignore email.

Crazy idea: as a stopgap, what if sashiko emails linux-kernel@vger.kernel.org if
the list hasn't opted-in?  It wouldn't be as good as getting mails sent directly
to me, but subscribing to and filtering linux-kernel@vger.kernel.org mails is very
doable.  And even without subscribing to linux-kernel@vger.kernel.org, for those
of us that heavily utilize b4, it would better than clicking through a web site.

Re: sashiko-bot@kernel.org's Cc list on reviews

[Sean Christopherson]

On Mon, May 18, 2026, Sean Christopherson wrote:
> On Mon, May 18, 2026, Roman Gushchin wrote:
> > On Mon, May 18, 2026 at 2:08 PM Sean Christopherson <seanjc@google.com> wrote:
> > >
> > > Hey Roman!
> > >
> > > Sashiko's choice of Cc is... odd.  The original mail has a whole pile of folks
> > > in both the To: and Cc:, but sashiko-bot is only replying to myself and (very
> > > bizarrely) linux-hyperv@vger.kernel.org.
> > >
> > > Oooh, is it perhaps because linux-hyperv@vger.kernel.org is the only mailing
> > > list on the original Cc that has opted-in to getting emails?  If that's the case,
> > > we might want to change that, even though it could result in people getting grumpy
> > > about unwanted bot mail, because IMO the alternative is worse.  I've been responding
> > > to some (mostly very good!) good sashiko-bot feedback, but only just now noticed
> > > that the vast majority of folks on the To/Cc of the original patch won't see the
> > > reviews.  That's bound to cause confusion, e.g. if someone reports the same issues
> > > and/or manually looks at sashiko.dev.
> > 
> > Hi Sean!
> > 
> > It's expected per configuration here:
> > https://github.com/sashiko-dev/sashiko/blob/main/sashiko.dev/email_policy.toml#L138
> > .
> > It can be configured to reply to all as well, but since there are some
> > very vocal anti-AI folks in the community, this might create some
> > friction. 
> 
> Heh, "might".
> 
> > In any case, I'm relying on the corresponding maintainers' decision on how
> > they want to configure Sashiko.
> 
> What level of configurability does the parser have?  Specifically, can it key off
> the Cc/To in addition to the list?  I ask because I really, really, want to enable
> auto-mailing for at least KVM x86, but kvm@vger.kernel.org is wildly overloaded.
> I could try and drive consensus across all 30+ maintainers, but if there's a way
> to precisely enable it for just KVM x86...
> 
> E.g. is there a way to do something like:
> 
> [subsystems.kvm_x86]
> lists = ["kvm@vger.kernel.org" && <email is To: seanjc@google.com>]
> reply_all = true

Actually, shouldn't this be a line item in MAINTAINERS?  E.g. an "A:" section
to describe the subsystem's preferences for AI reviews (and in the future, other
things?).

Re: sashiko-bot@kernel.org's Cc list on reviews

[Roman Gushchin]

On Mon, May 18, 2026 at 4:27 PM Sean Christopherson <seanjc@google.com> wrote:
>
> On Mon, May 18, 2026, Sean Christopherson wrote:
> > On Mon, May 18, 2026, Roman Gushchin wrote:
> > > On Mon, May 18, 2026 at 2:08 PM Sean Christopherson <seanjc@google.com> wrote:
> > > >
> > > > Hey Roman!
> > > >
> > > > Sashiko's choice of Cc is... odd.  The original mail has a whole pile of folks
> > > > in both the To: and Cc:, but sashiko-bot is only replying to myself and (very
> > > > bizarrely) linux-hyperv@vger.kernel.org.
> > > >
> > > > Oooh, is it perhaps because linux-hyperv@vger.kernel.org is the only mailing
> > > > list on the original Cc that has opted-in to getting emails?  If that's the case,
> > > > we might want to change that, even though it could result in people getting grumpy
> > > > about unwanted bot mail, because IMO the alternative is worse.  I've been responding
> > > > to some (mostly very good!) good sashiko-bot feedback, but only just now noticed
> > > > that the vast majority of folks on the To/Cc of the original patch won't see the
> > > > reviews.  That's bound to cause confusion, e.g. if someone reports the same issues
> > > > and/or manually looks at sashiko.dev.
> > >
> > > Hi Sean!
> > >
> > > It's expected per configuration here:
> > > https://github.com/sashiko-dev/sashiko/blob/main/sashiko.dev/email_policy.toml#L138
> > > .
> > > It can be configured to reply to all as well, but since there are some
> > > very vocal anti-AI folks in the community, this might create some
> > > friction.
> >
> > Heh, "might".
> >
> > > In any case, I'm relying on the corresponding maintainers' decision on how
> > > they want to configure Sashiko.
> >
> > What level of configurability does the parser have?  Specifically, can it key off
> > the Cc/To in addition to the list?  I ask because I really, really, want to enable
> > auto-mailing for at least KVM x86, but kvm@vger.kernel.org is wildly overloaded.
> > I could try and drive consensus across all 30+ maintainers, but if there's a way
> > to precisely enable it for just KVM x86...
> >
> > E.g. is there a way to do something like:
> >
> > [subsystems.kvm_x86]
> > lists = ["kvm@vger.kernel.org" && <email is To: seanjc@google.com>]
> > reply_all = true
>
> Actually, shouldn't this be a line item in MAINTAINERS?  E.g. an "A:" section
> to describe the subsystem's preferences for AI reviews (and in the future, other
> things?).

This would work too. But when people send patches (including
LLM-generated patches) get_maintainers.pl
will produce a combined list of all relevant maintainers and mailing
lists. There is no option for "I don't want to be cc'ed
on patches from mailing list X". It sounds like some folks are looking
for this option for AI reviews...