driver reviewer rotation

driver reviewer rotation

[Jakub Kicinski]

Hi!

As some of you may have noticed we have restarted a structured reviewer
rotation. There should be an uptick in the number of reviews when you
post patches.

Here is some details, and background info.

The majority of submissions we get are for drivers. We have tried 
to create a driver review rotation a while back, to increase the
review coverage, but it fizzled out. We're taking a second go at it.

The new rotation is limited in length (4 weeks) and focused on
involving NIC teams (nVidia, Intel, Broadcom, plus Meta/FB to make 
it 4), rather than particular individuals. I picked the NIC vendors 
for multiple reasons - with small exceptions they send more patches 
than they review. Secondly they have rather large teams, which makes 
it easier to create a stable rotation - employees from the same org 
can load balance and cover for each other. Last but not least, I have 
a possibly unfounded belief, that in a vendor setting the additional
structure of a review rotation is doubly beneficial as "organized
efforts" are usually easier to justify to corporate overlords.

Please feel free to reach out if you'd like to also be a part of 
a review rotation. We can start a second circle or double up one of 
the shifts... we'll figure something out. Also please reach out with 
any comments / concerns / feedback.

FWIW any "corporate involvement" in the community makes me feel uneasy
(and I hope that other community members share this feeling).
So please don't view this as any form of corporate collusion or giving
companies themselves influence. This is also not an indictment against
the community members who are already investing their time in reviewing
code, and making this project work...

HTH

Re: driver reviewer rotation

[Dave Taht]

On Tue, Dec 6, 2022 at 11:04 AM Jakub Kicinski <kuba@kernel.org> wrote:
>
> Hi!
>
> As some of you may have noticed we have restarted a structured reviewer
> rotation. There should be an uptick in the number of reviews when you
> post patches.
>
> Here is some details, and background info.
>
> The majority of submissions we get are for drivers. We have tried
> to create a driver review rotation a while back, to increase the
> review coverage, but it fizzled out. We're taking a second go at it.
>
> The new rotation is limited in length (4 weeks) and focused on
> involving NIC teams (nVidia, Intel, Broadcom, plus Meta/FB to make
> it 4), rather than particular individuals. I picked the NIC vendors
> for multiple reasons - with small exceptions they send more patches
> than they review. Secondly they have rather large teams, which makes
> it easier to create a stable rotation - employees from the same org
> can load balance and cover for each other. Last but not least, I have
> a possibly unfounded belief, that in a vendor setting the additional
> structure of a review rotation is doubly beneficial as "organized
> efforts" are usually easier to justify to corporate overlords.

+1.

I am assuming this is principally ethernet?

Because it would be so wonderful if the developers of drivers for wifi
& 5g, (especially) on phones, and aps, and clients of all sorts, all
somehow were enabled to more regularly got together... to do something
other than commiserate.

I was looking over the 2016 make-wifi-fast project description (after
gfiber pulled out) the other day, wishing that somehow, some set of
corps that cared about good wifi and lte experiences would get
together to make it better for everyone.

https://docs.google.com/document/d/1Se36svYE1Uzpppe1HWnEyat_sAGghB3kE285LElJBW4/edit#

And while it did get a bit better, the work remains kind of lonely. I
find sitting in a virtual shared videoconference to be faster (and
more fun) than reviewing code by yourself.

"Pain shared is decreased. Joy shared, increased." - Spider Robinson.

> Please feel free to reach out if you'd like to also be a part of
> a review rotation. We can start a second circle or double up one of
> the shifts... we'll figure something out. Also please reach out with
> any comments / concerns / feedback.
>

My principal thing on the ethernet front has been merely to try and
ensure subsystems like BQL are in new ethernet drivers. If there was
an AI other than me that could get "triggered" on that front that
would be great. BQL itself is showing its age, tho...

> FWIW any "corporate involvement" in the community makes me feel uneasy
> (and I hope that other community members share this feeling).
> So please don't view this as any form of corporate collusion or giving
> companies themselves influence. This is also not an indictment against
> the community members who are already investing their time in reviewing
> code, and making this project work...
>
> HTH



-- 
This song goes out to all the folk that thought Stadia would work:
https://www.linkedin.com/posts/dtaht_the-mushroom-song-activity-6981366665607352320-FXtz
Dave Täht CEO, TekLibre, LLC

Re: driver reviewer rotation

[Andrew Lunn]

> My principal thing on the ethernet front has been merely to try and
> ensure subsystems like BQL are in new ethernet drivers. If there was
> an AI other than me that could get "triggered" on that front that
> would be great. BQL itself is showing its age, tho...

Feel free to review very specific parts of a new driver. You don't
need to review it all.

I would say in general, the parts of a driver actually moving frames
around is under reviewed. So if anybody does have an interest in that,
i'm sure reviews for just that would be welcome.

    Andrew