On Sat, Mar 11, 2023 at 10:54:59AM -0800, Eric Biggers wrote:
On Sat, Mar 11, 2023 at 01:26:57PM -0500, Sasha Levin wrote:
"job"? do you think I'm paid to do this work?
Why would I stonewall improvements to the process?
I'm getting a bunch of suggestions and complaints that I'm not implementing
those suggestions fast enough on my spare time.
> One of the first things I would do if I was maintaining the stable kernels is to
> set up a way to automatically run searches on the mailing lists, and then take
> advantage of that in the stable process in various ways. Not having that is the
> root cause of a lot of the issues with the current process, IMO.
"if I was maintaining the stable kernels" - why is this rellevant? give
us the tool you've proposed below and we'll be happy to use it. Heck,
don't give it to us, use it to review the patches we're sending out for
review and let us know if we've missed anything.
It's kind of a stretch to claim that maintaining the stable kernels is not part
of your and Greg's jobs. But anyway, the real problem is that it's currently
very hard for others to contribute, given the unique role the stable maintainers
have and the lack of documentation about it. Each of the two maintainers has
their own scripts, and it is not clear how they use them and what processes they
follow. (Even just stable-kernel-rules.rst is totally incorrect these days.)
Actually I still don't even know where your scripts are! They are not in
stable-queue/scripts, it seems those are only Greg's scripts? And if I built
something, how do I know you would even use it? You likely have all sorts of
requirements that I don't even know about.
https://kernel.googlesource.com/pub/scm/linux/kernel/git/sashal/stable-tools/
I've last updated it about two years ago, but really it's not out of
date - it just doesn't get that many changes at this point.
This is a mess we want to solve too: having a single repository with
tools for "maintaining stable kernel trees" would be ideal and awesome,
but it's quite the lift.
We ended up with different scripts because we started trying to solve
different issues, and ended up converging into the same tree: even now,
each of us handles different subsection of commits going into the kernel
tree, we just end up pushing them into the same stable branch at the
end.
I've been playing with this in the past - I had a bot that looks at the
mailing lists for patches that are tagged for stable, and attempts to
apply/build then on the multiple trees to verify that it works and send
a reply back if something goes wrong, asking for a backport.
It gets a bit tricky as there's no way to go back from a commit to the
initial submission, you start hitting issues like:
- Patches get re-sent multiple times (think stuff like tip trees,
reviews from other maintainers, etc).
- Different versions of patches - for example, v1 was a single patch
and in v2 it became multiple patches.
I'm not arguing against your idea, I'm just saying that it's not
trivial. An incomplete work here simply won't scale to the thousands of
patches that flow in the trees, and won't be as useful. I don't think
that this is trivial as you suggest.
There are obviously going to be edge cases; another one is commits that show up
in git without ever having been sent to the mailing list. I don't think they
actually matter very much, though. Worst case, we miss some things, but still
find everything else.
Consider the opposite, which I just saw earlier today with a commit that
was tagged for stable: https://lore.kernel.org/all/20230217022200.3092987-1-yukuai1@xxxxxxxxxxxxxxx/
Here, commit 1/2 reverts a previously broken fix, and is not marked for
stable. Commit 2/2 applies the proper fix, but won't apply cleanly or
correctly unless you have patch 1/2.
In this case you need both commits in the series, rather than none of
them, otherwise you leave the trees broken.
If you disagree, and really think it's trivial, take 5 minutes to write
something up? please?
I never said that it's "trivial" or that it would take only 5 minutes; that's
just silly. Just that this is possible and it's what needs to be done.
If you don't have time, you should instead be helping ensure that the work gets
done by someone else (internship, GSoC project, etc.).
My personal experience with this approach was that:
1. It ends up being more effort mentoring someone who is unfamailiar
with this work rather than doing it myself.
2. There are very *very* few people who want to be doing this: to begin
with the kernel is one of the less popular areas to get into, and on top
of that the stable tree work is even worse because you do "maintenance"
rather than write new shiny features.
And yes, I am interested in contributing, but as I mentioned I think you need to
first acknowledge that there is a problem, fix your attitude of immediately
pushing back on everything, and make it easier for people to contribute.
I don't think we disagree that the process is broken: this is one of the
reasons we went away from trying to support 6 year LTS kernels.
However, we are not pushing back on ideas, we are asking for a hand in
improving the process: we've been getting drive-by comments quite often,
but when it comes to be doing the actual work people are quite reluctant
to help.
If you want to sit down and scope out initial set of work around tooling
to help here I'm more than happy to do that: I'm planning to be both in
OSS and LPC if you want to do it in person, along with anyone else
interested in helping out.
--
Thanks,
Sasha
