Conversation

Krzysztof Kozlowski

I think many kernel developers - including myself, a long time ago - believe that becoming a Linux kernel maintainer is a difficult process. As if some secret group has to choose you, you need to prove yourself, and then a Masonic lodge will select the best candidates from a huge pool of competition.

That's simply not true. Becoming a Linux kernel maintainer is actually quite easy.

The difficulty lies in staying one.

Every now and then, I notice subsystems that need more maintainers. Do you want to be one?

Of course taking maintainership position and being a true maintainer are a bit different things, but I hope you get the point...
2
8
16

@krzk be a maintener or a weekend hobbyist, writing kernel code is fun. I agree staying in that role takes a tremendous amount of effort and expertise.

0
0
1

@krzk I agree but with minor caveats - for established stuff, esp. core you need community buy-in and to be 'known' and appropriate for the role, in fact for me it was more the case that I (and my co-maintainers) were already doing the work _de facto_.

So it wasn't so hard to then make it 'official' :) in fact that's the case for all the M's I have really.

The key bit though is sustaining it in a workable way.

And we could, of course, talk for hours about what being a 'true' maintainer entails :)

For me it is a constant sense of doubt, dread and guilt :P

(to be clear I do love what I do! But one has to face reality ;)

2
0
0

@krzk the most obvious 'I get this' aspect of this though is it was a lot easier for me to get that first M than I thought it'd be.

I expected years, and people umming and ahhing and etc. etc.

Nah lol I made the case, chatted to other maintainers, then sent a mail to Andrew and boom...

1
0
0

@ljs @krzk Well, anyone can send a patch to MAINTAINERS. Getting an Acked-by is the tough part: You must somehow get to the same bar as the other maintainers.

1
0
1

@ptesarik @krzk and drink as much as them at that bar

1
0
1

@ljs @krzk Now, that would set the bar a bit too high.
At least for some subsystems.

1
0
0

@ptesarik @krzk you know you've reached the right level when the conversation turns to plum farming

1
1
0

@ljs @krzk Another faux pas. I thought B stood for Babka.
@vbabka

0
0
0
@ljs My post was of course simplifying or generalizing, because no one will take freshmen for example into maintaining MM or scheduling. What I wanted to say there are many non-complicated driver subsystems which need new maintainers and it is relatively easy to start there. If you have patches already in the kernel and the community recognizes you (vide xz case), then just start reviewing and in no time you will be there.

Assuming of course one knows Git, because that's another problem with young folks these days - they know Github but do not understand Git...
2
0
3

@krzk yeah of course.

I mean it matches my experience broadly speaking, I was surprised at how easy it was to get an M - once I had community buy-in and was in effect already doing the work.

But that first step is harder in some cases than others.

Honestly another aspect of this I think is how rare the very specific kind of talent you need to do kernel work is.

if you can show you have it, you are already on the road to an M if you want it I think.

0
0
1

Harry (Hyeonggon) Yoo

Edited 5 days ago
@krzk @ljs

Me crying ever since I randomly picked MM in college :'(

.
.
.

Of course, just kidding—it just takes a significant amount of time to even start reviewing when you look at complex subsystems. But no matter how complex the subsystem you're looking at is, as long as your curiosity doesn’t stop and you keep spending time on it, you’ll eventually get there.

And yeah, being good at using Git is such an important skill in a project with a long history (and being able to find and read old mailing list threads too...)
0
0
2