LWN: Comments on "Stable kernel updates"

Stable kernel updates

ErikF — Tue, 28 Nov 2017 23:42:01 +0000

Constructive criticism is fine: I'm sure that the kernel maintainers know that there are issues and welcome useful feedback. However, responding to all articles about kernel releases with derogatory comments like "Lame!" and "Ludicrous!" doesn't help matters, because there's nothing actionable about those; they seem like ad hominem statements. What exactly is ludicrous? Have those issues been reported upstream?

For me, I had some issues with the 4.13 series on a couple of my test computers, but that's why I keep several older copies of the kernel around. If you're using the most recent builds, you're going to encounter some issues because there are going to be less people who have tested them. If you want kernels that have had more exposure, don't use 4.14 yet; let people who are more willing to accept risk test them first.

Stable kernel updates

nix — Tue, 28 Nov 2017 22:59:33 +0000

Well, posts attacking people for doing important and tiresome stabilization work, distilling something resembling stability from a firehose of commits, is not merely "not sycophantic". I'd call it pretty appalling actually.

(Sure, there are regressions, but the rate is much lower than it used to be and frankly given how hard the job is I'm amazed it's not far higher. I can say that without the stable tree, it would be very hard to use anything but enterprise kernels for any serious work at all: you can't use the kernels Linus releases without your heart in your mouth, since they just have too many serious bugs and not enough real runtime, and in the absence of the stable kernels they'd only get fixed at the same time as more bugs are *introduced*, leaving you with nothing usable at all.)

So please stop acting like getting given bugfixes for free is appalling. You are not obliged to install every stable kernel. You can install them whenever you want, or not at all. It's up to you: but if there were only very rare releases (or none) as you seem to want, you would have no choice at all.

Stable kernel updates

flussence — Mon, 27 Nov 2017 22:48:10 +0000

>How do you know whether I contribute or not?

The overwhelming King Canute complex you're exhibiting. You're acting like kernel version numbers are some word of power bringing 0-days and system-crashing bugs into existence and then trying to loudly shoo them away in terror as if it'll accomplish anything except make you look like some kind of digital anti-vaxxer; anyone with even the most distant familiarity with any kind of software development knows the opposite is true: bugs *already* *exist* in what you're running *right* *now*, updates are there to cure them.

Likewise nobody's holding a gun to your head to deploy each and every change - how many similar updates have you missed for your browser, or your phone, or your router? Did the world end without them?

Use some common sense, please. If your rollout process depends on upstream's numbering process (not even the actual rate of commits, just some arbitrary number!) being chronically lethargic, your process is *broken*.

Stable kernel updates

bib — Mon, 27 Nov 2017 21:40:50 +0000

So what you are interest in reading is only posts from sycophants?

Stable kernel updates

bib — Mon, 27 Nov 2017 21:39:28 +0000

How do you know whether I contribute or not?

I have no idea whether you do or not and nor do I care.

I am giving an opinion, just like everybody else.

These rapid release do not allow for proper testing. It's not unheard of that a new release has had to be release quickly due to a bug that was introduced.

Stable kernel updates

roc — Mon, 27 Nov 2017 05:28:44 +0000

In fact, information about the security impact of bugs/fixes is deliberately omitted from changelogs and commit metadata.

Stable kernel updates

itvirta — Sun, 26 Nov 2017 18:01:42 +0000

Yeah, hey, thanks for reminding again that the comment forum here is often so unreadable it's just sad.

Stable kernel updates

arekm — Sat, 25 Nov 2017 21:44:03 +0000

Changelogs rarely explain real life impact.

Stable kernel updates

hmh — Sat, 25 Nov 2017 16:37:35 +0000

That bcache issue was present in 4.14 and 4.13, but not on any other long-term branches (such as 4.9, or 4.4).

Stable kernel updates

am — Sat, 25 Nov 2017 15:20:40 +0000

This one fixes a nasty Block I/O bug that resulted in disk corruption ("at minimum") in bcache.

> Michael Lyle (1):
> bio: ensure __bio_clone_fast copies bi_partno

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/...

https://www.theregister.co.uk/2017/11/22/linux_4_14_bcach...

Stable kernel updates

flussence — Sat, 25 Nov 2017 14:48:31 +0000

Do something with your life that would make your mother proud instead of raging against a product you don't even contribute to.

Stable kernel updates

jrigg — Sat, 25 Nov 2017 10:24:33 +0000

Is it too much trouble to look at the changelog and decide if you need to upgrade or not?

Stable kernel updates

Otus — Sat, 25 Nov 2017 07:02:48 +0000

You still haven't explained how anyone would benefit from getting larger patches less often. How would it be better than just upgrading every second time or whatever.

Stable kernel updates

bib — Fri, 24 Nov 2017 21:39:42 +0000

3 days later...

ludicrous...