Kernel patches

raymano · 2007-11-09 12:08:45

Not too long ago I posted a request to the forums and flyspray for adding the fbcondecor patch to the arch kernel:

http://bbs.archlinux.org/viewtopic.php?id=38966

Quite a few other people wanted it as well. Needless to say it was turned down:

Closed by Thomas Bächler (brain0)
Tuesday, 23 October 2007, 12:49 GMT-5
Reason for closing: Won't implement
Additional comments about closing: We don't randomly add features to our kernel, we like to stay as vanilla as possible.

No problem. I saw the point. I moved on and patched my own kernel. However I got really pissed off when I saw the undervolting patch among other patches being added in the next kernel release, kernel26 2.6.23.1-7. It sure looks like features are randomly added to the arch kernel. Doesn't it?

EDIT:
To make this more of a constructive criticism than a rant let me ask the question:
What is the process used by the devs for deciding what patch is added to the arch kernel?

Last edited by raymano (2007-11-09 12:27:24)

Dusty · 2007-11-09 14:09:57

As far as I know, the devs use a combination of their good judgement from experience and an evaluation of the tradeoff between the work required to maintain a specific patch vis the amount of benefit gained from that patch.

Dusty

tomk · 2007-11-09 14:27:40

Now under discussion on the arch-dev-public ML, if you would like to have a look.

Cerebral · 2007-11-09 14:34:46

raymano wrote:

However I got really pissed off when I saw the undervolting patch among other patches being added in the next kernel release, kernel26 2.6.23.1-7. It sure looks like features are randomly added to the arch kernel. Doesn't it?

So because you didn't get the feature you wanted, and someone else did, that makes you mad? I can assure you that patches aren't "randomly" added, and I'm sure if you take a step back you can agree that an undervolting patch has a greater overall benefit than a bootsplash patch.

(-edit- Hm, upon re-reading it sounds like I'm supporting the undervolting patch. I suppose I should add that I have no use for either undervolting or bootsplash. I could do without either, and I recognize that undervolting in and of itself can be dangerous and unstable, however the heat and battery savings shouldn't be overloooked either.

I don't intend to support anything over anything else - my point was that undervolting offers tangible benefits, while bootsplash is simply eye candy. Regardless, I leave these decisions up to the kernel maintainers, as I lack the knowledge and experience to make informed decisions on such things. -/edit-)

Now, that aside, as tomk pointed out we're currently having a discussion on what gets included and what doesn't as far as kernel patches go. Hopefully a solid policy will emerge from this. However, the attitude of "I didn't get what I wanted so now I'm mad!" won't get you very far.

Last edited by Cerebral (2007-11-09 14:54:12)

raymano · 2007-11-09 14:51:57

Thank you tomk. I will follow the discussions on arch-dev-public to understand the patch selection process better.

djclue917 · 2007-11-09 16:41:04

I think that one reason why the fbcondecor patch shouldn't be added at all is the existence of userspace solutions like USplash and Splashy. I've developed the initial implementation of Splashy on Arch (although I don't maintain it anymore and it now uses a wrapper script developed by the current maintainer) and I must say that Splashy can undoubtedly deliver bootsplash goodness.

http://archlinux.org/pipermail/arch-dev … 02843.html

raymano · 2007-11-09 17:22:51

I maybe wrong but as far as I know none of the userspace solutions are able to display a boot splash as soon as the boot option is chosen from the boot loader and some text still needs to get displayed. In case of just using gensplash without the fbcondecor patch the splash screen doesn't show up til after udev has completed.

Last edited by raymano (2007-11-09 18:07:00)

phrakture · 2007-11-09 17:59:12

If it's worth anything to ya, I side with raymano here. His feature was denied with a "we don't do this due to the Arch philosophy" rationale, which seems to have been overlooked with other things (undervolt patch).

Also, not to mention that the undervolt patch gives you the ability to break your hardware with echo. Hooray!

Cerebral · 2007-11-09 18:04:43

phrakture wrote:

If it's worth anything to ya, I side with raymano here. His feature was denied with a "we don't do this due to the Arch philosophy" rationale, which seems to have been overlooked with other things (undervolt patch)

Oh, I understand his rationale, don't get me wrong - I also don't claim to know that much about the kernel or patches aside from a cursory google search I did on the undervolt patch before posting. I just didn't like the attitude it was presented in.

I probably shouldn't have bothered posting. Bad me. *slaps own hand*

raymano · 2007-11-09 18:07:12

The discission not to include the fbcondecor patch is perfectly fine with me. I can always patch the kernel myself. Not a problem. The reason I was originally irritated was that I did not understand the kernel patch selection process. I might have come across too harsh in my original post and for that I appologize.

djclue917 · 2007-11-10 05:22:12

raymano wrote:

I maybe wrong but as far as I know none of the userspace solutions are able to display a boot splash as soon as the boot option is chosen from the boot loader and some text still needs to get displayed. In case of just using gensplash without the fbcondecor patch the splash screen doesn't show up til after udev has completed.

You can actually run Splashy from initramfs via the mkinitcpio hooks. When doing that, Splashy would load even before /etc/rc.sysinit is executed. I haven't tried using gensplash though that's why I don't know if it runs in initramfs fine. The only downside with Splashy running from initramfs is that Splashy still has some initramfs-related bugs (some of which are pretty annoying).

raymano · 2007-11-10 11:29:54

In the first release of FaunOS I was using splashy and I moved away from it because of the bugs you mentioned. On many machines it would just hang the boot process.

tachy · 2008-10-08 10:56:32

Maybe not completely related to the topic but as response phrakture's comment about undervolting:
Without undervolting I will break my hardware (bad notebook thermal design).

Btw I'm running Splashy with no major problems. The only bug is that the splash disappears on battery power for some reason.

bender02 · 2008-10-08 13:03:50

@tachy: you should be aware of the fact that this discussion is a year old.

Arch Linux

#1 2007-11-09 12:08:45

Kernel patches

#2 2007-11-09 14:09:57

Re: Kernel patches

#3 2007-11-09 14:27:40

Re: Kernel patches

#4 2007-11-09 14:34:46

Re: Kernel patches

#5 2007-11-09 14:51:57

Re: Kernel patches

#6 2007-11-09 16:41:04

Re: Kernel patches

#7 2007-11-09 17:22:51

Re: Kernel patches

#8 2007-11-09 17:59:12

Re: Kernel patches

#9 2007-11-09 18:04:43

Re: Kernel patches

#10 2007-11-09 18:07:12

Re: Kernel patches

#11 2007-11-10 05:22:12

Re: Kernel patches

#12 2007-11-10 11:29:54

Re: Kernel patches

#13 2008-10-08 10:56:32

Re: Kernel patches

#14 2008-10-08 13:03:50

Re: Kernel patches

Board footer