Alternative 200 Lines Kernel Patch

[rivenathos]

http://www.webupd8.org/2010/11/alternative-to-200-lines-kernel-patch.html

Has anyone seen or tried this yet? I tried it on my 64-bit machine, and I "think" I can tell a difference. If anyone else tries this, please share your results.

[libssd]

I just applied it to an Acer netbook running Ubuntu 10.04; it seems faster, especially for scrolling and videos, but I could be fooling myself. An Atom N270 needs all the help it can get.

However, the explanation of how it works is gibberish to me:

Each task’s signal struct contains an inherited pointer to a refcounted autogroup struct containing a task group pointer, the default for all tasks pointing to the init_task_group. When a task calls __proc_set_tty(), the process wide reference to the default group is dropped, a new task group is created, and the process is moved into the new task group. Children thereafter inherit this task group, and increase its refcount. On exit, a reference to the current task group is dropped when the last reference to each signal struct is dropped. The task group is destroyed when the last signal struct referencing it is freed. At runqueue selection time, If a task has no cgroup assignment, its current autogroup is used.

[rivenathos]

I have not benchmarked anything, but this little adjustment does seem to make things smoother. Maybe it is just wishful thinking, but I know it did not hurt anything.

[sylvainsjc]

Hi,

I applied it on my laptop and on my desktop (LM9 KDE 4.4.5) and it rocks
Applications are not more fast but the fluidity gain is awesome

[SiKing]

I have read a bit about this. One thing that I failed to find: do you need to have a particular kernel version for this?

[rivenathos]

Not entirely sure, but this is supposed to work on recent kernels. There is probably some info deeper in the articles, but I have not researched that far yet. I am running 2.6.35-22-generic on my systems.

[unexistance]

Hi,

Tested on my old laptop (HP nx6120), it does seems faster

Anyway, I believe the alternative patch works with 2.6 kernels

Regards,

[sylvainsjc]

I have read a bit about this. One thing that I failed to find: do you need to have a particular kernel version for this?

I've applied those scripts with my 2.6.32-25 kernel and as I said, fluidity seems really better

[vincent]

I have read a bit about this. One thing that I failed to find: do you need to have a particular kernel version for this?

If I'm not mistaken, damentz has also applied this kernel patch to his Liquorix kernels (for those of you using Debian or LMDE).

[SiKing]

Well I tried it on my LM9 - 2.6.32-21-generic kernel. It seems to mess with the screensaver? Whenever my screensaver kicks in, I press a key to get out of it, and it immediately goes into screensaver again. Have to press a key once more to get out of it. No big deal, but odd?

[azathoth]

You know, I was looking at that thing and it is intriguing but I am already using Liquorix 2.6.36-0.dmz.13 which has the autogrouping patch from Mike Galbraith.
He had to pull it out for dmz.14 though 'cause some folks were still getting freezes during modprobes. But if I am not mistaken, dmz.13 does have that same functionality.
And it also seems to me that you need to be using a vanilla 2.6.36 for that patch to work. I think that leaves the aptosid kernels out too lol.

edit - oh yeah, there is a no patch method to achieve similar functionality here

[vincent]

You know, I was looking at that thing and it is intriguing but I am already using Liquorix 2.6.36-0.dmz.13 which has the autogrouping patch from Mike Galbraith.
He had to pull it out for dmz.14 though 'cause some folks were still getting freezes during modprobes. But if I am not mistaken, dmz.13 does have that same functionality

From http://liquorix.net/debian/pool/main/l/ ... an.tar.bz2 (look in the Changelog file inside the tarball):

linux-liquorix-2.6 (2.6.36-14) unstable; urgency=low

* correct a 2.6.37 CFS backport
* update sched_autogroup patch from Mike Galbraith to 2010/11/18
* add more sched/urgent fixes from Ingo
* add radix tree rcu deadlock fix
* update ABI to 0.dmz.14

-- Steven Barrett <damentz@gmail.com> Thu, 18 Nov 2010 19:53:20 -0600

linux-liquorix-2.6 (2.6.36-13) unstable; urgency=low

* remove inactive_file_ratio tweaks
* remove vmscan backport for 2.6.37 - may have been causing oops on boot
* update ABI to 0.dmz.13

-- Steven Barrett <damentz@gmail.com> Thu, 18 Nov 2010 01:43:53 -0600

So no, I don't think the autogroup patch was removed. Sounds like it was "updated", and it's working for me (whereas dmz.13 didn't), so that's a good thing.

[azathoth]

You know, I was looking at that thing and it is intriguing but I am already using Liquorix 2.6.36-0.dmz.13 which has the autogrouping patch from Mike Galbraith.
He had to pull it out for dmz.14 though 'cause some folks were still getting freezes during modprobes. But if I am not mistaken, dmz.13 does have that same functionality

From http://liquorix.net/debian/pool/main/l/ ... an.tar.bz2 (look in the Changelog file inside the tarball):

linux-liquorix-2.6 (2.6.36-14) unstable; urgency=low

* correct a 2.6.37 CFS backport
* update sched_autogroup patch from Mike Galbraith to 2010/11/18
* add more sched/urgent fixes from Ingo
* add radix tree rcu deadlock fix
* update ABI to 0.dmz.14

-- Steven Barrett <damentz@gmail.com> Thu, 18 Nov 2010 19:53:20 -0600

linux-liquorix-2.6 (2.6.36-13) unstable; urgency=low

* remove inactive_file_ratio tweaks
* remove vmscan backport for 2.6.37 - may have been causing oops on boot
* update ABI to 0.dmz.13

-- Steven Barrett <damentz@gmail.com> Thu, 18 Nov 2010 01:43:53 -0600

So no, I don't think the autogroup patch was removed. Sounds like it was "updated", and it's working for me (whereas dmz.13 didn't), so that's a good thing.

That's absolutely a good thing. I was going by this in our thread at techpatterns-

Posted: Nov 18, 10, 16:50
Can you boot with the noautogroup kernel parameter? If that still freezes then I'm getting rid of the autogrouping patch from Mike Galbraith.

I have both installed on my lmde setup but I haven't used 14 very much yet. Anyway, Damentz rules.

[libssd]

For reasons unrelated to this mod, I had to restore from a backup that I made immediately before applying the mod. This gave me a chance to do some comparisons. In day-to-day usage, one of the most dramatic changes is with scrolling. Before, scrolling through a 20,000 line text file by holding down the PgDn key, the text would continue to scroll for a while after releasing the key. After, not only is scrolling noticeably faster, but text stops scrolling as soon as the key is released. Similar improvements appear to exist with browser page loads, and file copying.

[vrkalak]

I too, applied this 'patch' to my #!Crunchbang (Debian-testing) -Openbox-x64 system . . .
I have a high-end computer with lots of everything (don't need to use a light-weight desktop) I just like them.

Even with everything I use for my work running/opened I never use SWAP or max out my RAM or CPU.
I only, applied this patch to my upgraded 2.6.35 kernel, out of curiosity.
I didn't download the BFS kernel - just added the patch.

At first, I didn't notice any improvement in performance . . . but, now after a day or so . . . my #!Crunchbang/Debian system is even faster than before. Very, very fast ... Noticeably!!

This morning I added this patch to my LMDE on my AAO Netbook with the default 2.6.32 kernel. We'll see how this patchworks there.

[kwisher]

Anyone have any time line when this will be available via Mint Update or apt-get?

[vincent]

Anyone have any time line when this will be available via Mint Update or apt-get?

The kernel-space patch will be implemented in either kernel 2.6.37 or 2.6.38, as far as I know. Essentially, due to Ubuntu's kernel update policy, that means that you'll have for Natty Narwhal / Mint 11 before you'll see this patch in the kernel. The user-space variation of the patch won't be included in the kernel, you'll have to implement it yourself.

[vospy]

Or use the CK patch set!

I have a howto in my forum.

http://www.aofh.us/node/5

works for ubuntu and linux mint... duh...

[ikegzqrl]

Good evening,

Is it still recommended to apply the "alternative to the 200 lines" kernel patch? I was under the impression this fix was included in recent Linux kernels, but when I applied the patch, Linux Mint 13 (KDE) seemed faster than before applying the patch?

Did I just imagine it was faster, or was it? Is this patch now obsolete of is it worth installing it?

Many thanks for your thoughts and reply.

[zerozero]

ikegzqrl,
you have the answer to your question 2 posts above your. and please don't bump topics more than 6 months old without activity (the information might not be relevant)