Given the hype generated by this Phoronix article I decided that the bloody patch (:D) was worth a try. An updated 2.6.36 kernel package is now available in our testing (sabayon-limbo) repository and will be moved to mainline in a dozen of hours.
It looks like we’re the first providing these responsiveness patches for production usage. To know how to use the sabayon-limbo repository please read our wiki.
And then install sys-kernel/linux-sabayon-2.6.36 package and reboot with the new kernel.
Attention: make sure to update your video drivers and other external modules you may have on your system before rebooting. DO NOT try this if you don’t know what you’re doing. The risk of turning your hard drive into a paperweight is close to 100%.
On The Other Hand 
Indeed the first distro who implements the patch. Awesome dudes !!
This is why I love Sabayon the most, you guys don’t hesitate but you just act upon changes.
That CPU scheduling patch so far appears to work very well for me (I already had it from zen-sources). Useful if you use some TTY to do high-CPU-load stuff.
— Additional bits for still bored / unsatisfied powerusers: —
Have a look at systemd if you want to go beyond that patch’s tty groups and also schedule users, processes etc. with cgroups.
Also notable in new kernel patches: Wu Fengguan’s IO-less dirty throttling v2.
Does the patched kernel in the Sabayon overlay have CONFIG_SCHED_AUTOGROUP=n or CONFIG_SCHED_AUTOGROUP=y by default?
There is alternative :
please see
http://www.webupd8.org/2010/11/alternative-to-200-lines-kernel-patch.html
Well, but unless you have a kernel that enables the not-in-vanilla-kernel /sys/fs/cgroup, you cannot use the easier solution on that link.
So you’d need to customize the more complicated Ubuntu solution (the part of it that does create stuff in /dev on every startup).
And either way I think you’d need to do it with some sophistication to do one of the most useful bits – namely to get X.org into its own group.
Can’t really recommend going down that route vs. getting a patched kernel.
patch the kernel or apply other changes, would be recommended for advanced users, the links is references to “ubuntu” but the fact remains that could be applied to any other distribution (if it is linux), obvious, with appropriate changes .
bye
The whole point is that the appropriate changes will be difficult to realize for most people, especially in such a way that the same effect as the kernel patch is achieved.
Certainly more difficult than the alternative of getting a patched kernel.
@Radtoo: I’m not using a PC running SL ATM but, if I recall correctly, CONFIG_CGROUPS_SCHED=y in the stock 2.6.35-r5 and 2.6.36 Sabayon kernel config, isn’t it? So the userspace mod outlined in the blog post referred to by sabayonino (with suitable changes to use /etc/conf.d/local instead of /etc/rc.local in the case of Gentoo/SL) should work, shouldn’t it?
/etc/conf.d/local comes very late in the startup. At that point X.org has already been attached to tty7…
Does that amke much difference, though? Presumably the benchmarks made by Markus Trippelsdorf and documented in his post http://lkml.org/lkml/2010/11/16/392 were performed under the same circumstances (i.e. after editing .bashrc and creation of cgroups post-loading of X)?
Instead of adding the commands to /etc/conf.d/local, what would stop one putting them in a new initscript /etc/init.d/cgroups and adding it to the default runlevel (rc-update add cgroups default)?
To answer my own question now that I have tried it, it is indeed possible to use an initscript instead of /etc/conf.d/local — see http://forum.sabayon.org/viewtopic.php?f=59&t=22330&start=10#p125773.
^Afterthought: maybe CONFIG_SCHED_DEBUG=y is also needed? Or is CONFIG_CGROUPS_SCHED=y sufficient?
@Fitzcarraldo: Well, you even documented the whole thing, something which I couldn’t be bothered to do. So I am not going to complain much more, then. 🙂
But if you want to actually ensure openrc starts this service before xdm rather than just “some” time in the default run level, I think you’d need something like
depend() {
need localmount
before xdm
}
in your init script. And the rest should belong into start(), Gentoo requires it (why? no clue… but it may be unwise to ignore): http://www.gentoo.org/doc/en/handbook/handbook-x86.xml?part=2&chap=4 , right below code listing 4.1.
Very true. Nevertheless, it’s working as intended and I won’t bother tinkering further: there is no big incentive for me to do so because my machine runs like a rocket anyway.
However, the kernel guys are still evolving the patch and have now moved from TTYs to sessions: ‘[PATCH v4] sched: automated per session task groups’ (http://lkml.org/lkml/2010/11/20/91). Now I’m more interested, because I think this could be more relevant to normal desktop users.