Could it be the cut down functionality of BusyBox?

You could try installing coreutils-gnu (and set $PATH accordingly), this will replace the BusyBox utils so maybe read into it first.

http://www.webupd8.org/2010/11/alter...nel-patch.html

In just 4 lines?

This is *NOT* a patch to improve performance.

It just balances work differently: with groups. You could already do that in linux although with this patch you include the functionality inside the kernel.

Besides, you'll only notice a difference if you run staff from a TTY's...

has anyone tried out this on the n900? I'd be curious to see if theres any noticeable change in performance.

edit: perhaps a more appropriate question would be can this fix/patch/whatever even be applied to the n900? It calls for editing the .bashrc file but the n900 uses ash instead of the bourne shell.

you are asking for the patch that "optimizes" things by TTY. would the patch work on the n900? type ps. do you see any TTYs? no? okay... not good. go search on google. people talking about how it's disabled in our kernel. so, are we done here? what if it did work? well, then, you do realize, all processes launched by the same terminal, get the same tty, and assigned into one cgroup. things not launched by a terminal get another one. uhhh, so, do you guys have a lot of terminals open or something? i could be wrong, but this seems to be... stupid.

what about the bash script? this might work better, since it is based on sessions, which is the parent of a process group, a bunch of processes. better, because we have those. don't we? it's not in ps either. and it's almost as dumb as using TTY since if you launch two things from the same parent (ie the menu on your phone) it's gonna have the same session. yes, almost everything is going to have the same session. no point.

OH, if only there were some way to assign per process cgroups, like a customizable list of processes we could modify, and assign priorities to!

it's like you guys can't read... or something.

By the way, I don't find BFS makes my browser choppy, but that may be related to my swap being 1.5GiB rather than the default 768MiB?

So, the first one is nice, the second isn't, any reason you don't want the more IO-intensive one to be nice? Of course, nice relates to CPU scheduling, not IO scheduling, for which you could use ionice (e.g. ionice -c 3 apt-get upgrade).

Here's an update:

Well, Lennart Poettering, a RedHat developer replied to Linus Torvalds on a maling list with an alternative to this patch that does the same thing yet all you have to do is run 2 commands and paste 4 lines in your ~/.bashrc file. I know it sounds unbelievable, but apparently someone even ran some tests which prove that Lennart's solution works. Read on!

Basically, Lennart explains you have to add this to your ~/.bashrc file


if [ "$PS1" ] ; then
mkdir -m 0700 /sys/fs/cgroup/cpu/user/$$
echo $$ > /sys/fs/cgroup/cpu/user/$$/tasks
fi


And run the following commands as super user:


mount -t cgroup cgroup /sys/fs/cgroup/cpu -o cpu
mkdir -m 0777 /sys/fs/cgroup/cpu/user


Further more, a reply to Lennart's email states that his approach is actually better then the actual Kernel patch:

I've done some tests and the result is that Lennart's approach seems to work best. It also _feels_ better interactively compared to the vanilla kernel and in-kernel cgrougs on my machine. Also it's really nice to have an interface to actually see what is going on. With the kernel patch you're totally in the dark about what is going on right now.

-Markus Trippelsdorf


Some more information, benchmarks, and instructions for Ubuntu distros can be found here:

http://www.webupd8.org/2010/11/alter...nel-patch.html

http://talk.maemo.org/showthread.php?t=65745

Bigger swap should make it less choppy, though in normal use my N900 takes about 2-3 days of uptime before swap rolls over and swapping starts taking more and more time per request