Towards Linux 2.6

An anonymous reader writes "The impending release of a new stable kernel promises greater adoption for Linux, as it becomes more reliable and scalable over a larger variety of processors. This article highlights some of the changes, both big and small, with some code samples."

Interactive Performance

[GreyWolf3000]

Interactive performance With the new scheduler, the user should not see the system taking longer to respond to things like mouse clicks or key taps, even under very high loads.

One of the biggest problems new users face expecting Linux to be "faster" than Windows, and booting up for the first time to a slow and klunky desktop with either Gnome or KDE. When things get real slow, having the mouse and keyboard unresponsive make the system feel really rough and unpolished. The more I think about this one the more I think 2.6 will start the next generation of distributions that are finally going to feel like an operating system as opposed to a midnight hack session.

Re:Interactive Performance

[dthable]

One of the biggest problems new users face expecting Linux to be "faster" than Windows, and booting up for the first time to a slow and klunky desktop with either Gnome or KDE. When things get real slow, having the mouse and keyboard unresponsive make the system feel really rough and unpolished.

I just skimmed the article, but I hope this doesn't degrade performance for server applications. I could care less if my system feels sluggish when it's running huge batch jobs or high load web processes.

The onl

Re:Interactive Performance

[GreyWolf3000]

I think the kernel has gotten smarter; i.e. it won't reserve cpu time for the mouse/keyboard until the system is already at high loads and the user is trying to do stuff.

At any rate, the amount of resources needed to maintain usability are nothing compared to huge networks of threaded jobs.

My first Linux install. Success!

[scumbucket]

About two weeks ago I decided to try and install Linux on my old K6-2 450mhz machine gathering dust in the basement.
A friend of mine gave me a few cd's that had something called 'Mandrake' on it.

He said "This is supposed to be the most user-friendly 'distro' out there. Give it a try."

So with trepidation about wiping out my beloved win98se install on the old machine, I jumped right in.

On firing up the install disk, the Man-drake installer asked me if I wanted to remove the win98se partition that already e

Re:My first Linux install. Success!

[slittle]

baited breath

Have you been stealing Tux's dinner again? Tsk tsk tsk.

Code samples

[ThyTurkeyIsDone]

This one caught my eye:

int arr[NR_CPUS];

arr[smp_processor_id()] = i;
/* Chris, I'm gonna insert this bit from SVR4 here
so we can sue these commie pinkos later, hehe */
j = arr[smp_processor_id()]
/* Darl, you're a freaking genius! ROFL Chris */

No MMU? Finally!

[grunthos]

including support for MMU-less systems that do not have a hardware-controlled memory management scheme

Well, it's about time-- I've been waiting years to run this Linux thing directly on my original IBM AT. Now I can yank out that Intel Inboard/386 hack I had to add in to get it to work. That has always bugged me.

At least it can just be unplugged with no regrets. Unlike cutting my other original IBM AT case with the Dremel to get that stupid ATX motherboard in there; no going back on that one. Stupid progress.

Re:No MMU? Finally!

[brianjcain]

Seriously, sounds like it would make a great replacement for maintaining uClinux [uclinux.org] separately. iPod on linux [sourceforge.net], anyone?

Serial ATA

[Apreche]

How about some serial ATA support? I just built a new gentoo this weekend, and it does it all. Except for serial ata, which I need.

Re:Serial ATA

[Bernie]

Drivers have existed [linux-ide.org] for some chipsets for quite some time now.

"Affine"?

[Nynaeve]

From the article:
Processors should affine to one CPU and will not bounce between CPUs.
Affine is not a verb! [m-w.com]

Re:"Affine"?

[VaderPi]

Yeah, and I hope that he meant "Processes" instead of "Processors." I was afraid for a moment that he was talking about preventing each CPU from forgetting which one it was and telling the OS, "No, really, I am not CPU0 any more I am CPU3. If you want to talk to CPU0 you should ask CPU2. Unless, of course, CPU2 is really CPU1 in which case you need to look for CPU1."