Chk this out ppl http://dontdospamhere.com/

Duplicate threads of this closed - see here for some responses.

Pity they weren't merged here.

A user called K-Z did have an identical thread in ubuntu forums (this am my time) on the same thing, but I cannot access it now. Using their search function got zip, suggesting it got erased?! The last time K-Z talked about shedulers was november then december last year. That's not bad going actually.

Presumably the implementation of a (genuinely) novel scheduler would be quite a curiosity. Interesting to see what K-Z has to say.

Thnx for the replies to all.

First of all....I would like to say that I am very new to Linux and all stuffs related to it like ur forums. So I apologize for posting multiple threads.

Second, the paper signified by Simon (http://www.academypublisher.com/ijrt...0202244247.pdf) has been written by me under the guidance of my prof. I am a final yr graduating student who just opted UFS as final yr project coz of its novelty which I hope it is. And if not...it hardly matters....

Third, the link in http://ufslinux.blogspot.com/ has been corrected long ago.....my fault that it was a kind of spam initially...I apologize...especially to syg00.

Forth...I dont have much idea about licenses and all. All I referred http://www.gnu.org/copyleft/gpl-howto.html and tht too just formally. If there are other things...please point me out.

Fifth..I am not a Linux or technical expert who would work on i7 or other similar stuffs for scheduler testing...I am a simple student....thts it

Sixth..I just want to distribute it to u all and want ur views over it, on its novelity, quality and usability. I used the blog for this purpose only. Its all ur wish otherwise.

Thnx

Your best bet is to distribute it as a free software project. That way you don't need to be an expert. You'll want to use GNU GPL v2 to be compatible with the kernel (iirc) - you need to read it and make sure you understand the terms.

Syg00 will be able to advise you on how best to get advise where it counts - probably on the kernel mailing lists.

1 members found this post helpful.

Thnx Simon...thts a good advice. Can u give ur views abt UFS, if u dont mind??

Hi. Can u plz help me to distribute UFS (http://ufslinux.blogspot.com/) among linux users. I dont have any experience of this. Also, can u plz tell me if its licensing ok? I hope tht wont take much of ur time to answer...

You must start by writing the best english you can. Developers are particular about this - when you deliberately use ad-hoc abbreviations like you'd do on a cell phone or leet speek you only harm your projects image.

This can be tough on second-language=english people - but look like you are trying.

For background:
Read the "advise" link in my sig.
Read "The Cathedral and the Bazaar" by Eric Raymond.

Your scheduler is supposed to go in the kernel right?
Creating it as a patch against the latest vanilla kernel, then submit the patch to the kernel project.

Since you are academic, you should also be able to access and read the latest scheduler papers and compare with yours. Make sure you are still current. Also read about the current schedulers (iirc there are 7 or so but CFS is popular):
http://kerneltrap.org/node/8059
http://www.osnews.com/story/7623
... these are easy-to-read, you can find the technical stuff from there.

See also:
http://kerneltrap.org/
http://tldp.org/LDP/khg/HyperNews/get/khg.html
http://kernelnewbies.org/KernelHacking

From what I see, I cannot test your scheduler because I have a dual-core machine and your code is for a single processor. I cannot comment on the underlying ideas because I don't know enough about scheduling. To me it looks like the basic ideas are already implimented in CFS but you will know better than me.

As-is, it does not look useable - but all free software projects start out like that.
See if you can interest actual scheduler devs.

1 members found this post helpful.

In addition to (or further to) what Simon wrote above, here's two tips/ideas:

1) Check the MAINTAINERS file inside the kernel source, and see if there is a particular person or people who specifically maintain this part of the kernel. Contacting them may get you a better idea what exactly you should do first, with regards to getting your code considered for the kernel.

2) Patches submitted to the kernel (or maintainers) must be made against the latest "Linus GIT Tree" of the kernel. They will not be accepted or considered if they won't patch cleanly against that tree.

Best of luck,
Sasha

1 members found this post helpful.

A few too many assumptions being made here- myself included.
This is *not* a patchset for the scheduler, but an application that runs in userspace and queries the user how they want to use their machine. Then runs setpriority.

Not the way I would have chosen to do it, but each to their own. I don't currently have a machine available to test on, but will have a look around.
As for disseminating it, the current blog is as good a means as any I guess.

1 members found this post helpful.

Thank u all...