Slashdot Mirror
A Kernel With Everything
jd writes: "Ever thought the Linux kernel lacked features? Wanted to out-do that kernel guru next door? Well, the FOLK project might just be the answer. 34 additional Linux kernel projects, rolled into one gigantic mega-patch, with more on the way. (* Stability not included.) Projects include the obscure (eg: HP's scheduler plugin system), the arcane (eg: MPLS and SCTP), the bizzare (eg: Software Suspend) and the insanse (eg: VAX & PPC-64 architectures)." Note: this is neither necessary nor called for. It's just a symptom of hackers having fun and poking at boundaries. Don't put this on any production servers unless you are very disgruntled.
This page is interesting and shows the full history of the life of the kernel.
..as budha said.
On the other hand, EvStack seems to have bitten the dust, and I'm not sure I'm brave enough to try turning Berlin into a kernel module.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
when you call it instant on.
"I ping nothing...NOTHING!"
Yeah, but then someone would trademark it,
and sue the people who used it.
KernAll [tm]
dammit people its "kernel" not "kernal".
built in tetris
--- d'oh
When was it that the "necessity" of a given piece of free software became something larger than the author's wish do code the said piece of free software?
Or has the bazaar turned into a board-managed IPO-driven shopping mall and someone forgot to tell me?
Obviously it is not necessary (except for those who had fun putting it together) nor called for (who is there to call?). Just take a deep breath, a spare weekend, a spare machine and go have fun.
uh, PPC-64 is coming to the desktop and other mainstream markets in the form of the G5, folks. is it really insane to start developing for it now? was it insane when Linux supported Itanium the day it shipped while Microsoft only had beta software? no. it was a Good Thing.
Just raise the taxes on crack.
GRSecurity is a similar project. It's a big kernel patch that contains a lot of unofficial security enhancements for a linux kernel. Some unmaintained patches have been ported to latest kernels, other patches have been merged to work together, and some extra improvements have been added. /proc restrictions, setuid capabilities drop, and LIDS.
The most important things are PAX and Openwall to reduce stack smashing exploits,
I'm running GRSecurity since it was announced on Freshmeat on various loaded production servers. It works like a charm. I just found PAX somewhat slow, it's why I dropped it for Openwall.
-- Pure FTP server - Upgrade your FTP server to something simple and secure.
{{.sig}}
...is the patch that, instead of starting /sbin/init on boot, starts /usr/bin/emacs.
Linux supports ACPI and PnP, I have no idea what you are talking about.
"The Funcionally-Overloaded Linux Kernel" seems to be represented by a tool with a poking attachment and a bottle-opener that is too round to work.
I would like to point to jd's quote from the article, "(* Stability not included.)"
Use my userscript to add story images to Slashdot. There's no going back.
King Kong of Kernels:
Hackers love you, and create
Rootkits infernal.
Anybody else thinking of using this like FreeBSD's LINT kernel? You can look at it to find something new, then apply that to a stable kernel for use.
chris@xanadu:~$ whatis /.
/.: nothing appropriate.
Uhh - I seriously doubt Berlin is either desirable or even a possibility for kernel inclusion. That would sort of be like sticking all of X into the kernel (although I don't think Berlin is anywhere near that big yet.) You REALLY don't want to put all of Berlin's stuff into a kernel - that's asking for trouble. You might be able to use KGI to create an in-kernel interface, but considering what happened to the stability of Windows when they tried that I'd recommend against it. Of course, more power to you if you try. There are a couple of X replacements besides Berlin being worked on which are smaller and maybe better suited for kernel work (fewer features, but depending on what you want...): http://dinx.sourceforge.net/ http://pgui.sourceforge.net/ http://microwindows.censoft.com/
"I object to doing things that computers can do." -- Olin Shivers, lispers.org
How does one deal with getting the hardware state restored? If I run a tool which sends a command to a piece of hardware and changes its mode (e.g. hdparm, X), doesn't the kernel need to keep track of this so that if I suspend and then later resume, the resume process can get the hardware back in the state it was at suspend time?
If I suspend with X being displayed, how does the resume get the video card back in the right graphics mode and configured correctly?
This needs to be done for all devices before such a feature is ready for widespread deployment.
Just because it CAN be done, doesn't mean it should!
UNlike in microsoft OS's, where the only real flexibility as far as multiple kernel versions goes is a kernels binary images for seperate architectures [two], and for SMP vs. Uniprocessor [they charge a lot more for the former]. In linux, because the source is available, it's all always optional.
It's not bloat when you can choose yourself whether or not to compile it / patch it into the kernel. Even if some of these things migrated into the mainstream linus kernel, they sure as heck will be configuration options for them to be rolled in or not.
And I personally think that software suspend is a really good idea, moreso then winME which [I think] now has it. The hugely greater stability of linux allows you to have the nearly-instant on convinience of resuming, because memleaks and the like don't gradually destablize the system the longer it runs. I always loved the perk of working on a laptop wherein I could have a bunch of applications open, be in the middle of a gdn session, and hibernate to disk and power off, only to come back to precisely the point I was before when the power is back on. Software suspend alone is going to make me check this out.
---
---
the pen is mightier then the sword. the sword is mightier then the court. the court is mightier then the pen.
Seriously, just use xdm. Same effect, but without doing something so foolish as forcing someone into a GUI. Makes debugging startup problems a lot harder. I can just see it now.
---
---
the pen is mightier then the sword. the sword is mightier then the court. the court is mightier then the pen.
Kernel Hogan!!
Well, even on the project's own webpage it says "Lastly, this project is not intended to be "useful". If the patches used were all "useful", they'd be in the main tree by now, or Alan Cox' branch. This is much closer to the "silly putty" end of things.". So the /. editor didn't just make it up himself. In either case, it's a cool project and hopefully it will serve as a testing grounds so that some of the more interesting patches make it into to the official kernel.
--
Use di and si too, giving you six. Of course, you need to be sure to push those before going into your spiffy texture mapping routine, but I'm sure you can do that.
Exactly, and where might innovation start? With a little experimentation and forward thinking. This project is more useful than people are summing it up as. Pushing the limits of the kernel is a good thing, it may help us see the road ahead and what a particular path may yeild.
Damn, I'd really like to be able to watch The Matrix and listen to it on my Aureal Vortex Soundcard while I wait for HP Scheduler to crash like a cranky bitch.
The next Slashdot story will be ready soon, but subscribers can beat the rush and slashdot the links early!
133t h4xx0r: "What is this VAX doing on a cable modem?"
Make it Mac. Most l33t h4xx0r flee on the very sight of it, guarantee.
No it's not a flamebait. I in fact compliment Mac...
Is this thing just to prove a point?
I know, size doesn't matter...
Careful though, you might get sued by criMoSoft for ifringing on the bloated operating system (tm) patent.
Having fun is the whole point. It just happens that the result is useful for real work (most of the time).
It never occured to me that open source groups operated in the same way as the Hitchhiker's Guide to the Galaxy. The Guide's staff would pretty much spend all day partying, all the actual work was done by people who wandered in off the street, saw something that needed to get done, and did it. More than one open source project that operates with this strategy. Just another way Douglas Adams was ahead of his time.
Anne T was the highly-trained Cisco tech who...
No, she wasn't
I have no reason to doubt roblimo here... Damn folklore will live on forever though...
501 Not Implemented
If one can compile most of these using modules,
why is it bloat? Some even improve performance.
802.1Q VLAN (vlan.1.0.1)
The 802.1Q VLAN protocol allows multiple virtual LANs to reside on a single ethernet cable.
ABI (2.4.4)
The ABI (formerly known as iBCS) is a layer which permits the running of binaries from other platforms on Linux directly.
Alan Cox' Patch Series (2.4.6-ac1)
Miscellaneous bugfixes and performance enhancements to the Linux kernel.
ALSA (0.9.0-b5)
ALSA is the Advanced Linux Sound Architecture, and provides a powerful interface between applications and audio devices.
Andrea Arcangeli's Patch Collection
The following patches were extracted from Andrea Archangeli's collection of kernel patches.
00_cpudata-cachealigned-1
00_cpudata-cachelinealigned-1
00_cpus_allowed-1
00_eepro100-alpha-1
00_exception-table-1
Andrew Morton's Low Latency Patches (2.4.6-pre2)
These patches should reduce the kernel latency (delay) under a variety of circumstances.
Bad Memory (4.7)
A patch to provide support for partly buggy memory modules.
Compressed Cache Support (2.4.5-0.16)
An intermediate layer in the virtual memory hierarchy which caches pages to reduce swapping.
Ext3 (0.0.8)
Ext3 is the Journalling version of the Ext2 filing system. It is compatiable with Ext2, except insofar as Ext2 ignores any journals Ext3 has made.
i2c (2.6.0)
Drivers and system calls for monitoring hardware health.
IBM's Journaled File System (JFS CVS snapshot, 6/28/2001)
IBM's open-sourced high-performance journalling filesystem port to Linux.
IBM's Next Generation POSIX Threading (1.0.0)
An implementation of an M:N threading model. Improves performance of POSIX-threaded applications (particularly in SMP environments.)
lm_sensors (2.6.0)
Drivers and system calls for monitoring hardware health.
Real-Time Scheduler (2.4.4-1.1.patch)
An implementation of a real-time scheduler for Linux.
Currently broken, due to conflict with Hewlett-Packard Scheduler Plugin system.
SGI's POSIX Asynchronous I/O Support (1.3.1-2.4.2)
A high-performance I/O system which reads/writes asynchronously to ensure optimal contiguous throughput.
This has not been completely integrated, as yet.
SGI's XFS (1.0.1)
An advanced 64-bit journalling filesystem (with access control lists.)
This has not been completely integrated, as yet.
Software Suspend
Enables the possibility of suspending the machine. It doesn't require APM. You may suspend the machine by pressing Sysrq-d. It creates an image which is saved in your active swap-space. The next time the machine is booted, the kernel detects the saved image and restores the machine to the saved state.
Ummm... the FOLK thing exist purely for testing purposes. No one is required to install it. Their bloat is available by *choice* (how big was the last Service Pack you downloaded?)
--
I'm sure "SlashdotMedia" will improve on all the wonders that Dice Holdings blessed us all with
Hehe, some of those sound great ;) That IPPersonality patch could entertain me for a bit, and entertain those l33t h4xx0r5 for a long time ;) I wonder what all I could make my box look like... heheheee ;)
l33t h4xx0r: "What is this VAX doing on a cable modem?"
-= rei =-
"This may be presumptuous..." "That's my favorite kind of 'This'."
Never underestimate how much you can learn from a "useless" project like this. When I'm trying to figure something out, I usually have to come up with a "useless" project to play with. For example, right now I have a truly hideous PHP site hiding in a directory on one of my web sites. The site will never amount to anything, but I'm learning a lot about PHP in the process.
This project strikes me as the same sort of intellectual exercise. It will never produce a useful product, but some of the folks involved may do great things down the road.
** The opinions expressed here are my own, and do not reflect those of my employers - past, present, or future**
Blockquoth bonzoesc:
I have moderator access at the moment, and was quite tempted to add to the ``troll'' moderation you've already recieved, but I've decided to reply, instead. I'm not into wasting mod points on games of mod-the-troll.
FOLK is good for one thing, and one thing only: experimentation. And it does that thing extraordinarily well.
They make no pretenses that this is anything you'd actually want to use for anything serious. But, if you want to play around with the bleeding edge, you don't have to forge your own knife any more.
b&
All but God can prove this sentence true.
Then you better start making Micropayments...
sulli
RTFJ.
So, it installs NT for you? :)
---"What did I say that sounded like 'Tell me about your day?'"---
Slashdot is policing what is good software now! Amazing how once a group gets mainstream acceptance, they too join the punditocracy in deciding what is "called for" and "useless". Hey, maybe these guys might maybe just discover something. Or perhaps it's because people get frustrated dealing with the intelligentsia on LKML. Worse than a bunch of professors arguing over grant money.
So, a bunch of people take an entenepenurial (sp.) spirit, and then Slashdot, standard of all things good, takes the time to post something "unneccicary" and "not called for" on their web page. YEAH! 'Cause only Linus' or Alan's kernel is the best one.
Why is it that every time I read slashdot, it gets to be more painful? Need... new... news... source...
Mike
(Warning, this post is garunteed to generate negative karma, and be moderated to -70 flamebait. But that's OK -- I don't care! Screw you guys, I'm going home.)
Cult of Personality
My one gripe with them is that it'd be nice if they could release each of the patches separately as many of the patch writes have stopped maintaining them. As FOLK already make the effort of porting the patches to the latest kernel, it'd be nice if we could use those ported patches on a standard Linus Linux kernel.
Haven't any of you heard of compiling your own kernel? This is a source patch. That means you have access to features that aren't available in the official kernel. It doesn't mean you have to compile all of them in. In other words, the "bloatware" jokes aren't really funny, they just make you look ignorant.
You had me at "dicks fuck assholes".