Glimmers From The 2.4 Horizon

Oestergaard writes: "We're finally going pre-2.4! Linus posted this on the LKML (Linux kernel mailing list): >>I just made a 2.3.51 release, and the next kernel will be the first of the pre-2.4.x kernels. That does NOT mean that I'll apply a lot of last-minute patches: it only means that I'll let 2.3.51 be out there over the weekend to hear about any embarrassing problems so that we can start the pre-2.4 series without the truly stupid stuff. There's some NFSv3 and other stuff pending, but those who have pending stuff should all know who they are, and for the rest it's just time to say nice try, see you in 2.5.x. The pre-2.4.x series will probably go on for a while, but these are the 'bug fixes only' trees. These are also the 'I hope a lot of people test them' trees, because without testing we'll never get to the eventual goal, which is a good and stable 2.4.x in the reasonably near future. Thanks, Linus

Comments

Re:What about Screenshots ?

[Harvey]

Devfs is in, but still experimental IIRC.

Re:What about Screenshots ?

[Teferi]

one word: xcruise. :P
"If ignorance is bliss, may I never be happy.

Re:What about Screenshots ?

[Harvey]

I don't believe ReiserFS is included (and neither are ext3 and jfs?), and if it was, as an AC pointed out below, it would be pretty hard to take a screenshot, a benchmark or a time to reboot comparison would be more accurate.

Re:Is the LKML Public?

[Harvey]

The FAQ is here. Check out subscribe info here.

New features of 2.4

[Virtex]

Anyone anxious to take the 2.4 plunge, but wondering what has changed might want to take a look at http://lwn.net/1999/0819/a/wwol2_4.html . It somewhat dated -- August of 1999 (anyone know where to find one more recent?), but hopefully most of its content will still be valid.

--

Re:New features of 2.4

[zrpg]

A more recent version that I could find, February 7, 2000, is available at: http://linuxtoday.com/stories/15936.html

Re:FIRST POST!

[Anonymous Coward]

What's this talk about Linux 2.4? I thought Linux 6.1 was the latest one out! Why should I use this 2.4 Linux instead of 6.1?

(yes, I understand the word "irony". Do you understand the words "satire" and "humor"? :)

helping with kernel 2.4

[HeUnique]

Alan Cox just posted a list of jobs that need to be finished before final 2.4 could see the light of day...

Any help around with this?

Re:How to work Linux pre-2.4 kernels?

[The Man]

f you can't program you have no business using a pre-release kernel!

Yes and no. Remember that even most kernel hackers aren't familiar with every part of the kernel and can't fix everything that goes wrong. There are kernels that should legitimately be avoided by everybody, and then there are those that pretty much just work. Linus is now aiming explicitly at the latter; the kernels near the end of a development cycle need to get used so that bugs can be found and fixed. With this in mind, it's time for everyone to start using 2.3/2.4pre on all the non-production machines you can find. Bugs won't get fixed if nobody is using these kernels.

In answer to the original question, the best places to look are the Documentation/Changes file, which describes what software you need to build and successfully run the latest kernel, and the mailing list, which is archived at www.kernelnotes.org and other places.

Re:FIRST POST!

[Harvey]

Poor fools. Windows is already at version 2000. Just more proof that this "Open Sores" crap will never make it in the real world. Come and use a real OS with a real version number sometime.

Re:Is the LKML Public?

[Adnans]

Web archive HERE

good idea?

[Ixnorp]

Looking at past development versions I see:
(Ignore stupid typo or date calculations, I got this info from looking at the dates on kernels from kernel.org so I could be wrong)

1.1.13 - May 22 1994 (cant find a 1.1.0 so I'm guessing its about mid-april or there abouts)
1.1.95 - Mar 1 1995
time elapsed: ~11 months

1.3.0 --- Jun 11 1995
1.3.100 - May 9 1996
time elapsed: ~11 months

2.1.0 --- Sep 30 1996
2.1.132 - Dec 22 1998
time elapsed: ~27 months

2.3.0 -- May 11 1999
2.3.51 - Mar 10 2000
time elapsed: ~10 months

So we've run about the typical time elapsed between dev kernel versions but the current kernel version is half that of normal so it makes me wonder if we're ready yet.

.

Wanted: one clue, will accept good to mint condition.

Re:good idea?

[T-Ranger]

It is, because linus said so.

At the begining of this cycle it was made very clear, in lots of mediums from /. to the 'normal' press that the developement cycle was too long. Less features were added to the kernel, and some that are 'almost' ready that would have held up 2.4 just diddnt get in. This will give everybody some new feature now, and make the upgrade less painful.

For you to have gone to the effort of finding those date but missed this philosiphy change.. well I wonder :)

Re:good idea?

[garver]

What are you worried about? If 2.3 looks totally different from 2.1 and 1.3 we must be in good shape! 1.3 was plagued by the release-a-day syndrome (remember the YAGWRs, yet another greased weasel releases?) and 2.1 just bit off more than it should chew in a release. It seems to me that in 2.3 we have a nice set of new features and best of all, we get to see them in a reasonable amount of time.

Kudos to the kernel developers.

Re:good idea?

[warpeightbot]

So we've run about the typical time elapsed between dev kernel versions but the current kernel version is half that of normal so it makes me wonder if we're ready yet.

Yes.

Linus set out to only include a limited set of stuff in the new kernel, so that that thumping huge 27-month gap wouldn't happen again. The fact that there have only been 51 blockpoints is a reflection of the decreased new-feature count. As it is, he'd have liked to have been done by December, but you know how things go... as it is we're getting one a year and being lucky to do so, at this rate.

It's still a helluva job, and everyone hacking on it deserves every scrap of recognition they get, in the form of long green or otherwise.

--
"Uncle Cosmo ... why do they call this a word processor?"

"It's simple, Skyler ... you've seen what food processors do to food, right?"
-- MacNelley, "Shoe"

Re:good idea?

[dlc]

This definitely is an interesting phenomenon, but I don't think it indicates that the new kernel might not be ready to see the light of day. If you think back to who was using Linux in 1994, and who's using it now, or, rather, who was aware of it then and now, there is significantly increased awareness now. I think the shortened development time is due to more eyeballs.

Cthulhu for President!

Re:Status of 2.3?

[CentrX]

Different devel kernels are broken and unbroken all the time. Some hardware configurations will work perfectly, while others will crash and burn. Consequently, it is not definitive that a 2.3. (x + 1) is going to be any more stable than a 2.3. (x)

Chris Hagar

Re:How to work Linux pre-2.4 kernels?

[jdike]

Sorry about the blatant self-promotion here, but if you want to run the new kernel, but don't have a spare machine and are worried about what it might do, check out my user-mode port.
It is the 2.3.51 kernel running in a set of Linux processes. No need for a spare machine, and it doesn't have access to your hardware or filesystems, so it's less worrisome than a native kernel.
Downloads are available here.
Jeff

Huh?

[Daniel]

I was under the impression that COM/DCOM was some sort of an object-oriented system for remote procedure calls (I guess if you're using objects that should be 'remote method calls'..)
CORBA and RPC don't need kernel support. (well, aside from requiring some sort of network layer, but the kernel doesn't have to know about them) What's special about COM that requires it to go in the kernel? Why not put an NFS server and an httpd in the kernel while you're at it..er, nevermind..

Daniel

e2compr?

[Daniel]

Does anyone know what the state of e2compr support in 2.4 will be? I read in Kernel Traffic that it's not ready for the official kernels, but that someone finally explained to the e2compr maintainer how to port it to the new buffer-cache system..but I haven't heard anything that indicates that such porting is actually occuring. I'm starting to think that I may have to decompress my whole hard drive to try this..which may be difficult given that I'm operating close to capacity ;-)

Daniel

Re:So my question for Nvidia Boosters is

[be-fan]

A) I could really care less. I use Windows for any 3D stuff.
B) The new kernel better not have nVidia support. OpenGL in the kernel would not be a good thing. (Or are you talking about the hardware independant DRI driver?)

Re:Mostly agree.

[Toshio]

It could be argued what is right and what is wrong way to do it. I (for nice example) used to be C/C++ freak programmer (I still prefer C for personal stuff, that I try to write cross platform) and hated all those VB/Delphi/Builder "babies". With time I was forced to use other programming environments and languages and I found out each and every one of them has distinct advantages and handicaps (you just don't get REAL caffeine by drinking de-caf, no matter if you're in NYC or in Calcuta).

At the moment I'm on project involving MS backends, MS tools, MS components, ... basically all MS. It's just the policy of customer to use MS only stuff. I can't go to the customer and say that all his systems are piece of crap, since I've been working with them for some 8 years now, I can't tell him to trash all his existing solutions so I can bring in one open source solution, I can't make a product that won't interoperate with existing software. That is privilege of MSs, IBMs, Suns and others.

What I learned is that nothing can be done by hitting your head against the wall. Yes, it might be rotten publicity, but are we employing PR people to take care of our "faces" (used in japanese sense of the word) or are we trying to do something for ourselves! -- Exclamation, not question mark.

I think we're trying to do something for ourselves. KDE has it's licencing quirks, but it's still popular. It obviously didn't die of bad publicity. Gnome has no licencing quirks and it's popular. What's the catch... I can have Corba, I can have DCOM, I can have my own protocols running around wild... The point is that user doesn't care what does it run on as long it runs well. COM, DCOM, Corba, ... are just tools to achieve that. Nothing more. No bad PR. If MS doesn't want to comunicate with other solutions, it's their problem. Right now it's my problem as well... If DCOM/Linux takes off, it won't be my problem any more [I still like MS DevStudio though... No Linux viable counterpart yet :(].

Stuff about kernel 2.4

[be-fan]

I think that those people running the dev kernels could enlighten us about a few things. (Those of use who don't use Linux or don't feel like DLing 12 megs of source and ripping out our current kernel.)
1) Is it an FASTER?
2) How is the stability? Since this is a pre release, it better be pretty stable. The 2.2.0pre series laster 10 kernels or so, so this is fairly close to release.
3) Any new features that would warrent upgrading (aside from the afformentioned speed/stability)
I also have another question. What kind of resource usage are we looking at in this kernel compared to the 2.2x series? I say this because I have yet to see a major OS vendor pull a Be and actually make an already memery efficiant systems use even less memory at the same time it added a bunch of features.

Re:Stuff about kernel 2.4

[PhilBrut]

I think that those people running the dev kernels could enlighten us about a few things. (Those of use who don't use Linux or don't feel like DLing 12 megs of source and ripping out our current kernel.)

You don't need to rip out your current kernel - I have 2.2.14 sitting around as a backup on some of my systems that run 2.3.

1) Is it an FASTER?

Visibly, especially if you have an SMP system.

2) How is the stability? Since this is a pre release, it better be pretty stable. The 2.2.0pre series laster 10 kernels or so, so this is fairly close to release.

I haven't had problems, although things newer than 2.3.48 have been unusable for me for various reasons (not stability related, though). YMWV (your mileage *will* vary), of course.

3) Any new features that would warrent upgrading (aside from the afformentioned speed/stability) I also have another question. What kind of resource usage are we looking at in this kernel compared to the 2.2x series? I say this because I have yet to see a major OS vendor pull a Be and actually make an already memery efficiant systems use even less memory at the same time it added a bunch of features.

The biggest non-speed related advantage is netfilter (a replacement for ipchains that's quite a bit more efficient). Everything else is for performace, scaleability, or both.

On the memory front, it's a mixed bag between taking up less memory or taking up more. The kernel marks more memory as unusable in 2.3 than 2.2 (dmesg indicates that that memory is where ACPI sits, even though I have ACPI disabled on the motherboard).

However, it certainly swaps less, even after several days of use.

WRT memory consider it to be about the same as 2.2.x.

Re:Stuff about kernel 2.4

[CentrX]

Just because it's a pre does not mean that it is stable at all. It merely means that there will be no more non-pending features added. In other words, the focus will be on making it stable, rather than on adding new features. Consequently, this pre/2.3.51 should not be expected to be any more stable than 2.3.50 or any other 2.3.x It is merely the beginning of making the kernel stable for 2.4

Chris Hagar

Re:So my question for Nvidia Boosters is

[Anonymous Coward]

What, you don't use Be for 3d stuff? You use Windows for 3d and you could [sic] care less? Why then do you use any other operating system at all?

Just talking DRI here. As of 2.3.4X only 3dfx and 3DLabs boards may use DRI. However the specs to implement DRI are available for Ati and possibly Matrox. DRI is necessary to give user space processes using Mesa or real OpenGL direct access to the video hardware. When you build a 2.3 kerrnel you'll see what I'm talking about. There's currently no support for Nvidia boards to use the DRI interface. Only Nvidia can enable this because of their hostility regarding open source hardware support. And they haven't been very timely with working Glx modules (try their latest? ugh!) or DRI.

I'd like to know what Nvidia boosters think about the route Nvidia has taken with regard to Linux and open source--are we happy campers? Do we love our closed spec, obfuscated source buddy Nvidia NOW ??? And how do you imagine you'll feel when 2.4.50 comes around enabling, say fast journalling and other goodies, but changes to the DRI interface break direct rendering for Nvidia hardware again? And no one but Nvidia can remedy the situation ? Well Be-fan you may go back to using Windows and be happy, but Windows has no place on my drives.

Maybe Nvidia based hw has no place in an Open Source system.

What about Netfilter?

[Turmio]

Are they going to integrate the code for the Netfilter modules into Linux 2.4 or do people need to download/compile/install separate Netfilter modules if they want to do some NAT'ing? Somewhere I read that the Netfilter team goal is to get it into 2.4 but the current version (0.1.18) doesn't even compile with kernel 2.3.51 source. I guess they have to hurry up if we want to have NAT support out-of-the-box for 2.4...

Re:What about Netfilter?

[Paddi D.]

Most current version of netfilter is 0.90.4. It seems that they are currently meging the stuff into the kernel tree. The last kernel known to work with 0.90.4 ist 2.3.48 (2.3.49 can be made working with a patch from the netfilter list).

Re:Status of 2.3?

[W6BI]

I started using the devel kernels on three of my machines at 2.3.42. It ran fine, a few of the later versions had compile problems for me, so I just skipped them. I'm currently on 2.3.49, and it runs fine. 2.3.50 fails with a compile error. I'll give 2.3.51 a shot.

I think they're more than good enough for non-critical boxes. YMMV, of course.

- me -

Firewall changes

[tilly]

One of the things that was a big headache for a lot of people going from 2.0 to 2.2 was firewalls.

Well one of the changes that people don't appear to be aware of was that it was completely rewritten again.

But relax, the new stuff was designed to be something to be easy to develop stuff on top of. So 2.4's firewall code will transparently work both like 2.2 and like 2.0 did, and there are hooks to do virtually anything you want.

But still if you want to find out what changed, wander on over to the Netfilter page.

Cheers,
Ben

Re:Firewall changes

[UnknownSoldier]

> and I've been kind of hoping/petitioning certain sites like Linux.com to to run an article on transitioning from 2.2 ->2.4::ipchains -> netfilter+iptables

If you want a good starting walk through, you could start here. It doesn't answer all your netfilter setup questions, but it at least its a great start on Linux security:
http://www.ecst.csuchico.e du/~dranch/LINUX/TrinityOS.wri

You can find David Ranch's homepag here:
http://www.ecst.csuchic o.edu/~dranch/LINUX/index-linux.html

Cheers

Version numbers...

[osu-neko]

The problem is, using the current version numbering scheme, it'll take forever to get to Linux 1776.0, which of course is when the revolution really begins. :) (Yes, I realize that was extremely US-centric, not to mention historically inaccurate...)

--

Re:Can't wait to begin spreadin the love

[chialea]

just take a look at the Linux on Laptops page. it was vastly helpful setting up my thinkpad (of course, most IBM's are easy, becasue they use pretty standard hardware -- there's even a Linux driver for the Lucent Winmodem). in any case, you'll get an idea of how hard/easy/impossible it'll be to set up. in most cases, it's pretty easy, especially if there's a CDROM drive built in, (I don't have one -- ftp install is your friend) except for X.

and of course, there are all these helpful /. people around... everyone I emailed about my thinkpad was very helpful (thank you, if you're reading this). nice thing about the Linux community -- there is a Linux community, and people are really nice about helping.

Lea

Re:There's more good news than this mentions.

[Fizgig]

Crap, I think I accidentally moderated this up! I didn't mean to! Posting to remove it.

Re:So my question for Nvidia Boosters is

[be-fan]

A) I don't use Be for 3d stuff because it doesn't have HW accelerated 3D for anything but 3dfx; Just like Linux.
B) I don't like using windows for 3D, but it has the only usable 3D modlers out there. (BTW the guy who designed the Blender interface was on crack.)
C) I love our closed spec obfuscated source buddy nVidia still. I'm a pretty happy camper. My card runs fast as hell, and I have no religious attachment to Linux/Be/etc.
D) I think you missed the subtlety abou the DRI thing. They only DRI compenent that goes in the kernel is the DRI kernel driver. That is accelerator non-specific. Thus it will work with any card, it is just a system for the DRI driver to communicate through the kernel. It is a symantical thing, no DRI support for a particular card acutally goes in the kernel, it is loaded by the X server which uses the hardware independant kernel driver to talk to the hardware. I really don't care if Linux has no place on your drive. If you aren't willing to use a GeForce just because it doesn't run on your precious Linux, thats your problem. Don't, however, blame nVidia for it. Their product kicks ass. Have you ever run 3D Studio on a GeForce? You'd think you were on an SGI! They will have accel. OpenGL on Linux soon. If you have some problem with it being closed source and propriotry, fine. But I'm just sitting here waiting for Redhat 7 and nVidia's super tweeked OpenGL support. (BTW nVidia's drivers won't use DRI.)

Re:FIRST POST!

[anonymous cowerd]

> Windows is already at version 2000.

So what? I seem to remember reading somewhere about a 68000 version of Linux, which would be, like, 34 times cooler and more up-to-date.

But for really a unbeatable version number I don't see anybody anytime soon topping Google.

Yours WDK - WKiernan@concentric.net

Re:Ultra DMA 66 Support?

[Logicon]

2.4 will be the first stable release with UDMA66 support included with the source. 2.3.x (I don't know what the earliest version that had it was) also has it included. There is also a patch for 2.2.x. Make sure to check the documentation there.

As for four IDE buses, yes, they will be supported. I've used both 2.2.14+patch and the 2.3.x with a Promise UDMA66 add-in controller along with the 2 controllers on my motherboard with no problems. The ABIT motherboards are also supported by the drivers.

Hope this helps =)

Same applies

[CentrX]

If a person cannot figure out how to read and assimiliate the abundance of information regarding the downloading, configuration, and installation of either the Stable or Development kernels, what I said above still applies. If they can figure out how to do that, then they know how to compile and install a development kernel, and consequently, what I originally said would not apply to them anyway.

Chris Hagar

Reiserfs Concerns

[Laven]

Hans Reiser very recently posted the following in response to Linus pre-2.4 announcement. This announcement created a largish problem for reiserfs, as this implies a true feature-freeze when reiserfs is so close to 2.3.

I do hope Linus accepts this last minute reiserfs addition. This is one component that would be of great benefit to Linux.

http://marc.theai msgroup.com/?l=linux-kernel&m=95276159801406&w=2
List: linux-kernel
Subject: Re: Linux-2.3.51, and the pre-2.4 series..(can reiserfs get in?)
From: Hans Reiser
Date: 2000-03-10 20:59:05

We now have a working port of reiserfs for 2.3.49, and I am not sure whether you consider us pending. Can reiserfs get in? Putting us in as an experimental file system until we are accepted by the community as known stable is just fine.
Our 2.2 version seems to be accepted by the users on our reiserfs mailing list as stable.

We'll port it to the new 2.3.51 starting immediately, the 2.3.49 version will hit our webserver in a few hours.

Sorry we tweaked longer than we should have, and created inconvenience for you.

Hans

Re:THIS IS NOT NEWS FOR NERDS!

[acb]

If you don't care about Linux, and you don't know what XFree86 is, then what kind of nerd are you supposed to be? Tell us WHAT YOU would want to see on Slashdot.

News on Star Trek.

Curious (Was Re:emu10k1 sources from creative)

[tweek]

Do you have to load the soundcore module first in debian? Just a bit curious. Basically I added this to my /etc/conf.modules

alias sound soundcore
post-install sound insmod emu10k1

And my sound starts up fine. I would like to look at that script though cause right now I just load gmix when I start X and it restores settings just fine but it would be nice to have it done after the module is loaded.

Re:Curious (Was Re:emu10k1 sources from creative)

[kcarnold]

1. All changes to /etc/conf.modules get wiped out when you do a update-modules (some package installations may do that, possibly the kerenel package; I don't know). Make all local changes in /etc/modutils/(whatever file is appropriate). It does what is equivalent to a run-parts in that directory, except it filters by architecture. 2. Depmod should handle all module dependancies, and when modprobe comes and loads the sound module, it'll see the dependancy, and automagically load the depended-on module first. Instead, in /etc/modutils/aliases: alias sound emu10k1 and in /etc/modules (this does not get update-modules-ized AFAIK): sound And you should be set. 3. See the other response to my comment for a real Debian package that does the same thing as my script.

Re:So my question for Nvidia Boosters is

[be-fan]

There is a reason Matrox, ati and 3dfx have taken the pledge to be open. They are either tiny in the case of Matrox, or dying in the case of ATI and 3Dfx. A market leader has no real incentive to support Linux until it becomes big. And when it does, they still won't need to open the source. Corperate users really don't have the time/resources to hack a graphics driver. Second, I'm pretty sure the nVida GL will not use DRI. They made a statement that DRI was not exactly appropriate for their graphics pipelines. Most likely, it wil just be a patch to the kernel. I don't play Quake that often. I do, however, have a problem with not using my hardware to its 100% just because some OSS guys want their precious, divinly inspired idea to succeed. Hey, there is nothing ethically ambigious about my point. I could care less if their wasn't a single Open Source driver on Linux. You may, and I have nothing against that. I just want my hardware to work, and work at 100%.

Re: stable 2.3

[CentrX]

All minor-odd numbered kernels are designated Development, a tree where new features can be added to the kernel without sacrificing stability. All minor-even numbered kernels are designated Stable. Consequently, by the very definition of the numbering of Linux kernel's, a 2.3.x kernel cannot be Stable.

Regardless, the very process by which one would make the 2.3 kernel stable would produce a kernel for the 2.4 kernel tree, and consequently be a Stable kernel.

If all you want is USB support, there are patches to add USB support to the current Stable kernel, if you just look around for them.

Chris Hagar