Last 2.5.x Linux Kernel Released

Kourino writes "Today on LKML, Linus released 2.5.75, which he said will be "the last 2.5.x kernel from me", and that he and Andrew Morton are going to start a 2.6-pre series soon. While this certainly does mean things could get interesting soon, don't hold your breath about seeing the actual 2.6 for a while; there are still many areas that need work. This essentially means that the development branch is going into maintenance mode, and new features probably won't get in after this point. Changes of note in 2.5.75 include a merge of the anticipatory scheduler from Andrew Morton's -mm tree and updates from several architectures."

Comments

Argh

[Anonymous Coward]

What about Reiser4?

Re:Argh

[Anonymous Coward]

Its not ready yet. Plus its only a filesystem, it can easily go in after 2.6.0 and marked as experimental. When it is ready.

Re:Argh

[Anonymous Coward]

I think he is still working for Captain Picard. Oh no wait, he got his own ship.

Re:Argh

[jonadab]

We already knew reiser4 wasn't going to make the cut, back when
the discussions were taking place about people who did and didn't
get their proposals in while Linus was on that cruise. Reiser
didn't finish in time, so it won't go in until the feature thaw.

This doesn't mean distributions won't add reiser patches to their
versions of the kernels, though. It just means kernel.org won't
carry it at this time.

Easiest way to fix the bugs

[Sabalon]

Just name it 2.6 - everyone will flock to it because 2.even means that it must be a stable release, never mind it's the first release.

Bitching will ensue, and the bugs will get fixed even quicker. Why mess around with all the pre-2.6 stuff, when this is obviously the fastest way to get it all working :)

Re:Easiest way to fix the bugs

[Muggins the Mad]

>Just name it 2.6 - everyone will flock to it because 2.even means that it must be a stable release, never mind it's the first release.

And those who made that mistake with 2.4.0 will continue to ignore 2.6 until it's proven itself stable and not find the bugs anyway.

(I'm not one of them, but I have time to spend
on following dev releases. Not everybody does).

I'm not a fan of the "it compiles, ship it! and we'll fix it in a service pack" mentality.

- Muggins the Mad

Re:Easiest way to fix the bugs

[Bill, Shooter of Bul]

I'm not a fan of the "it compiles, ship it! and we'll fix it in a service pack" mentality.
And I'm not a fan of waiting for beta testing that will never happen before releasing it. It is a thousand times easier to find bugs that have been found. Therefore the method that allows you to find the most bugs in the shortest amount of time is the best method. This is assuming that you are not actually selling the product for a profit. In other words, release the 2.6 kernel because no one important is going to use it until it gets put into a distribution. So there's sort of always a testing period after the release but before most people start using it.
ok, I've managed to completly disagree with myself several times. I guess that means I must be right. Its pretty clear to me that there *isn't* a best solution. So for heaven's sake just do something, it will be better than nothing.

Re:Easiest way to fix the bugs

[Sabalon]

C'mon...it's easy.

We ship it broken (but earlier than expected), we come up with some crappy ass program that people can use to upgrade it, gather all the info on their computer, ..., profit! :)

Re:Easiest way to fix the bugs

[Kynde]

>>Just name it 2.6 - everyone will flock to it because 2.even means that it must be a stable release, never mind it's the first release.

>And those who made that mistake with 2.4.0 will continue to ignore 2.6 until it's proven itself stable and not find the bugs anyway.
(I'm not one of them, but I have time to spend on following dev releases. Not everybody does).

I agree with you, making sure that 2.6.0 will not contain any obvious showstoppers is worth the while. Besides, 2.4.0 ran beautyfully straight away on my laptop aswell as desktop. Those were pretty standard ix86s, but still.

The 2.5.X series have been occasionally runnable, but on and on during it's cycle it's been uncompilable, unrunnable and all other issues. And not that many people have been running them outside of actual developers.

The whole point of releaseing 2.6pres is to get _a_lot_ more testers. 2.5.7Xs are definitely there.

Re:Easiest way to fix the bugs

[Squarewav]

don't worry mandrake will, they have an annoying habit of taking beta versions of linux software such has xfree 4.2.9 and calling it 4.3, they do it with the kernel too the'll take a pre version of 2.4.x add more drivers and and put them in the distro makes it a pain in the ass to install drivers that require the unaltered src

Re:Easiest way to fix the bugs

[op00to]

>> It is a thousand times easier to find bugs that have been found.

Spoken like a true politician!

Re:Easiest way to fix the bugs

[iabervon]

Actually, naming it "2.6-rc1" will be sufficient to get the distros to test it, which is a substantial ammount of testing. On the other hand, it's not nearly ready for even this; getting tons of bug reports isn't helpful when you already know about a lot of important bugs.

And bugs don't actually get fixed more effectively when people are complaining and pressuring the maintainer; remember that it was mid teens in the 2.4 series where the version that was totally broken got released. 2.4.1 was actually reasonably good, and, up to about 10, there are a number of reasonable kernels which work fine under most workloads.

Re:Easiest way to fix the bugs

[jonadab]

> It is a thousand times easier to find bugs that have been found.

This is why you do a release when the supply of known bugs dwindles
to next-to-nothing and the deluge of bug reports is an intermittent
trickle. The kernel guys know what they're doing, I suspect: no
sense doing a release to _find_ more bugs intil the found ones are
all fixed up.

Re:Easiest way to fix the bugs

[netsharc]

And you might as well make 2 versions: 2.6 Full Speed and 2.6 High Speed!

Re:Easiest way to fix the bugs

[Descartes]

The 2.5.X series have been occasionally runnable, but on and on during it's cycle it's been uncompilable, unrunnable and all other issues. And not that many people have been running them outside of actual developers.

I dunno, I've was running 2.5.x for quite a while before I switched to gentoo. I found it to be quite a bit better than the 2.4's I'd rolled and way better than the one RedHat forced on me.

I think you're right that the 2.6pre's are going to get a lot more people running the new kernel. Personally I don't see any reason for a home user not to upgrade at this point. Unless you have some critical stuff going on on your PC, why not give 2.6 a whirl.

Re:Easiest way to fix the bugs

[Kynde]

I found it to be quite a bit better than the 2.4's I'd rolled and way better than the one RedHat forced on me.

Gut feelings. I bet you've never opened up the .src.rpm of recent RH kernels and looked into the patch set within. The RH 2.4.Xs have been way better than stock 2.4s. (I even did some measurements on the UI interactivity and on some other stuff that are important on my desktop feel during sw developement.)

And for a very good reason. There's Alan Cox in charge of what's in there. And trust me, RH patchset has been good.

The 2.5 series is totally another story. But with 2.4, a lot of the other distros understandably imitate the RH kernels.

So

[Anonymous Coward]

Will the 2.6 "stable" kernel series actually be stable?

The 2.4 series had this public cloud of wierd problems hanging over it its entire existence. It seems like 2.4 never really seemed "trustworthy", they kept making huge and highly experimental changes and 2.4 seemed just kind of like a work in progress for its entire run. Will 2.6.0 be totally safe to download and run and install in a production environment, or is that going to be kind of a "well thats still sort of experimental be careful"? And if the latter, why the heck aren't they staying in 2.5 until it's ready for production.

Am I just too paranoid, or do you know what i mean?

-- anonymous and terrified

Re:So

[Anonymous Coward]

The main problems with 2.4 were the VM problems, especially on big boxes. Now with massive amounts of testing and resources especially from IBM, they are very much improved. There hasn't been a mention of a VM problem for quite a while.

2.6.0 I think will be a lot more stable than 2.4.0, but I don't think its for a critical environment. Its more to start getting testers on board. The number of people using 2.6 probably increases exponentially with each point release for the first few, so leave the more critical stuff for a while. Do you really need it?

Re:So

[aussersterne]

All stable kernel series take a while to sort themselves out. Stable series doesn't mean bug-free, it means working toward such as a matter of priority instead of actively adding new bells, whistles and capabilities. Don't install and run 2.6.0 (or 2.4.0 for that matter), give it some time to stabilize. No kernel since 1.2.x has been particularly stable early on.

As for the eventual stability of 2.4.x, I have an SMP file/print/Web/DHCP/DNS server running in one of the labs that I volunteer to run that has been running 2.4.18 since it was released sometime in late February 2002... it has only had one reboot, to replace a UPS whose battery went dead. It runs 24/7 and has crashed/frozen exactly zero times. Average load is generally above 2-3 during open hours.

That's not bad for a "work in progress" kernel!

Re:So

[N7DR]

Will 2.6.0 be totally safe to download and run and install in a production environment, or is that going to be kind of a "well thats still sort of experimental be careful"? And if the latter, why the heck aren't they staying in 2.5 until it's ready for production.

It seems to me that the problem is that the number of people who try to use 2.6.0 will be far greater than the number that try 2.5.x. Therefore, the probability that a whole new set of bugs will appear (probably not major ones, but a fair number of minor ones) is quite high, and there's nothing that the kernel developers can really do to prevent this happening. This is even more true than int the past because of the ever-increasing ratio of Linux users to Linux kernel developers.

Re:So

[Mr Bill]

I started running the 2.4.0 pre release kernels on my desktop as soon as they were available and never had a problem.

I started running the 2.4 kernel on some production boxes around 2.4.6 and never had a problem.

Yes there will be some problems with the code, but unless you use every single feature in the kernel, chances are it will not bite you... I can't remember the last time I had a kernel panic (besides me mis-compiling modules) on a running box. Probably not since the 2.0 days for me.

Re:So

[73939133]

All stable kernel series take a while to sort themselves out. Stable series doesn't mean bug-free, it means working toward such

I dunno--I think 2.4 has actually deteriorated throughout its lifetime. For example, for my hardware, 2.4.20 seems to have serious bugs in USB2, while 2.4.19 is working fine. Also, I have encountered numerous compilation problems in different 2.4 kernels with different configuration settings.

It runs 24/7 and has crashed/frozen exactly zero times.

Yes, in my experience, there hasn't been much flakiness in those kernels--if you manage to find one that configures correctly, compiles without complaints, and boots up, chances are that it will work well. But if my experience is any guide, that's a big "if".

Altogether, I think Linux kernel development is really having some serious problems.

Re:So

[jbardell]

I love Linux as much as the next zealot ( :) ), but there will never be a 100% stable kernel. And Linux has and always will be an experimental work-in-progress, so why should 2.6 be any dif.? All we need to worry about is keeping the bug reports flowing, and the source constantly modified for the best stability/usability we can manage. Version numbers are just that; only numbers.

Re:So

[Your_Mom]

if you manage to find one that configures correctly, compiles without complaints, and boots up, chances are that it will work well. But if my experience is any guide, that's a big "if"

Wow, seriously, what /are/ you doing to your kernel? I have yet to see any kernel not configure correctly, very rarely seen a compile blow up (and 90% of the time that happened it was user error ['make depend'? We don't need no 'make depend'!]) And most of the time it fails to boot is because I don't have the right support compiled in (ReiserFS used to always bite my ass when I first started using it, never compiled support in for it, I just did ext2)

What /are/ you doing, or what platform are you compiling on? I've only seen errors remotely related to what you are talking about when I was playing with ARM processors.

Re:So

[shepd]

Well, I dunno about you, but I'm waiting for the time when I don't have to have this in my rc.S to get DMA functioning on my Si 3112 without crashing (and yes, the sleeps and syncs are required, and yes, I'm running the very latest -ac patch of the kernel [regular branch or not, it makes no difference]):

/bin/sync
/usr/bin/sleep 5
/usr/bin/echo "max_kb_per_request:15" > /proc/ide/hdg/settings
/bin/sync
/usr/bin/sleep 2
/bin/sync
/usr/sbin/hdparm -X66 -d1 /dev/hdg
/bin/sync
/usr/bin/sleep 2
/bin/sync
/usr/bin/echo "max_kb_per_request:15" > /proc/ide/hde/settings
/bin/sync
/usr/bin/sleep 2
/bin/sync
/usr/sbin/hdparm -X66 -d1 /dev/hde
/bin/sync
/usr/bin/sleep 2
/bin/sync

Blech! And that's not the first time bugs in the 2.4 series have locked me solid at boot -- for a time, creating a monolithic kernel with the bttv drivers would lock a system solid.

Re:So

[umm qasr]

90% of the time that happened it was user error ['make depend'? We don't need no 'make depend'!])

You don't actually need a make dep with the 2.5 kernels. So, we dont actually need no make depend!

Re:So

[ADOT Troll]

It will be the processor optimisations that will determine the success of 2.6 Linux. The solutions that Microsoft has embrased with hyperthreading might be a non-starter for Linux. The Intel/MS world view is to optimise compilation for Windows and software emu the 32 bit environment, a difficult if not stupid way to do things.

If the Amd world view of how to achieve 32 bit without emu on a 64 bit platform are to fly then the adoption of AMD by the server world is essential for Linux in the future. Blindly following the Intel/MS lead may lead to kaos. The same as blindly imitating Microsofts functions by reverse engineering, is for programmers.

Re:So

[Error27]

It seems like 2.4 never really seemed "trustworthy", they kept making huge and highly experimental changes and 2.4 seemed just kind of like a work in progress for its entire run.

For many uses 2.4 was fine from the beginning... The 2.4 VM issues didn't really get sorted out until 2.4.14. After that it was pretty good.

There are still tons of minor glitches with 2.5, but nothing huge like the 2.4 VM. The big problems with 2.5 are the IDE layer and the TTY code.

Re:So

[anshil]

Will 2.6.0 be totally safe to download and run and install in a production environment

To quantity of such kernerls is equal to the quantity of unsinkable ships.

totally safe does not exist! And who ever sells you such is unserious.

Re:So

[NerveGas]

If history repeats itself, my experience with 2.3/2.4 was that even late 2.3 kernels were buggy, and definitely not stable. 2.4.1, on the other hand, did me just fine, running perfectly stable on UP and SMP machines for long enough that I had no reason to upgrade until the o(1) scheduler was released.

steve

Re:So

[BrokenHalo]

I don't know about the grandparent poster was doing, but I have run into similar things from time to time. Like 2.4.20 -> 2.4.21 as follows:

make mrproper
make oldconfig... (I like to do my patching after I've got a kernel that works)
make dep clean bzImage modules modules_install

I don't think there's anything abstruse about that, but 2.4.20 works fine, while 2.4.21 panicked on boot-up (unable to mount root fs). Haven't had time to go back and work out why this happened, but it really shouldn't.

Re:So

[minus9]

So you're saying no commercial vendor would ever ship a product knowing it to contain tens of thousands of errors?

Re:So

[lgftsa]

Me, too. I grab either a kernel.org or debian source package of 2.4.21 and build it in the exact same way as 2.4.18/19, and I get the same error. The initrd works fine, but when it tries to pivotroot to /dev/hda(x) where the root fs is, it barfs. Happens on four systems, two at home, one at work and a vmware image as well.

Re:So

[WindBourne]

Yes, in my experience, there hasn't been much flakiness in those kernels--if you manage to find one that configures correctly, compiles without complaints, and boots up, chances are that it will work well. But if my experience is any guide, that's a big "if".
I tried useing kde' nice front-end for config and had simillar experiences. Since I switched back to xconfig, I have had 0 problems.

Re:So

[WindBourne]

This is even more true than int the past because of the ever-increasing ratio of Linux users to Linux kernel developers.
Yeah, but there is also a more systematic approach to testing being done at IBM and OSDL( I would guess this includes some fence sitters suh as HP). The amount of on-going generalized testing is much improved over what 1-2 years ago was. In fact, I suspect that this kernel will be the most stabile release that we have had for some time.

Re:So

[dpilot]

But as they've said, they've flushed about as many bugs out of 2.5.x as they can with the current pool. They need more people running the new kernel in order to find and fix more bugs, hence 2.6-pre

Re:So

[AftanGustur]

Yes, in my experience, there hasn't been much flakiness in those kernels--if you manage to find one that configures correctly, compiles without complaints, and boots up, chances are that it will work well. But if my experience is any guide, that's a big "if".

I know. I'we had the some compile problems also.. The solution is to save your configuration to a file (F.ex. Config-2.4), do a "make mrproper", and then compile the kernel with your saved configuration..("make dep && make clean && make bzImage")
Is has worked every time for me..

Re:So

[cyb97]

The reason it's up and having such a high load is box it's probably rooted beyond recognition...
2.4.18 is a lethal kernel, especially with such a public server as file/print/web/anything-you-can-throw-at-it...

I suggest atleast patching the ptrace-bug or upgrading to a fixed kernel ;-)

Re:So

[BrokenHalo]

,,,Alternatively, is root a filesystem you forgot to compile support for? This really could be caused by a number of things

Remember, I said I used "make oldconfig". The whole point of that is that it shouldn't just "forget" what I built in before.

Re:So

[SubtleNuance]

Ive always found 2.4 solid on my servers and desktops... though i ran it on x86 w/ common hardware, Ive not felt your paint.

Re:So

[bfields]

Remember, I said I used "make oldconfig".

Hmm, but it looks like you did that *after* a make mrproper. You do know that mrproper deletes .config, right? So what you want to do is something like

mv .config ../
bzcat ../patch-2.4.21.bz | patch -p1
make mrproper
mv ../.config .
make oldonfig

--Bruce Fields

Re:So

[The Almighty Dave]

Commercial developers release software with disclaimers like "no warranty" and "not fit for any particular use". This leads me to beleive that testing is not a big concern.

Re:So

[Hard_Code]

I can tell you that I have had nightmarish times compiling kernels due to various obscure chipset bugs that may or may not be causing problems. In the end, the solution is just enabled all of the damn chipset bug fixes, even if you are not sure they apply. Sure, this is an issue of the user not knowing their hardware, but considering that a huge amount of linux installs are on previously OEM machines, which have scant hardware documentation, it seems pretty unreasonable to expect users to know the ins and out of every piece of hardware. Of course I have also done dumb things like disable virtual terminals and then wonder why my screen is blank on boot ;)

Re:So

[jensend]

If everybody figures they won't install and run 2.6.0 but will give it some time to stabilize, it never will stabilize. Since not enough people tested late development kernels, they started calling the kernels 'pre-stable'. Since not enough people test pre-stable kernels, they just put it out as 'stable version.0'. As more and more people start avoiding stable.[012] kernels, they'll have to start bumping the revision number to compensate.

Re:So

[Josh Booth]

Well, download it and find out. Keep your old kernel and cross your fingers and hope that there are no bugs in the filesystem drivers!

Re:So

[aussersterne]

There is a patched, compiled and lilo'ed 2.4.20 waiting, prepared in case this runtime fix crashed the kernel. It didn't, so we're still 2.4.18 and next time we reboot we'll be at 2.4.20.

In the meantime, the box is behind two layers of NAT and handles only internal requests which internact with a database server hosting a several-hundred gigabyte database. The lab is generally full and users use the http server on this machine to access the database server, then print and/or save their results in small personal (not login/shell, purely storage) areas.

Re:So

[Crazy Eight]

I just hooked up a SATA drive last night, booted into a .21 kernel, and found it odd that hdparm is needed to tweak the interface. Any ideas as to why it seems to support nothing better than udma2?

Re:So

[zerocool^]

Yeah, i noticed that when i got sys-kernel/development on gentoo - following their instructions, I ran make dep && make clean bzImage modules modules_install.

Yeah. No make dep needed.

But, holy hell, did 2.5.74 give me problems. I'm back to 2.4.20 now. It would cause just random things where processes would hang and the load would be 203 on the machine when it was doing nothing.

~Wx

Re:So

[shepd]

>In the time it took you to do that script, you could have patched the IDE source to add your SiS chipset to the various internal "blacklists" and throttled it back to ATA-66 with 15k DMA transfers. Seriously, check the source.

So, if it's that easy, why didn't Alan do it rather than telling us to do this?

Just wondering, because you seem to know more about this than him.

Of course, then again, perhaps you didn't notice the sleeps? This interface doesn't like switching to DMA mode quickly, and will lock up. It would probably be a bad thing to have a near 10 second pause inside the kernel per controller. But maybe that's standard procedure now, and I really don't know what's going on.

All I can say is that unlike a lot of people, my SATA is running in DMA mode right now without random lockups, despite this wacky initialisation.

Re:So

[Physics Nobody]

"...it writes less crap on your screen..."

What?! Noooooo! I love the crap on the screen. The crap on the screen is great. I like to show it to people who know nothing about computers in order to make them think I'm working on highly technical computer stuff that they could never understand so they should just leave me alone when in actuality I'm just reading slashdot while I wait for a new kernel to compile. And damn that was a long sentence.

Re:So

[jonadab]

I had a kernel panic just two weeks ago, with Mandrake 9.1. (Of
course, I had re-arranged my IDE cables so that the drives had
different identifiers (e.g., hda became hdc), and I got the panic
while booting before mounting any read/write filesystems, and I
fixed it by re-installing LILO, but nevermind all that stuff, I got
a kernel panic, and I deserve geek points for that, or something.)

Re:So

[greenrd]

That's wrong. make mrproper deletes your .config file.

Re:So

[(startx)]

I'm guessing none of the moderators have read this post by ratfynk, which the clearly labeled troll has copied?

Re:So

[73939133]

What /are/ you doing, or what platform are you compiling on? I've only seen errors remotely related to what you are talking about when I was playing with ARM processors.

I'm just compiling for regular Athlon and P4 machines. I've been compiling Linux kernels since around 1995 (and UNIX kernels since the early 1980's).

Where are the problems in 2.4? There have been compilation problems in the IEEE drivers, inability to use fb consoles after upgrading, bugs introduced into the USB2 drivers, numerous non-ANSI C constructs that GNU C starts rejecting in later versions (gcc-3.3 will not compile 2.4.19), and on and on. I'm sure many of those are documented in the release notes, and people do trip over them.

I can just repeat: my impression is that kernel development is collapsing under its own weight; I think the Linux kernel and kernel source need a major architectural overhaul.

Re:So

[Zarquon]

Well, I had a problem with 2.4.20->21. Turns out the debian package managers switched around the module order in the initrd config.. fun. Well, that's why it's "unstable".

Re:So

[oscarcar]

Don't forget to do this if you need modules loaded at boot-up, so you can access things like your filesystem, etc.

mkinitrd /boot/bzImage-2.4.21 2.4.21

Re:So

[FueledByRamen]

I haven't ever had a problem with 2.4-series kernels, other than one. I've used 2.4.2-2 (RedHat 7.1 stock), 2.4.9, 2.4.16, and 2.4.[18-21].

The _only_ problem I've ever seen (besides the cpqfc driver not having the correct PCI ID for my card - a few minutes with VI fixed that) was that the tg3 driver (Broadcom gigE chipset, found on NetGear GA302T cards) is REALLY buggy under 2.4.20 - as in kernel panicing when bridged with another interface (eepro driver, Intel 10/100 card). 2.4.21 fixed it, and the server has a current uptime of 5 days (upgraded the kernel 2 weeks ago, no crashes, but I had to take it down to install a DLT3 tape drive).

Re:So

[BrokenHalo]

It dies, indeed, but that doesn't matter if you know if you know to put it back :-)

Re:So

[dotwaffle]

make dep && make clean bzImage modules modules_install Why??? I never understood this method. My favourite line is (in Debian): make dep clean bzImage modules modules_install install I never understood the need for (&& make). Why not just put it all in one statement? And why does no-one do make install? copying the bzImage and altering LILO etc, stupid! make install does it for you, and is sooooo much better! HAve you seriously not done that?

Re:So

[zerocool^]

no, i seriously have never done make install on a kernel. I just do what the gentoo docs tell me, cause i figure that it works. I don't have any problem editing lilo.conf, putting the bzImage in the boot sector, and running lilo. Plus, i want to be able to manage multiple kernels.

I have done make dep clean bzImage etc on one line.

In my current kernel, i even turned off module support, so i don't bother make modules / modules_install. If I add hardware, i have to recompile, but, who gives a fuck? I don't need stuff like USB, etc.

~Wx

Re:So

[dabootsie]

And why does no-one do make install?

make install assumes you're using LILO and doesn't work on grub.

Re:So

[dotwaffle]

Grub sucks... Well, I've never tried it... But I've always used LILO and anything else is tantamount to blasphemy!

I'm glad they put in

[toddhunter]

the anticipatory scheduler, because I haven't been expecting them to do that yet.

Re:I'm glad they put in

[ticklemeozmo]

Well, I've been anticipating iT! Sorry..

Re:I'm glad they put in

[SeaEye420]

I must say that the anticipatory scheduler is the most noticeable improvement from the 2.4 kernel. I'm glad it is going in the official tree. It's been rock solid(for me anyway) for months.

Re:I'm glad they put in

[toddhunter]

It follows then the most noticeable improvement for the 2.6 kernel will be that they rename the anticipatory scheduler to something that is just a little bit easier to say.

Re:I'm glad they put in

[anthony_dipierro]

You should have prepared for it anyway, just in case.

Re:I'm glad they put in

[dpilot]

The real question is how the heck did they take the essence of theotimoline and code it into software?

(Asimov reference)

Re:I'm glad they put in

[einstein]

and a pretty obscure one, at that. it's a chemical in a short story that is bonded to a tachyon, when used with a schrodinger's cat type device can be used to predict the future, right?

Re:I'm glad they put in

[dpilot]

Actually no, it refers to a different story.

Asimov did his PhD (Chemistry, he really was Dr. A.) work on solubility constants. He was getting bored out of his skull measuring how long it took various chemicals to dissolve in water, and in response made up a fake substance called 'theotimoline' that dissolved *before* being put into water. He then wrote a fake/humor research paper on theotimoline and sent it into a magazine for publication. (Later there was another story, "Theotimoline to the Stars" that made up some practical uses for the stuff.) He had second thoughts about having submitted the story when he realized that it might be in print prior to defending his dissertation, and some members of his review panel might take offense.

He knew he'd made it when, after the hard questions, one professor asked him a theotimoline-related question.

Re:I'm glad they put in

[einstein]

in the short story, wasn't it used to determine if space launches were going to be sucessful?

Re:I'm glad they put in

[dpilot]

Don't remember. I'll have to see if I still have the book.

Re:I'm glad they put in

[Fenris Ulf]

If I remember correctly, thiotimoline had electron orbits that actually extended into the future/past instead of the traditional three dimensions.

Don't you hate it when people say....

[Goalie_Ca]

"Linux 2.5.x? Pssh. That's old! I run linux 9.0!"

Re:Don't you hate it when people say....

[pv2b]

Linux 9.0? Bah. I run Mac OS X 10.2, because version 10 is newer and better than version 9.

Mac OS X 10.2? Bah. I run Windows 2000...

Re:Don't you hate it when people say....

[sixdotoh]

so then, wait... if you run winXP .... does the alpha value take precendence over the numerical??? even if you convert XP to ascii it doesn't work, AGGGGGGHHHHHH.

slow day...

Re:Don't you hate it when people say....

[ripleymj]

I thought that was 9. Didn't they drop the .0 ? :)

Re:Don't you hate it when people say....

[burns210]

heh, what is even worse is when some new kid comes and says he is "running Unix 4.8" .... *shiver*

Re:Don't you hate it when people say....

[dhart]

The scary thing is, when I read this post, the voice in my head was that of Christian Bale (Patrick Bateman from American Psycho ).

Re:Don't you hate it when people say....

[Arandir]

You laugh, but I've seen it in action. Two guys standing in front of various shrink wrapped distros in the aisle at Fry's, arguing over which one to get, finally deciding that a Distro 7.2 was better than Kernel 2.4. Sigh.

Re:Don't you hate it when people say....

[Haeleth]

Well, that depends on the user, doesn't it? I have 201,585 XP, so it would be a good idea for me to upgrade from Win2k. But if you're still stuck at level 1, I recommend you go and kill some more goblins before considering it.

What about C2 -style auditing?

[StupidKatz]

The SNARE folks say they are working to get C2-style auditing capability in the kernel, since the old hooks were broken/fixed in 2.4.21. This is a big feature that is keeping Linux from being a "serious" player in "secure" environments, such as certain government-controlled areas.

Re:What about C2 -style auditing?

[StupidKatz]

Well, see, I want a shark with a frikkin "laser" beam attached to its head. Also, period inside the quotation marks: very good!

New features probably wont get in??

[Curtman]

So why would they set the feature freeze for Halloween if new features wont be allowed in after mid July?? Or does the feature freeze have nothing to do with features being added? I'm highly confused.

Re:New features probably wont get in??

[phantomlord]

No more large changes are going to take place... just bug fixes, driver updates, etc. Today Linus said he would reject the HUGE (40k+ lines) ARM merge excepting stuff that only touched the ARM specific source (ie, arch/arm) even though ARM doesn't currently compile. The only thing he says must be working out of the box for 2.6.0 final is x86 and he doesn't care if other architectures are broken on release if fixing them destabilizes what's already there.

Re:New features probably wont get in??

[Trogre]

The Feature Freeze did indeed occur in 2002. This is something different, I think referred to as the Code Freeze.

AFAIUI, no major features have been accepted since the Feature Freeze, and from now on, nothing that will majorly alter the code will be accepted. Only bugfixes.

The 2.5.x series could be real cool

[MoThugz]

...if released in the near future as stable. I did play around with the 2.5.69 build and I must say it's somehow feels faster.

However, the new module handling procedures in the 2.5.x series makes some of my frequently used apps to behave strangely (iptables for example)... even with module-init-tools.

Ah well... as long as there's progress :)

maintenance mode

[Anonymous Coward]

Maintenance mode? No, 2.2 is in maintenance mode.
2.4 isn't even in "maintenance" mode yet - it is _the_ stable tree, and its getting new things added to it with each release (slowly, and after being tested in other trees, and RCs). Just recently new ACPI for example.

2.5 is going into "stabalisation" mode, to get it ready enough for 2.6.0 that it won't piss too many people off who try it. 2.5 has been a good cycle and 2.6.0 will be quite stable, but it needs to go through a few 2.6 point releases during which more and more people will start testing it.

Then _2.4_ will go into "maintenance" mode.

Jiffies in 2.5/2.6

[Sexy Commando]

Does 2.5.x use 64bit jiffies for x86? I've seen it mentioned somewhere but forgot. I'd hate to see my uptime roll over every other month.

Re:Jiffies in 2.5/2.6

[TedCheshireAcad]

What's a jiffie?

no troll, i really don't know.

Re:Jiffies in 2.5/2.6

[pv2b]

Linux definition: On most hardware platforms, a jiffy is 10 milliseconds in duration.

In other Engineering and science diciplines there are other definitions of "jiffy".

In English, it means "a short amount of time" as in "I'll do it in a jiffy".

Re:Jiffies in 2.5/2.6

[Anonymous Coward]

Yes it does. Your uptime won't roll over for half a billion years or so.

Re:Jiffies in 2.5/2.6

[Kourino]

A jiffy is the kernel scheduler quantum. The kernel wakes up once every jiffy to update internal state and reschedule processes, basically. There's a constant called HZ that determines how long a jiffy is; on x86 HZ used to be 100, so a jiffy was 1 / 100 seconds. HZ is now 1000 on most architectures.

Re:Jiffies in 2.5/2.6

[zulux]

HZ is now 1000 on most architectures.


Damn, the Linux kernal wakes up and reschedules things more times per second than my first computer could do operations per second.

(TRS-80 if you're curious)

Re:Jiffies in 2.5/2.6

[Daengbo]

Oh, the memories of 4KB RAM (expandable to 48!!), a tape drive, and 50 for loops per second. I learned everything that I know about computers from that one. Later, I hacked on networking through the serial port and am posting this from a browser I wrote for it in trash BASIC. Okay, that last part isn't true, but the Model I was my first computer, and I remember the pain of having to program my own games in both assembly and basic because there weren't really very many for it.
OTOH, it was a very "open source" experience for me because I learned the basics from the small manual they supplied with it and looking at the BASIC code that most of the games were written in.

Re:Jiffies in 2.5/2.6

[faaaz]

In Sweden we have something called "jiffypåsar", jiffy-bags (translated). They come in various sizes (Letter size is most common) and have soft walls to protect contents. I'm sure they exist everywhere, but perhaps with different names?

Re:Jiffies in 2.5/2.6

[oever]

What's a jiffie?

A jiffy is a quick quicky.

That's because

[hackwrench]

you're not an anticipatory scheduler.

A couple notes on hardware

[phantomlord]

Some notes from my personal experiences with 2.5 on certain hardware:

Those of you who want to use the closed NVidia drivers with 2.5 can find the necessary patches here

2.5.71 also introduced a new native mode driver for synaptics touchpads. You'll need to download the X11 driver and I saw it mentioned that the cvs version of GPM has support if you use that as well.

Re:A couple notes on hardware

[dunham]

As of 2.5.73, the synaptics driver had issues on my dell laptop. (Neither it nor the APM BIOS reset it to absolute mode after a suspend/resume cycle.)

Re:A couple notes on hardware

[pair-a-noyd]

I wish I could find a synaptics driver for my serial touchpad.
I have one of those ergonomic weird looking keyboard with a ysnaptics tp built in and I can not turn off the stupid tap to click crap.

I freaking hate it.

Re:A couple notes on hardware

[phantomlord]

if the 2.5 synaptics driver picks it up, I know for sure that you can disable it under your synaptics settings in X. I had the time set too low on the touchpad on my laptop and tap to click didn't work

Dependencies?

[Duncan3]

Sounds like it's about time to try out 2.5.

Is there a "the mile long list of things to update before trying to boot 2.5.x" list?

I remember the "fun" of updating 2.2...

Re:Dependencies?

[uhmmmm]

the main new dependency is module-init-tools, which replace modutils for module loading, etc.

Re:Dependencies?

[r00zky]

read Documentation/Changes, there's the list of software needed to compile/run it, and their required versions.

If you're running 2.4 in an updated box now, you'll only need module-init-tools, me thinks

Re:Dependencies?

[DarkMan]

There is only really module-init-tools, rather than modutils.

The main reason for this change is that there is now a kernel level module loader. This is for various changes, so that drivers will be handled in exactly the same manner whether they are loaded as a module, or included in the main kernel image. This makes a number of problems in driver writing, and a bunch of idiosyncrities just go away. For example, you should be able to load multiple copies of a driver, compiled into the main kernel. Previously, you had to use them as modules to work that trick. This is important in situations like three identical soundcards.

I believe that is the only required (significant and normally needed) chage to userland tools. Other tools will benefit from updating, to support new features, but that's always the case, and not required. Note that the kernel aught to boot without it - just be less useful if you use modules.

It's worth noting that the sound infrasturcture changed from OSS, to primerly ALSA. OSS is still in, but marked DEPRECATED, so at some point over 2.6, you aught to expect to shift to ALSA sound. It aught to be painless - ALSA supports OSS emulation, so you can phase apps through that. I can't think of any other userland level changes for 2.5 (at least, that impact on your average commodity PC desktop / server - If your're using LVM / md stuff, I think that there might be a shift in there).

One fun change is that you shouldn't need to use ide-scsi emulation to drive CD burners anymore (though that'll require updating userland tools). That's a really useful one, particulary for newcomers [0].

[0] Windows actually also does the 'pretend it's a SCSI device' trick too - but hides it a lot better.

Re:Dependencies?

[norweigiantroll]

[0] Windows actually also does the 'pretend it's a SCSI device' trick too - but hides it a lot better.
I've never heard this before, where did you hear that from?

Re:Dependencies?

[GreatOgre]

I've had the problem with a couple of 3rd party CD-burning utils. It was a bitch trying to find the right dll's for that program. Anyhow, one I got my SCSI cd burner it really didn't matter.

Re:Dependencies?

[red floyd]

If you use Cdex, you get that trick exposed really quick!

DISCLAIMER: I use and love Cdex!

Re:Dependencies?

[DarkMan]

Um, rip open your favourite cd burning package. Poke around it till you find the ASPI dll's. Nero includes it's own ASPI managers - others might use Windows own ones, so you might have to break a debugger or similar (hurtin' for ldd there) over them to spot it.

ASPI is a SCSI interface standard. It stands for SCSI Programing Interface - it's designed so that ASPI complient hardware can all use a single driver, for a specific type of device. If it's using the ASPI code, then it's working a SCSI device - at least in emulation.

A couple of links, to back up that cd burners use ASPI in windows:
http://aspi.radified.com/
http://www.nc f.carleton.ca/~aa571/aspi.htm

Re:Dependencies?

[DarkMan]

Um, sorta maybe.

cdrecord supports two different methods of working with ATAPI burners. dev=ATAPI uses a packet writing interface in the cdrom suport (in 2.4.X). However, the program actually uses the sg device (SCSI generic), so yes, it's still thinking it's talking to a SCSI device at some level. However, this is prety damn efficent - there's no full blown SCSI layer in the way. Given that ATAPI is basically SCSI over IDE (hence all the SCSI crap in the first place), this is more or less fine, with one exception.

If you're using the 2.5 new interface, then maybe (don't really know, not played around with it). However, it does support DMA, which is rather tricky to get working over the other methods (too many layers in the way, or solething like that).

Re:Dependencies?

[Ace Rimmer]

It's worth noting that the sound infrasturcture changed from OSS, to primerly ALSA. OSS is still in, but marked DEPRECATED, so at some point over 2.6, you aught to expect to shift to ALSA sound. It aught to be painless - ALSA supports OSS emulation, so you can phase apps through that.

Hmmm... It sounds like the the bad practices like
cat wow.au > /dev/audio
won't disappear for another period. I was looking forward to default ALSA in 2.6.x but till the programs keep using the old API there is no chance for something better than OSS. Opening raw device, what the hell should it be good for?

Call for new API, better integrated with kernel new abilities. It is a shame that Linux really can't use nearly nothing that modern soundcards provide without many glitches....

Projects like OpenAL don't seem to be evolving and things like artsd, esd and whateverd don't do any good...

Re:Dependencies?

[demon]

I remember running SCSI spy tools from Adaptec on a Windows 95 box with a SCSI slide printer and a SCSI scanner plugged into it - the IDE devices showed up as their very own SCSI bus to those tools. So yes, Windows definitely does an equivalent trick, as was mentioned.

Hopefully it will be more stable then 2.4x in smp

[Billly Gates]

I have a feeling I am going to be modded as a troll but I have karma to burn.

I use to be a big fan of linux but the latest 2.4x came with a bad vm in the so called stable release branch and I heard of dismal uptimes for smp systems with 4 or more processors. Infact Debian still uses the 2.2 kernel by default because of the bugs sorrounding 2.4

I am no longer in IT but if I was I would be more in favor of FreeBSD. I heard 5.0 is alot more scalable then the 4.x branch.

Anyway its reputation for those who are not Linux fanatics on slashdot will be better. Linux 2.0 was rock solid. However the quality has gone down hill recently. Yes Linux 2.4 can scale quite well but in real world uses filesystem corruptions, xinet freezing, and kernel panics happen on smp hardware.

Since Linus now wors at OSDL he can now test these features on high end hardware. Linux is stable on pc class hardware but that is all most kernel hackers have to test the kernel.

Saving time

[MarkCollette]

Linus Torvalds and Alan Cox made a joint press release today. In an effort to save time, Linux 2.5.75 will be renamed 2.6.75, to reflect how mature they assume the code is. "We don't feel like bothering with all that 'pre-' crap, so we thought we'd save some time and just jump right into 2.6" reasonned Torvalds. Alan Cox elaborated that "when MS Windows went frm 3.11 staight to 95, they really left us behind. Now that they're at 2003, we've really got to get our shit together to catch up".

Re:Saving time

[Zachary Kessin]

well if you want a higher year number, how about 5763? And it will be 5764 in a few months.

--Zach

How to help test the kernel

[MichaelCrawford]

Back around when 2.4 was released I wrote a couple articles about how to help test the kernel. They are also helpful when evaluating a new kernel for production systems - you should never just run even a stable kernel on a production system, for while it may work OK for everyone else, it may not work for you.

• Why We Should All Test the New Linux Kernel - Japanese
• Using Test Suites to Validate the Linux Kernel - Japanese

The Open Source Development Lab's Japanese facility was kind enough to provide the Japanese translations.

I am looking for translations into other languages for all my Linux Quality Database articles - there are other articles on web application quality and C++ programming, and more will be posted from time to time.

They are all under the GNU Free Documentation License, but for reasons explained in Which License for Free Documentation? I am planning to change the license soon to another one.

Re:How to help test the kernel

[dpilot]

Then the other thing needed is a clear set of guidelines about getting ready to test the new kernel. The 2.4 kernel required a new modutils, and maybe a few other things. We need two things, which of course may already be done, but I haven't looked in the right place:

1: Additional requirements to run kernel 2.6 above and beyond what ships in current distributions, like modutils.
2: Gotchas that will show up using lilo/grub to switch between 2.4.x and 2.6.x, and how to work around them.

Stuff like this needs to be FAQ status.

In my day...

[Valar]

You kids and your new fangled "2.5". Back in my day, the kernel was 0.1, and the only supported boot device was a piece of toast. And we liked it better that way! Stable, unstable? Kids these days are so ungrateful. Back in my day, when linux crashed, not only did it erase your hardrive, but it put you into seizures! But it built character, and that's the way we liked it!

Re:In my day...

[r00zky]

hardrive? you lucky bas****!
Back in my day we booted the kernel from punched cardboards!

Re:In my day...

[serial frame]

Why gee sonny, we did have hard drives back then! I remember it like it was yesterday in that old desert base...But they were nothing like they are now, because we had to suck droplets of water up with a straw and drop them right in the exact location on a spinning merry-go-round! And it was your own fault if you got hit by a bar! And to top it off, you could only do it on a rainy day! Sure, we didn't have any space left over for the userland, but it was good, and we were happy!

Re:Needs work

[Anonymous Coward]

Who modded this "Insightful"? Poor attempt at humour, possibly, but INSIGHTFUL?

Linux is a kernel. Neither the lack of apps nor the windowing system has something to do with anything.

Oh, and it's Windows that has bluescreens, not Linux. Linux has kernel panics and blinking keyboard LED's. (In morse with the correct patches.)

GNU/Linux is the operating system in itself. I'm usually not anal about the distinction, but to post a comment about the GNU/Linux system in response to the Linux kernel is not really appropriate.

Either way, if the parent is to be modded up it should be "Funny". Definitely not "Insightful". Better yet, blow your mod points on something more worthwhile.

Re:Hopefully it will be more stable then 2.4x in s

[Anonymous Coward]

Yeah, FreeBSD 5 being more scalable than 4 isn't saying much, its still probably below 2.4. No troll, look at the archives - some guy couldn't run 3 SETIs and 1 ssh session on his 4 processor box without them grinding each other to crap on kernel locks... and SETI only enters the kernel a couple of times every 5 seconds or so!! FreeBSD 5 is also having stability problems. FreeBSD isn't a bad OS, but come on, FreeBSD 5 isn't ready yet my good chap.

Your next paragraph about filesystem corruption, etc is misinformed, but if you like, show me some postings of problems: I can't prove its stable by making you read through months of archives to find few problems, but you can show me it has problems by pointing to a few posts.

And yes, don't worry about the scalability of 2.5. IBM have been benchmarking it on 32 way POWER4s with 256GB memory, and 32 way x86 with 32GB memory for just about its entire life. SGI has been doing 64-way and higher IA64 global shared memory machines... etc etc.

Oh, and debian uses 2.4 by default I think.

Top 10 New features over 2.4 ....are what?

[spineboy]

I'm happy with the stability of my machine, but what I want to know as an experienced end user is what can I expect from the 2.6 series. How about someone knowlegable listing the top 10 gains over the 2.4 kernel?

Re:Top 10 New features over 2.4 ....are what?

[crimsun]

Here's a good pointer: Dave's "post-Halloween" doc.

Re:Top 10 New features over 2.4 ....are what?

[Kourino]

Well, some of the nice things are:

o New i/o scheduler, which seems to improve a lot of people's desktop performance;
o Better scheduler performance under loads with lots of processes;
o Rewritten scheduling and threading code, which, coupled with Ulich Drepper's NTPL library, greatly improves threading performance;
o ALSA for sound, and AGP 3 support;
o Faster and cleaner framebuffer support;
o Faster CD recording that doesn't need ide-scsi;
o Upgrades for NFS (v4), NTFS, and HFS+, as well as merges of JFS and XFS;
o System-level in-kernel profiling support;
o CPU Frequence scaling
o IPSec

More information can be found in Dave Jones' list of things to expect in 2.6. Personally, I think it's great to see features that benefit both big and small systems.

Re:Top 10 New features over 2.4 ....are what?

[fanatic]

o Faster CD recording that doesn't need ide-scsi;

It's about goddamn motherfucking time. God the IDE CD running SCSI emulation is the biggest piece of shit I've seen on Linux. I have a box that has run rh 7.2 and now runs rh 9 and the only thing that locks this box up is anything out of the ordinary having to do with the CD-r (such as the pathetic piece of SHIT that GNOME calls CD player software - who made this crap the default desktop in RH? [And yes, I know I can run KDE (in fact I use the CD player kscd from that environment) but having heard how RH mucked around with KDE to 'unify' it with gnome, I'm leery of that]). Last night, that junk locked my machine so solid I couldn't login over the network, the Xserver froze (except for the mouse) and I had to hardware-reset it, fsck all the filesystems on reboot [even though I'm using ext3] and then it spent all fucking night resyncing the raid mirrors.

I love Linux, I've been a fantic about it since 1999, but that ide-scsi was terrible fucking hack.

Re:Top 10 New features over 2.4 ....are what?

[Anonymous Coward]

Try reiserfs instead of ext3. The default ext3 mode used by RH and others is prone to losing data... it's basically ext2-plus-two-bits-and-we'll-call-it-better.

I tested ext3 for a while for my prod systems and was determined it make it work well because it's "where we're headed" but I never was quite satisfied with performance or with reliability.

The default resier mode, on the other hand, has never been bad to me.

Re:Top 10 New features over 2.4 ....are what?

[fanatic]

oops. Guess I got carried away. I just hate it when Linux acts like Winblows.

Re:Top 10 New features over 2.4 ....are what?

[hughk]

I use RH, but I stop using it at the ide-SCSI level. I use a locally compiled MPlayer CDRtools at the latest release level and XCDroast. Its a bit more fiddly to configure, but it works well on my older machines. I also have it on a true SCSI machine and again it works ok.

Re:Top 10 New features over 2.4 ....are what?

[Anonymous Coward]

Wow, they don't call you fanatic for nothing!

Well, I've never had any problems with ide-scsi. And I'm not paranoid about Red Hat's KDE patches doing -- whatever the hell you think it will do. Eat your children, maybe? Steal my refrigerator? Yeah. I mean. Calm down?

I steer clear of that whole GNOME/KDE thing. I do most everything from an xterm. CD burning software? cdrecord. CD playing software? cdplay. I use FVWM as my window manager and I don't run any gnome bits, other than the occasional application, run from the command line.

Also, if you hate red hat so much, why don't you switch?

Re:Top 10 New features over 2.4 ....are what?

[Daniel Phillips]

yes, I know I can run KDE (in fact I use the CD player kscd from that environment) but having heard how RH mucked around with KDE to 'unify' it with gnome, I'm leery of that

Have no fear:

xinit /usr/bin/kde3

(you know what to do if you have kde2)

Or if you want to run both at the same time, go to a text console and:

xinit /user/bin/kde3 -- :1

I feel your pain re Gnome. At least the desktop itself and panel don't seem to crash any more. It's getting more useable too, but you need time lapse photography to notice the changes. I'll even go out on a limb and suggest that the closer Gnome moves in the KDE's direction, the more usable it gets, and frankly, I don't have a problem with that.

By the way, Gnome2 boots about 30% faster than KDE3. They're both patheticly slow in that department, though :-)

Re:Top 10 New features over 2.4 ....are what?

[lgftsa]

Strange, I have been using ide-scsi on four different debian systems, and it's been extremely stable. For example, we had to do an emergency CD burn for a launch a few months ago, and between the three CD burners I scrounged, I did 180+ with only 3 coasters in 3.5 hours.

Re:Top 10 New features over 2.4 ....are what?

[fanatic]

Here's a big one, a large plus in my book, that Solaris has had forever: POSIX ACLs and Extended attributes. And here's an place with examples showhing why this is a Good Thing. Hooray!

Re:Top 10 New features over 2.4 ....are what?

[fanatic]

In rh 7.2 I used windowmaker, with much of the heavy lifting done from rxvt (better than xterm, I think) - no sign of it in rh9, which I just installed a week ago. I went with the default wm because oh it's the default and I haven't had time to mess with it yet. I guess my concern with the futzed around KDE is if I want to download anything for it from other than RH, it won't compile. I haven't tried it yet, but I thought that was the issue with what redhat did to try and unify them - did I misunderstand?

GNOME/rh9 has some dissappointments. Where are the terminals? It looks like it's xterm, which I'm not carzy about (text highlighting and scrolling both are non-intuitive), or the pathetic gnome-terminal, which doesn't even have a functional '-e' option, or nothing. I've already made my displaeasure with the Gnome CD plyer known. Where is windowmaker? I see fvwm and fvwm2 and and maybe I'll give them a go.

On the other hand, my vido performance is much better than in 7.2 (I've got some sort of almost-recognized integrated trident chipset). And having Mozilla, Galeon and OpenOffice pre-installed are good things.

cdrecord and cdrdao are what I use for burming and they are mostly OK. The hangs I get seem to be in the gnome cdplayer, and, earlier, in grip under certain circumstances. Once again, it was not the standard application problems - the system locked solid, required a hard reset and fsck on restart (under other circumstances, a hard reset or inadvertent power-cycle almost never wants an fsck - it recovers the journal and keeps on going).

Re: switching - a different distro will not do cd writing w/o ide-scsi. I tried a prior version of mandrake (8.1, I think) and ditched it when I saw that the MTA (postfix) didn't work and wasn't going to without some major screwing around. I've nothing against postfix - moving away from sendmail is probably a good thinkg given it's history over the years, including last year, but I have a policy: if I load up a distro and some core function doesn't work with at most minor tweaks, I move on.

Having discovered Knopppix, I may use an HD install of it to jumpstart me into debian. apt-get sounds awfully nice and I'm not afraid of the command line at all (I've been a sysadmin on and off since '95, starting with, of all things, SCO, then Solaris, running Linux at home and work since '99).

Re:Top 10 New features over 2.4 ....are what?

[fanatic]

ls -l /usr/bin/kde3
ls: /usr/bin/kde3: No such file or directory

Now what?

Also, you say:

Or if you want to run both at the same time, go to a text console and:
xinit /user/bin/kde3 -- :1

Run both at the same time? How would I get back and forth between them?

Re:Top 10 New features over 2.4 ....are what?

[fanatic]

I didn't know there was *any* way to burn CDs on IDE burners without ide-scsi. How does that work?

Re:Top 10 New features over 2.4 ....are what?

[Tet]

I'll even go out on a limb and suggest that the closer Gnome moves in the KDE's direction, the more usable it gets, and frankly, I don't have a problem with that.

Personally, I think they're both heading in the wrong direction. Both are a long way from where they should be, and neither are remotely well designed or thought out.

Gnome2 boots about 30% faster than KDE3. They're both patheticly slow in that department, though

Agreed, which is one of the reasons I use neither. Since fvwm2 does everything I could possibly want (including many things I can't do with either GNOME or KDE), and is 20 times faster, why would I consider switching? Eye candy does not a good desktop experience make...

Re:Top 10 New features over 2.4 ....are what?

[hughk]

There will be. I've been using SCSI and ide-scsi without problems. Note that some players seem to prefer IDE-SCSI when ypu go away from a pure ISO-9660 disk format.

What I don't understand is that at the same time, the developers are talking about changing the way that the IDE drivers work so that they use the SCSI driver backend processing routines.

Re:Top 10 New features over 2.4 ....are what?

[Wdomburg]

Odd how I've got over a million user's e-mail on ext3 partitions without any data loss or performance issues...

Re:Top 10 New features over 2.4 ....are what?

[einstein]

ctrl+alt+f7 and ctrl+alt+f8

Re:Top 10 New features over 2.4 ....are what?

[Le Marteau]

I hear ya. cdrecord is the only thing that has ever locked up my system.

I now only run cdrecord at the console, without any guis... system has not locked since. And if it does, there will be less to fsck because I'm not running gnome/kde.

Re:Top 10 New features over 2.4 ....are what?

[Paul Jakma]

IME, its completely the other way around. ReiserFS is /awfully/ prone to losing data. I know of several people who have lost a lot of data due to reiserfs, and i've had it lose data too (luckily just while testing it). OTOH, i've never had problems with ext3. (NB: to protect your data you want the data=journal option. The default protects just the metadata - which is fine for spool and tmp fs'es).

Re:Top 10 New features over 2.4 ....are what?

[Josh Booth]

One of the worst problems I've had with ReiserFS is that on reset, the wrong data gets written to the wrong files. Just today, my .xsession-errors got written directly to the end of /lib/modules/2.4.21/modules.pcimap and something funky happened to /etc/hotplug/pci.agent (I didn't examine it). Another time, /etc/X11/XF86Config was overwritten by some other file. I just hope that 2.6.0 fixes that. I may eventually switch to ext3.

Re:Top 10 New features over 2.4 ....are what?

[Crazy Eight]

The post that fanatic was responding to gave the impression that you were burning cds without using the ide-scsi kernel module on an ide drive. Like fanatic, I've always thought this wasn't possible. Cdrecord needs the interface presented by a scsi driver -- whether the drive is actually scsi or not.

Re:Top 10 New features over 2.4 ....are what?

[karlm]

What I don't understand is that at the same time, the developers are talking about changing the way that the IDE drivers work so that they use the SCSI driver backend processing routines.

SCSI is a more flexible protocol than IDE (according to my friend that used to do kernel development at EMC). The SCSI back end is probably fairly close to what they want for a generic nonvolatile storage back end. They can increase reliablility and reduce the size of the kernel (both the source and any binary that is compiled for both IDE and SCSI) by refactoring allof the common HD functionality into a generic back end. This may or may not minimally hurt performance in the short run, but will increase performance by increasing development speed in the long run. This would also greatly increase the amount of testing and stress testing on a good chunk of the SCSI back end.

I doubt they want to use some of the SCSI back end for IDE b/c it's SCSI. Rather, they want to use some of the SCSI back end for IDE b/c the SCSI backend is probably pretty close to a generic storage back end.

While we're on the subject, does anyone know of an IDE RAID controller that pretends to be a normal SCSI controller with a single device attached? How much of the normal IDE CPU load gets offloaded into a typical IDE RAID controller? I would assume the easy way to offload most of the IDE CPU load would to have the controller emulate a SCSI controller.

Re:Top 10 New features over 2.4 ....are what?

[Paul Jakma]

yes, thats a really common reiserfs problem. its due to its tail packing feature. (using tails of blocks allocated to one file for storage for other files). Use the 'notail' mount option to get rid of that problem.

Re:Top 10 New features over 2.4 ....are what?

[Bombcar]

Try this for SCSI emulatin' IDE RAID

Re:Top 10 New features over 2.4 ....are what?

[fanatic]

Please dont even try to use gentoo, we don't want you. And rh changed kde to *LOOK* like gnome, and everything will compile correctly if you install the god damn dev packages, ass wipe...

• You must know everything, since you post as an anonymous coward
• I'll use whatever I like. Fuck you, moron.
• Interesting theory that it'll all just work, but I'm pretty sure that redhat changed header files or symbol names or some such to accomplish what they did. There was considerable concern and displeasure about this on the kde lists, as I recall.
• Fuck you. You're a dipshit.

Re:Top 10 New features over 2.4 ....are what?

[fanatic]

I used to use grip... It was pretty clunky. Then I realized not to trust such a frontend when you can just use the stuff that IT uses in the first place.

Sure, this would probably save me some headaches, but it is awful nice to have grip go out to the CDDB server, tell me which tracks are which tunes, let me clikc on them, etc. I guess years of Windoze use has made me a little lazy. I *have* done the cdda2wav thing in the past, though. Gvien grip's problems, maybe a good idea.

This just confirms...

[pv2b]

...what we have known all along.
That Windows XP sucks? :-)

Re:distros with 2.5 ?

[getoblstr]

none that I know of.

Yeah? well what about

[gotr00t]

I know someone who uses Windows 98. Since the number 98 is higher that 10.2, 9, and 2.6 put together, that must be one kickass operating system.

Re:Yeah? well what about

[pv2b]

NEW scientific proof of Windows 2000's superiority over Windows 98!

We apply a highly sophisticated mathematical theorem named division to compute a ratio.

2000 / 98 == 20.4082

This is SCIENTIFIC PROOF that Windows 2000 is 20.4082 times faster, better, smaller, and stabler than Windows 98!

Order your copy today!

Re:Yeah? well what about

[einhverfr]

Well, technically---
Windows 98 is Windows 4.1 (4.0 was 95)
Windows ME was Windows 4.9....
Windows 2000 was NT 5.0
XP is Windows NT 5.1
Server 2003 is 6.0
Longhorn is ?????

This means that Microsoft is doing WHAT with version numbering???

Re:Yeah? well what about

[BJH]

Nah, "Me" is an acronym - stands for "Mostly evil".

Re:Yeah? well what about

[Abstrakt]

Server 2003 is 6.0
Longhorn is ?????

Actually, Windows Server 2003 uses the NT 5.2 kernel. "Longhorn" will be the big upgrade to NT 6.0.

Re:Yeah? well what about

[phorm]

Actually... 2000 was about the only one in the bunch with really notable stability without huge resource requirements. Too bad everyone had XP crammed down their throat before 2000 could get better.

Re:Yeah? well what about

[T-Ranger]

At least the version numbers are increase with time. Sun's version numbers on another hand....

Re:Yeah? well what about

[karlm]

Longhorn is NT 6.66, a marketing ploy to make you think it's only one percent evil from concentrate.

Re:Yes, but does it run Linux?

[Anonymous Coward]

Only with the userspace patches.

Get to it mister...

[AntiGenX]

Well with that pesky job out of the way, Linus should have more time on his hands to feed the hungry masses. Chop Chop mister!

On second thought, after reading the earlier discussion, Developers: "Quick 'n Dirty" vs. "Correct and Proper"?, take your time...

Re:Yes, but does it run Linux?

[crimsun]

Actually yes, it "runs Linux." While it's nothing unique to 2.5, there is user-mode-linux. For 2.5, there is Kexec. It's interesting that at least one co-lo service is hosted on a UML configuration (see the isp-colo list).

Brings new light to "Yes, it runs Linux."

Re:linux is dying

[pv2b]

You're forgetting, that this site is not only news for regular users.

In fact, this was posted in the "Developers" section.

Do developers care? Possibly.
Is it News for Nerds? Hell yeah.
Is it Stuff that matters? For some people.

Ergo, it's /.-worthy. QED.

LKML confirm Linux is Dying!

[Anonymous Coward]

Today LKML confirmed that linux is dying.

Linus was quoted as saying this will be "the last kernel".

This announcement, of course, has the same high validity as the claim that *BSD is dying.

Maybe if I'd written it earlier...

[MichaelCrawford]

I wrote "Why We Should All Test the New Linux Kernel" not realizing that the 2.4 release was just a couple days away. I was very surprised at that.

I protested the release of 2.4, saying its inclusion in distros would cause users to unknowingly run a poor quality kernel, but Linus said the reason he wanted it released was so that it would get more widespread testing.

The "stable" branch of the kernel is perhaps misnamed. Linus gets to release a new kernel whenever he wants, and I imagine he does some testing, but I don't think he puts a stable release through any kind of rigourous qualification, I think it's more like when the complaints about his pre- versions die down a little.

I know it's common for Linus to release stable kernels that are actually quite broken on some non-x86 architectures. People who run Linux on PowerPC use a branch that's extensively patched from Linus' releases.

Both the 2.4 and 2.2 kernels went through a number of releases before they were really usable. I think the reason 2.2 became reliable was that it was smaller and simpler, and fewer people were working on it.

I'm pretty sure a good part of the reason behind the establishment of the Open Source Development Lab and their hiring of prominent kernel developers like Linus and Andrew Morton is to make sure that 2.6 actually does turn out to be enterprise quality. IBM is a big backer of OSDL, and I don't think they want the billion dollar investment in Linux in general to go to waste.

Re:Hopefully it will be more stable then 2.4x in s

[NerveGas]

I only have one 4-processer machine, but the 2.4 kernels were wonderful on it. It got 2.4.3 when 2.4.3 was hot off of the presses, then ran without reboot or any glitch until it was shut down last month for hardware upgrades. At that point, I put 2.4.21 on it just for fun. Uptime was something like 1.5 years at that point. 2.4.3 is timestamped over 2 years ago, so it could be as long as that.

steve

Re:Hopefully it will be more stable then 2.4x in s

[zuralin]

Infact Debian still uses the 2.2 kernel by default because of the bugs sorrounding 2.4

Actually I (not so) recently installed Debian Woody/stable from CD and it installed kernel 2.4.18 by default. You must still be stuck in the times when Potato was the "Stable" Debian.

Party pooper

[smallpaul]

Doesn't it seem strange that of all the open source projects with constant upgrades and lots of user visible changes, the slashdot community gets all excited about a kernel? A kernel! What difference is this going to make to your day to day life, compared to a new version of your mail program or language runtime or desktop environment or editor?

Re:Party pooper

[Kourino]

Kernels are cool if you like hardware. I started learning x86 architecture and assembly programming soon after I started with software, so I've always had a soft spot for hardware-related stuff :D

It can make a huge difference in your day-to-day life if you have (eg) a laptop that could benefit from cpufreq CPU clock scaling, or from improved ACPI support.

I think the specific Slashdot obsession is because it's the package that's actually called "Linux" though -_^

Re:Party pooper

[arkhan_jg]

Well, there's several big differences that affect me that come from a kernel upgrade. The first lot is better hardware support. AGP3 means your top-end graphics card gets used to its full potential. Native ALSA support makes getting your soundcard up and running a whole lot easier. Better support for IDE burners makes it easier to get your burner running well, and faster, in linux without having to sort out SCSI emulation. The second chunk is scheduling algorithm improvements. What this means for me is, I can have a compile going in the background (using 100% cpu), xmms playing without skipping, xmule downloading things, and whatever I'm doing in the foreground, such as mozilla-firebird, isn't affected in the slightest. In fact, if I minimise the compile window, I can't even tell when it finishes any more, there's just no difference in responsiveness. And all of that achieved without having to use nice. This is something that has been recently improved substantially in the 2.4.20 series, and apparently 2.6 holds up under even heavier loads. In fact, the priority scheduling means that linux now responds better for me when i click stuff than windows does (you don't realise how often explorer sticks on the busy icon until you don't have to use it for a while)

Re:Party pooper

[NerveGas]

Wait until you buy some cutting-edge hardware that doesn't yet have full kernel support, *then* see how excited you get about a new release! ; )

steve

Re:Party pooper

[Overly Critical Guy]

Because a new major version of Linux is like a new major version of Windows. More hardware drivers, better performance, and so forth. Are you for real? The kernel is RUNNING those mail programs and desktop environments.

Re:Party pooper

[smallpaul]

New versions of Windows come with large GUI changes and entirely new utilities (e.g. photo management software). You can often tell what version of Windows you are working with just by the look and feel. (at least for the desktop-oriented ones). New versions of Windows also have radically new technologies like COM or .NET. A new version of Red Hat has many more user-visible changes than a new version of the Linux kernel. It isn't that there is anything wrong with the Linux development model, it is just much more modular and incremental so why get so excited about such a relatively small and day-to-day invisible component?

Re:linux problem

[pv2b]

Something is seriously broken on your system if it's taking you 20 minutes to copy a 17 MB file. That just takes me a few seconds on my computer.

I'd suggest you try a mailing list.

something easier to say

[sacrilicious]

perhaps "schedulatory anticipator"

Re:I guess this means that

[Kourino]

Think it'll run Duke Nukem Forever?

Time for some 2.6-pre Linux Hosting...

[rimu guy]

I host a bunch of VPSs based on Jeff Dike's UML (User Mode Linux) project.

One (of the many) cool things UML allows is for you to try out new kernels without having to dedicate a real box to it. Even if you're only dedicating the box to it between kernel swap reboots. Especially if you're not sure if the new kernel will corrupt your precious partitions.

The UML 'host' server can continue to run whatever stable 2.4 kernel you need (in my case 2.4.21).

You SSH from your 'host' server into your hosted UML kernel. Play around, test reliability, fiddle with new features, regression test your apps.

So anyway, I'm off to grab the new kernel and have a play. Maybe even see if there are any crazies out there who want hosting with the 2.5/2.6-pre kernel.

- Peter
RimuHosting - Linux VPS Hosting

Hardware optimisation will be the telling factor.

[ratfynk]

It will be the processor optimisations that will determine the success of 2.6 Linux. The solutions that Microsoft has embrased with hyperthreading might be a non-starter for Linux. The Intel/MS world view is to optimise compilation for Windows and software emu the 32 bit environment, a difficult if not stupid way to do things.

If the Amd world view of how to achieve 32 bit without emu on a 64 bit platform are to fly then the adoption of AMD by the server world is essential for Linux in the future. Blindly following the Intel/MS lead may lead to kaos. The same as blindly imitating Microsofts functions by reverse engineering, is for programmers.

The office desktop lock of MS is not the route that Linux should take, the applied advanced scientific computing and clustering is the best route. When a great scientific workstation can be had for the price of a Linux install on a 64 bit AMD system the business computing world will finally start to wake up and take notice.

Re:linux problem

[Chyeburashka]

Yep, same here. This is on a 450 Mhz PIII, 384MB, 5400 rpm IDE disk running 2.4.22-pre4.

20264 -rw-rw-r-- 1 myname mygroup 20749685 Jul 9 19:37 patch-2.4.22-pre4

$ time cp patch-2.4.22-pre4 testdir
0.01user 0.69system 0:02.19elapsed 31%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (137major+18minor)pagefaults 0swaps

It looks like his Mac box is sick. Even though it's probably fixable, maybe he's just gotten used to the pain, like a familiar nail coming through the sole of a favorite pair of shoes. If he fixed it, he wouldn't have that to complain about.

Re:I guess this means that

[mandolin]

we'll see the 2.6 kernel in Mandrake 23.2 in the year 2019..

Given Mandrake's penchant for advertising "ships with 2.4.xx" when it's really just a heavily-patched pre-release, I'm suprised they haven't already started shipping "2.6".

That said, I still run Mandrake at home.

Bah.whimper snapper..Re:In my day...

[zoid.com]

whimper snapper... In my days Linux version number was 1/0 and the only boot device was toast crumbs. We had to line them up burnt/crisp/crisp/crisp/burnt in order to get it to boot and we were happy. Soon after that Linus migrated to bagels and thigs got a little more stable.

Ugh, another newbie

[wonkamaster]

You just don't get it, do you?

Linux isn't about copying 17-MB files in under 20 minutes. If you want that, run Windows. Duh! Netscape (and Mozilla) are both dogs. Intentionally. Get used to it. If you want speed, use IE. Thanks for saving the rest of your details, since it's probably by design anyway.

What you fail to understand is that the economy is in a crunch. The idle get let go. What's the best way to not get let go you ask? Not be idle. Take 20 minutes to copy a 17-MB file. If you have a TURBO button be sure to press it until the light turns off -- you might even be able to get the copy up to 30 minutes. Maybe you'll even have to work late. That's sure to get some brownie points with the boss!

And what about the people running Linux not worried about getting let go? One word my friend. Coffee. IT guys love coffee. It's the caffeine really, but coffee is the main source. It's like water, or maybe air. And what's the best excuse for getting coffee? Pretend to be working at the same time. That's right, the ol' I'm busy doing the file copy flim-flam. 20 minute coffee break. For the Starbucks run be sure to copy a 50 or 60-MB file, you don't want to seem like you're slacking off.

It's all in the Linux handbook, copied right out of the GPL. It's at the end -- R.S. knows no one reads copyrights anyway. That's how we hide it. So don't tell. There's also a section on the secret handshake, be sure to read that part carefully!

Okay, in all honesty you either have a network duplex problem (assuming copying to/from the network) or bad hard disk/controller. The throughput is too slow to likely be a DMA problem but you should probably use hdparm to check your settings anyway. I'm leaning towards a hard disk problem which would account for why Netscape won't work during the copy.

It's also possible that you are out of RAM. I've got Linux running on a Mac as well, and HD performance under load is quite poor, even with DMA enabled. Copying a 17-MB file should be cachable unless you are really low on RAM. Swapping is a peformance no-no, especially on a MAC.

Get a local Linux guru that knows the performance tweaking commands to look at your system. Or add RAM/replace your hard drive, they're cheap.

Re:Ugh, another newbie

[akedia]

You, sir, are the newbie. You've been trolled.

FINALLY

[Gaccm]

NEW XCONFIG!!

Check out: http://www.codemonkey.org.uk/post-halloween-2.5.tx t

Now, when someone does make xconfig it uses the qt libraries. There is also a make gconfig for all you gnome people. While I like the advancement, it's annoying that even at the deepest level, the kenel, people are forced to repeat functuality for different libraries. While I love the choice, it is just annoying that we so much redundency for these libraries. It seems that programmers are programming more for the libraries than they are for the users. Unfortunitly, I can't think of a way to solve this.
However, it does suck for anyone who uses another window manager and doesn't have/want qt or gnome. I guess they have to live with ncurses.

Re:FINALLY

[misleb]

However, it does suck for anyone who uses another window manager and doesn't have/want qt or gnome. I guess they have to live with ncurses.

Huh? You don't need to run KDE to use a qt based app and you dont' need GNOME to run a gtk based app.

-matthew

Re:FINALLY

[zdzichu]

NEW XCONFIG!!

You can see the screenshot here.

Re:FINALLY

[Kynde]

While I like the advancement, it's annoying that even at the deepest level, the kenel, people are forced to repeat functuality for different libraries.

You have gotta be kiding! Forced? Ofcourse there are config and menuconfig for actual usage. The xconfig option has always seemed like a joke to me, it's there (with it's graphics library dependencies) for all those graphics junkies.

Your post is funny but not interesting...

Re:FINALLY

[samhalliday]

damn right... im glad `make menuconfig` is staying in, i was actually getting worried that its support was getting ripped out entirely! i heard rumours, but i never actually tested 2.5.x so i couldnt confirm them. phew! well, in hindsight, the rumours were stupid and i was a fool for even considering them to be true: who in their right minds would require a GUI library to build the kernel :-/

Re:FINALLY

[CentrX]

Isn't it used for the new make xconfig?

Re:FINALLY

[Fenris Ulf]

There's always 'make config' and 'make oldconfig'...

I must be the only person who still uses that. Well it was the only thing there was back then, we didn't even have documentation for each option, and had to print out the kernel HOWTO ahead of time to know what the options meant.

Re:FINALLY

[samhalliday]

i still use `make oldconfig` between kernel updates, but if you are using `make config`, then god help you!! :-D

Re:FINALLY

[Kynde]

i still use `make oldconfig` between kernel updates

Nothing "still" to it man. It's been the way to do that for ages and it's the way to do it today with 2.5 and will probably stay that way for quite some time.

Re:I guess this means that

[akedia]

we'll see the 2.6 kernel in Mandrake 23.2 in the year 2019..

No, no, no. The proper troll is:

"I guess this means that we'll see the 2.6 kernel in the Debian stable branch in the year 2019!"

(That said, I use Debian exclusively. Man, you kids today...)

Re:Hopefully it will be more stable then 2.4x in s

[meowsqueak]

It's a precaution rather than a requirement.

Re:distros with 2.5 ?

[Lobo93]

Try Gentoo You'll have to compile the 2.5-kernel yourself, but the distro will prep the system for you, among other things ;)

Re:Hardware optimisation will be the telling facto

[Korpo]

Though we're bordering on offtopic here, because this discussion isn't as closely related to the kernel as it could be, I'm fairly convinced needs no focus as you imply.

First of all, the Linux kernel is and will be the most important readily available high performance computing platform. I cannot imagine a design decision with more than temporary character that will slow down the kernel. Through constant improvement it will lead on all 64bit platforms, Dec Alpha, PowerPC, IA-64 and x86-64. We all know, in the long run, open source isn't beatable in improvement. The kernel is already far on the right side of that curve.

Now, should Linux developers at large focus on scientific computing, or the desktop, on both? Actually this is a "no-question". The development force of open source will always distribute itself along its own best interests, not because of what anybody told them. Till now the technical gurus of programming turned the core of the GNU/Linux OS in what it is, but the evergrowing developer community is attracting more and more apps developers (they are simply more readily available). So while the kernel project is readily scaling to bigger and bigger feats, the app world will still aim for the desktop, the poweruser's desktop first. Simply because there are many people that want to provide apps and simply will do. This will not impair kernel development in any way, and anyhow those people have no different needs from the kernel as the scientists have: stable, efficient, robust.

Since the POSIX and other standards strongly decoupled OS internals from the apps developers (what's going on behind the scenes is no business of the apps developer) we have the power to do it both, in parallel, with no friction.

".config"uring 2.5.75

[Mark_MF-WN]

Does anyone know of a website or mailing list on which to find a good .config file for building 2.5.75? I'm not much of a kernel-hacker, and it would be very helpful for me (and I'm sure for a lot of other people too). Just something general that will produce a kernal that can boot and will run Gnome, Mozilla, and Java.

Re:".config"uring 2.5.75

[aiyo]

Most distributions keep a copy of the .config file youre currently using. You can probably copy that over to your 2.5 source folder and compile away.

Re:".config"uring 2.5.75

[AsparagusChallenge]

...and compile away...

Actually, with the 2.4 series it would be running 'make oldconfig' first, and then compile away.

Re:".config"uring 2.5.75

[Mark_MF-WN]

I've heard that can cause problems because of some of the substantial changes that have been made in 2.5. Any truth to that, or should a bit of manual tweaking take care of it?

Re:".config"uring 2.5.75

[davFr]

Despite other bad-documented replies, you're right that a .config would be helpful to know which modules can't be build. And .config doesn't depend on what your configuration is, but on what you want to have compiled.

Well, my kernel goes to 11

[A nonymous Coward]

So there. All you weenies stuck at 10, I have no pity for you.

About scalability & stability

[vlad_petric]

Linux *does* scale better than FreeBSD in SMP.

Regardless, I've had an uptime of 214 days with Linux on a 2-CPU Pentium 4, with a stock RedHat 2.4.18 kernel. The reason it went down was because the computer had to be moved from one location to another ...

Yes, I've noticed instability on the stable kernel - but that happened mostly with my own kernel builds. That is why we have distros - let them heavily test, use their kernels *and* make your purchasing decisions based on their hardware compatibility lists. When you compile your own configs, there's clearly some risk that you're undertaking (compiling kernels isn't for aunt tillie - sorry Bruce)

Furthermore, don't make empty assumptions about what kernel hackers test their work on :). Trust me, RH & Suse *do* test on high-end hardware. Not to mention the gargantuan hardware that IBM ships with Linux as the OS.

Hardware issue?

[Anonymous Coward]

I have had simular problems in the past (with rh 9 and other linux versions). My problems were due to a buggy motherboard chipset. From what I have seen Linux does not act this way on most hardware.

Re:Hopefully it will be more stable than 2.4x in s

[snero3]

Correct me if I am wrong. don't really need to say that here now do I!!! But arent the other versions of the linux kernel IE ones for alpha, power4 high end hardware etc.. maintained by people other that linus and his cronies?

I also seem to remenber that Linus mentioning that the linux kernel in its current state doesn't scale well past 4 cpus?

The Reason I say this is because I have

Linux 2.4.18 smp #1 SMP Wed Sep 4 11:55:37 EDT 2002 i686 athlon i386 GNU/Linux

and it has been running completely stable at a load average of 2 for about 8 months now, the thing that is buggy is X but the kernel, file system etc stays up even after X has decide that is enough for tonight

Debian compatibility?

[PD]

Anyone working on Debian compatibility packages for the 2.5.x and 2.6.x series? I've been searching around and haven't found anything yet.

Re:Debian compatibility?

[heikkih]

The only compatiblity package needed is the new modutils, and they are already in unstable.

apt-get install module-init-tools

Download new kernel, configure, compile, install and reboot, and try do it all with kernel-package....

Worked perfectly for me (TM), but you may have to do some tweeking if you want some other stuff (like nvidia-drivers).

Just did it. 2.2

[phorm]

Just set up a server, it's 2.2 by default (unless you have an odd ISO).

Debian cd install:
linux (2.2)
bf24 (2.4)
vanilla (2.2 vanilla)
ramdisk0 (disk install 2.2)


Not to say that there's anything wrong with the 2.4 kernel though, often enough it's the only thing I can get certain hardware drivers under, and 2.2 has its issues too. I've noticed that one of our servers was choking on 2.4 (tape system amongst other stuff), whilst another really had issues with stuff in the 2.2 tree.

Re:Just did it. 2.2

[zuralin]

Ahh, well that would explain why, i used the bf2.4 cd. thanks

Re:How sad

[darketernal]

2.5 is a development kernel. It would be STUPID to run production systems on it... Many production systems still run on 2.2 kernels and have been up for years.

Linus is 'giving up' on 2.5 because he wants to make that branch of the source tree stable. Not because it's a flaming lump of shit.

Then again, I've probably just been trolled. But FYI for those people who believe him.

Re:Yes, but does it run Linux?

[Overly Critical Guy]

Geez. What a karma whore. Seriously.

Re:Hardware optimisation will be the telling facto

[ratfynk]

Hopefully the 2.6 kernel will gain rapid acceptance with hardware manufactures, and software writers world wide.

Yes you are right on target as to why Linux is doing great things. Not releasing 2.6 stable too soon is one of them. The problem is that the business world does not appreciate the reasons why .

My wife trains new users for an environment that is typical.
Their server is MS based because of the need to run MS office. The real power apps are served by a Linux server for the medical image data files. There in lies the problem, the office application lock of Microsoft.

The lack of a killer business app in Linux is the root cause of the delema. Sure there are great integrated office apps in Linux. But by and large you would need to pry the dead fingers of the average office computer user away from MS office addiction.

My point was that the world of scientific computing and real powerhouse apps would be the best chance for Linux in business till someone comes up with a new killer business application that is native to Linux. What might fill the Bill (pun intended) is an appliction that does all the necessary buisiness/computer chores but is easily customisable in house and scalable. Inventory data , linked to all departments, with or without cash register interface, and accounting software all linked, a kind of Swiss Army knife that is easy to secure and change. Hopefully this is the future of Linux in business.

When businesses see that using a powerhouse Linux server/workstation setup can make a difference to their bottom line their MS office/OS/server dependancy will quickly evaporate.

Re:Hopefully it will be more stable than 2.4x in s

[NerveGas]

As for scaling past 4 CPU's, that was much more true before 2.4. While it's still not going to scale linearly to an infinite number of processers, improvements along the way in the 2.4 series and the o(1) scheduler have helped out quite a bit.

I do believe that much of the non-scaling of Linux past 4 CPU's is, to some extent, actually the fault of the hardware, as the great majority of multiprocesser hardware has bottlenecks that impeded linear scaling.

As a simple example, take a look at the dual P4 Xeons on the 533 MHz bus. Sounds spiffy, right? Well, you're splitting the 533 FSB and memory busses between both processers, giving each one an effective *266 MHz* when under load. Seeing as how even the slowest single-proc P4 has a 400 MHz FSB, you can see off of the bat that you're hitting bottlenecks!

Look at the AthlonMP series: Each processer has it's own independent bus! However, the only available motherboards have a single-channel, 266 MHz memory architecture. That gives each processer only 133 MHz effective memory bus under load. The simple addition of a dual-channel memory controller (like the one on the NForce boards) would give the AthlonMP's a real shot in the arm.

Now, it may seem like I'm just talking about low-end multiprocesser machines, but here's another example: Even on some of the higher-end machines, there are restrictive bottlenecks. By naming names, I'll only start a flame-war with the zeolots, but suffice it to say that there are $35,000 "high-end" servers that have *less* total memory bandwidth than that $3,000 dual P4 Xeon. That is pretty pathetic!

It's pretty easy to see why someone who didn't realize that could plop $35,000 on a 4-way, big-name machine that had less memory bandwidth than a $3,000 dual P4 machine, see that under load both of them performed similarly, and say "Well, Linux must not scale well."

To make matters worse, the kind of applications that are run on multi-CPU machines tend to be things like RDBMS', which do not lend themselves well to clustering. Here's the catch: Those types of applications tend to be the most memory-demanding. So, take a single P4 with a 533 MHz bus, and install your RDBMS on it. Take a dual P4 Xeon with a 533 MHz bus, and try the RDBMS. You're certainly not going to scale linearly, but that's because you still haven't improved the memory bandwidth.

steve

Re:linux problem

[BenjyD]

Somebody should really write an updated version of that troll.
Although, I guess from the number of responses (and the mod point) it still seems to work.

Re:distros with 2.5 ?

[Spider[DAC]]

actually its fairly painfree
emerge sync
emerge development-sources
(should pull in module-init-tools and set the system ready)

then copy .config from /usr/src/linux to /usr/src/linux-2.5.7?/, cd there and
make oldconfig

Dont forget to enable VGA consoles, Slap in /dev/pts (dont forget your filesystems either) devfs (which has changed, yeeey)

make it install it (no make dep, finally) and copy the files into /boot....

still a bit hairy, but not really compared to building a virgin kernel in LFS ;-)

LWN interview with Linus

[Spider[DAC]]

This might prove interesting.. I found it hard to know where in the discussion to tuck it, but here seems good since Linus touches on the part of "new features" and so on.

here at lwn

Re:Needs work

[LouisvilleDebugger]

Thanks, RMS!

Man some people can bitch about nothing

[vandan]

I've been running a 2.4 kernel since the 2.4-test ones ( I think that's what they were called ) and I have never had any need to rush out and upgrade to the latest version. Sure ... when a newer version comes out I will upgrade for the heck of it, but seriously, if it wasn't for the ability to cat /proc/version, most Linux users wouldn't have a clue what minor or major version they were running.

So the vm stuff was ripped out half-way through. Did it cause any problems for end users? I didn't notice. If anyone is running a Linux box that is under so much load that they did notice, then maybe they should have upgraded their hardware. It gives me images of someone fanging their VW beetle down the road at 180km/h and complaining that the new suspension seems a little rougher than the old one.

I've read some examples of 'my sis motherboard craps out when I do this' or 'my oh-so-cheap raid controller doesn't like it when the kernel does this'. These are drivers people! They can't be considered a core part of the kernel. You can't brand a kernel unstable because someone's obscure, 5-year old POS hardware's drivers haven't been updated in years. Buy some real hardware. See above point. I challenge anyone with regular ( ie I can walk down to the local computer store and buy one because it's in production and regular use now ) hardware to tell me what problem they've had with any 2.4 kernel.

Funny thing is that most people bitching about the supposed instability of early 2.4 and 2.6 kernels most likely upgraded to them as soon as they came out ( just like me ) and bragged to all their Windows-using friends about how stable and fast etc their new kernel is. It's only on Slashdot where it's cool to whinge about how people can break kernel-x just by xxxxxx that they change their tune. And how many of the complainants actually submitted a bug report? MMMmmm?

Re:Man some people can bitch about nothing

[CaptainPhong]

See above point. I challenge anyone with regular ( ie I can walk down to the local computer store and buy one because it's in production and regular use now ) hardware to tell me what problem they've had with any 2.4 kernel.

You'd better qualify that a bit more. There have been some 2.4 revisions (usually short lived) that have had substantial breakage with some ordinary hardware. Most distros have a extra q&a process or have a significantly patched kernel, so these versions don't get to most users. Just the list of USB devices that have underwent some sort of problem during the 2.4 series would take days to recite. My Sandisk USB compactflash reader was broken in 2.4.19 and 2.4.20.

Oh, and don't forget, you can still go to the store and buy a Winmodem.

I've read some examples of 'my sis motherboard craps out when I do this' or 'my oh-so-cheap raid controller doesn't like it when the kernel does this'. These are drivers people! They can't be considered a core part of the kernel.

Well, they're in the kernel source tree, they link to the kernel, they run in kernel-space, they touch the hardware directly, the kernel can't boot without some of them, etc. But no, they're not part of the kernel.

Don't get me wrong, though overall I think 2.4 has been in very nice shape overall for a while now. But just because you haven't had trouble with earlier incarnations doesn't mean other people haven't, and it doesn't always mean they have crappy hardware. Why do you assume anyone using Linux can afford to replace every piece of hardware that isn't supported well?

On a more positive note, last night I got a call from my dad. Without any supervision, he installed SuSe and got it working WITHOUT destroying his Windows partition. I wouldn't have expected him to get through installing Windows, let alone surivive partitioning and installation of Linux. The day before he had called me and told me that he had purchased SuSe at the store because he was pissed at Windows. He couldn't boot off the CD (because his BIOS wasn't configured to). In shock (and a twinge of terror) I told him to hold off until the weekend or something so I could help him out. Well, he wasn't that patient, and it turns out he didn't need my help after all (though he is going to need my Courier I'm not using because he has a Winmodem). I'm talking about somebody who thinks their whole computer is their "hard drive" and that he has a "Western Digital" modem. He said, and I quote, "I can see why people like this Linux thing. It's really user friendly."

Re:Man some people can bitch about nothing

[CaptainPhong]

Oh, I should also mention, his Winmodem wasn't working under Windows, which is the thing that drove him to install Linux.

Re:Man some people can bitch about nothing

[vandan]

Yeah.
Come on then big server man.
Tell me how it is on your Pentium 2 300...

Re:Hopefully it will be more stable then 2.4x in s

[zmooc]

My car still, by default, requires a check-up every now and then. That just screams stability.

Summary of changes...

[eille-la]

Alexey Kuznetsov: o [TCP]: Delete obsolete comment

I like this one.

Coffee?

[stemcell]

Even n00bies know that tea has higher concentrations of caffeine than coffee.

Tastes better too.

Stem

Just leave out the dots...

[g.a.g]

And voilá, Linux kernel 2575! Zipped past Windows elegantly enough!

ACPI *not* in 2.4 yet

[ChrisCampbell47]

2.4 isn't even in "maintenance" mode yet - it is _the_ stable tree, and its getting new things added to it with each release (slowly, and after being tested in other trees, and RCs). Just recently new ACPI for example.

Er, ACPI is a bad example because ACPI has NOT yet been migrated into 2.4. The ACPI guys have been trying to get it in for a long time and finally threw a tantrum on LKML about it when Marcelo said he was planning on releasing 2.4.22 without it. They changed his mind and Marcelo now plans to include it, but it's not actually in there yet.

Re:ACPI *not* in 2.4 yet

[iabervon]

You're a month out of date; it went into 2.4.22-pre2.

HA HA HA

[paradesign]

you got trolled big time! you should read the apple section now and again. this troll pops up every time theres a Mac related enhancement... so all the time. all he did is replace Mac with Linux.

and dont feel too bad, it trips up some people in teh apple section too. must say, one of teh more sucessful trolls ive ever seen.

upgrade ?!?!

[cvbear0]

guess that means i should move from 2.2.12 to 2.4.x

hmm...

Its up to distros

[linuxislandsucks]

Its up to distros to make users aware of 2.6 bugs per hardware combinations as they have the most motive to test on every hardware combo of their customers...

Not to be ungrateful, but...

[gosand]

[snip]
More information can be found in Dave Jones' list of things to expect in 2.6. Personally, I think it's great to see features that benefit both big and small systems.

Thanks for the information. I find it kind of disheartening that this type of info doesn't come directly from the kernel maintainer himself. It seems like every time a kernel is released, I think "maybe I'll upgrade me kernel this time". But when I go out to see what is in the new version, I get essentially zero information. Oh, I see statements of what went in, but they don't tell me anything. What the hell is "anticipatory scheduler"? Why do I care? I shouldn't have to figure these things out, or diff the code, or any of that BS. Is it so hard to just state in plain and simple terms what the changes are? I am glad someone did it, but it would be nice if it came from the maintainer.

I suppose I'll get modded down for critisizing Linus now. Oh well.

world class ipv6

[clenhart]

Does this mean that world class ipv6 is out?

http://www.kernelnewbies.org/status/latest.html

stable vs. unstable

[oojah]

stable/unstable isn't supposed to refer to how likely the kernel is to die, more that it is in a state of flux with new features being added, internals being changed. As a consequence of this the kernels tend to be less reliable. Stable means (should mean!) that no new major changes are taking place and that the emphasis is on increasing reliability.

Cheers,

Roger

Debian installer

[ca1v1n]

So, maybe by 2005 we'll see the official Debian installer using a 2.4 kernel?

Re:Debian installer

[SuiteSisterMary]

Other than bf2.4, you mean?

Re:Debian installer

[ca1v1n]

Damn near every installer out there uses bf2.4, just not the official one, so far as I can tell.

Re:Debian installer

[SuiteSisterMary]

bf2.4 is one of the 'official' Debian install flavours; there's about five of them. To boot up bf2.4, boot off of CD 5 of the Debian Distro.

Use xplaycd

[autechre]

It's in Debian, but I don't know about Red Hat. Can't be that hard to get, since it's a simple program.

At the peak of my radio-station-ing (summer means slowness for college stations), I was reviewing at least two or three CDs per week, and I had to screen about five times that many. Almost every single Linux CD player was just too fragile for the job, messing up after about 3 CDs worth of abuse (heavy track skipping, seeking, frequent disc changing, etc.) xplaycd is the one I tried that worked (and continues to work).

As per another poster, I think that some of your problems might be in hardware. I have a Plextor ATAPI CD-RW (and they are so worth the extra $$), and I haven't had a single problem. Well, except for the fact that it can write dodgy media that other drives won't be good enough to read, but I tossed that stuff and haven't had a problem since I started using Computer Renaissance CD-Rs.

apt-get IS awfully nice :) When I want a new program, I'm surprised if I DON'T find it with "apt-cache search [program]". And having configured several MTAs, I'd have to say that Postfix is probably the easiest (although sendmail is not as hard as people say, even with a few funky options). But if all you need is a basic config, you may be happy to know that Debian packages ask you a set of questions to get themselves up and running when they're installed, rather than just dumping the files on your drive and saying "Good luck!"

Re:How sad

[rburt3]

Well, I got it.

Linux ok, FreeBSD Better

[jo42]

TheAge reports : The Top five of the top 10 hosting providers for the month of June, measured in terms of those whose sites experienced the fewest failed requests and provided the fastest connection times, are all running the FreeBSD operating system , data from Netcraft shows .

Obligatory Spinal Tap Quote...

[Man In Black]

So there. All you weenies stuck at 10, I have no pity for you.

Why don't you just make 10 better, and make the version number 10 and make that a little better?

Huff, Trice, Jiffy

[HiggsBison]

These are all vehicles.

He left in a Huff.
My Grandfather had a 1934 Huff. Classic.

He came back in a Trice.
The Trice, of course, had three wheels.

He did it in a Jiffy.
The first ones were built in the late '50s.
No, I've never "done it" in a Jiffy.

Sorry I couldn't be of any real help.

Violating their own numbering scheme..

[netsharc]

Whatever happened to the x.y.z numbering scheme, when y is even it's stable, when it's odd is development.. now you have x.y.z-pre-A, etc, etc. What's the point?

Kernel upgrades are not sexy

[IncohereD]

Thanks for the information. I find it kind of disheartening that this type of info doesn't come directly from the kernel maintainer himself. It seems like every time a kernel is released, I think "maybe I'll upgrade me kernel this time". But when I go out to see what is in the new version, I get essentially zero information. Oh, I see statements of what went in, but they don't tell me anything. What the hell is "anticipatory scheduler"? Why do I care? I shouldn't have to figure these things out, or diff the code, or any of that BS. Is it so hard to just state in plain and simple terms what the changes are? I am glad someone did it, but it would be nice if it came from the maintainer.

So you're a lazy S.O.B., is what you're saying? Try typing "anticipatory scheduler" into google, and see what you find.

How often doesn microsoft tell you what's changed in their kernels, except that "it should be 20% faster now" or whatever. And that's what the scheduler change means.

If you want it dumbed down, the new kernel offers better speed, better stability, and better support of new hardware and software architectures (i.e. ACPI and ALSA).

If you don't care, then don't upgrade your kernel. The kernel isn't sexy. It's the workhorse. Like, if someone told me they wanted to put a new engine in their car, I'd hope to god they would just tell me it's more powerful and more efficient, and not go on about pistons and shafts and what not. Because I'm not an engine geek. And I hope even more they're not pulling the lead engineer out of the BMW plant to come tell me about it, either. He should be spending his time working on engines.

So if you're not a kernel geek, don't worry about it. Wait for your distro to upgrade.

Comment removed

[account_deleted]

Comment removed based on user account deletion

And they shouldn't be!

[gosand]

So you're a lazy S.O.B., is what you're saying? Try typing "anticipatory scheduler" into google, and see what you find.

I see. So I should do a google search on every single change in the kernel, and piece together the information into something coherent? That is pretty stupid, when the person who handles the kernel could write up a paragraph or two in 5 minutes. It isn't like Linus isn't capable of doing this.

As long as this attitude prevails, Linux will have a harder time gaining any respect by the larger community. Not that I care much, because I will keep using it. But even though I have been using it since the late 90's, I am still amazed at the attitudes of some of the people in this "community".

How often doesn microsoft tell you what's changed in their kernels, except that "it should be 20% faster now" or whatever. And that's what the scheduler change means.

So the Linux community should follow Microsoft's lead on this one? WTF kind of logic is that?

If you want it dumbed down, the new kernel offers better speed, better stability, and better support of new hardware and software architectures (i.e. ACPI and ALSA).

No, I don't want it dumbed down - I would just like it in basic sentences so I can decide to upgrade it or not. Admit it, you don't know what is in the new kernel either, at least not from looking at the release notes. You are just willing to blindly upgrade it, pretending you are all cutting edge and cool.

And I hope even more they're not pulling the lead engineer out of the BMW plant to come tell me about it, either. He should be spending his time working on engines.

I'll guarantee you that the lead engineer at the BMW plant will be able to tell you in a nutshell what the improvements of a new engine are without using obscure terminology and engineering-speak that only an engine guru would know. BTW, Paul Rosche is my hero.

What's a jiffie?

[Khan Fused]

Jiffy (adjective) - Like, or similar to, smashed peanuts.

See also: Jif

good lord!!!!!!!!!

[XO]

i just went from 2.5.73 to 2.5.75.

All my previously broken modules compiled.
They all loaded.
No modules paniced the kernel.
I have 80MB free after boot (debian testing) before loading X (used to have about 45MB). Load X (fvwm2), and Mozilla, and still have 35MB free.

This is moving faster than I have ever seen Linux move. Ever.

PIII/600MHz, 128MB ram, 9GB SCSI and 4GB EIDE drives.

Re:I guess this means that

[pair-a-noyd]

Troll?!! Are you kidding me??

I'm stone cold serious! Have you ever played the waiting game with Mandrake??

Sheesh!

kernel prelease 2.6 is unclear

[dankelley]

Gosh. The plans listed at the linked discussion site are pretty unclear. I mean literally unclear; the site uses dark-grey text over a black background ;-)

Re:How sad

[Sloppy]

Then again, I've probably just been trolled. But FYI for those people who believe him.

Actually, I was just making a lame joke. No serious trolling was intended. If this has been an actual troll, it would have been modded up, not down. :-)