Linus Torvalds released on late Sunday the first test (-pre) kernel of the upcoming Linux Kernel 2.6. Download or get more information here and make sure you report any bugs you might find on the way.
Linus Torvalds released on late Sunday the first test (-pre) kernel of the upcoming Linux Kernel 2.6. Download or get more information here and make sure you report any bugs you might find on the way.
At long last – we are on our way to the greatest version of Linux yet. Once they’ve stabilised this beast, it will be a pretty big step forward for Linux systems both large and small. Now that all of the patches and add-ons that vendors have bolted onto 2.4 are now *in* the mainline kernel, we should see a lot more synergy / commonality coming through, not to mention less kernel maintenance effort from the vendors.
But one question – what ever happened to the 64-bit dev_t thing?
I see everyone hyping up this 2.6 kernel, but what are the new features that are really going to make it THAT good for Joe User (me)?
Just read Dave Jones’s “post halloween document”.
http://www.codemonkey.org.uk/post-halloween-2.5.txt
or check out the new feature list @
http://kernelnewbies.org/status/latest.html
http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.0-test1
I’ve been using the development kernels since 2.5.63, (usually Andrew Morton’s variant back then) and am now on to 2.5.75.
They are stable and glitch free for me, although my requirements are fairly lowly (just as my desktop). The 2.5.75 kernel is noticably faster than the 2.4 series, and does much better under load. I can run several CPU intensive things at once, such as compiling a couple of things and playing a 3d game or two, without a hitch.
I expect the BeOSers will be on in a mo saying, “but does it boot any faster?” Well, it’s probably 45s from power on to the GDM login prompt for me, but that includes loading up a crap load of services (things like Apache, ntpd, ssh, pureftp). If I cared enough (I don’t since I reboot about once a month if that, usually because of a kernel upgrade or kicking a wire out) I’d refine that so that the unecessary services were loaded post-GDM rather than pre-GDM and cut the boot time to 20-25s. But I don’t.![;)](https://www.osnews.com/images/emo/smile.gif)
Please tell me wut that reeee-boot thing is again ? I remember hearing about it back when my parents were young but please tell me what that reee-boot is all about.![;)](https://www.osnews.com/images/emo/tongue.gif)
Does Nforce (audio, ethernet) drivers + Nvidia Graphics driver work on 2.6.0-test1?
Which distro are you running 2.6 on? I don’t have linux on my comp right now, but 2.6 is intriguing me. I’m thinking about putting either redhat 9 or debian unstable back on my machine at some point so I can try out 2.6. Which would be easier to get this new kernel on? Any tips or pointers for compiling would be very helpful. Thanks in advance.
hey, how does this 2.6 compare with freebsd 4.8 or 5.1?
If stability and code is supposed to be as strong and clean as freebsd, I might consider using 2.6 once it comes out of beta.
Just a question, not flaimbait. 🙂
Thanks.
i might be wrong (i don’t have a chipset like that) but i recall reading somewhere that the problems where mainly in framebuffer support and grub/shared memory. I suppose not using framebuffer and choosing lilo as a bootloader and you should be fine with any kernel
I reccomend a distro like linuxfromscratch or gentoo. They don’t come with any kernel so they make it easy for you to compile one on their own.
Well given that this is only a test release of 2.6 don’t be expecting great stability, as for code quality, well the kernel janitors have being doing a fairbit of cleaning up old code.
As for comparing it to 4.8 and 5.1 well if Bascule was to run his benchmarks on it and compare it with the results he had for 2.4 and freebsd i’d be plenty interested myself.
yea, i shoulda said, “How WILL 2.6 compare with freebsd 4.8 or 5.1?”
o well… any more thoughts and predictions?
Following the subject, my question is:
Are now included (by default on 2.6.0-test1) the beautifull patches coming from http://kernel.org/pub/linux/kernel/people/akpm/patches?
Regards
Anyone knows the requirements of the 2.6 series of the kernel? I mean if its necesary to install new versions of utils like modprobe or libs like glibc from a fairly new distribution like RH8…
Thanks to everyone…
> Which distro are you running 2.6 on? I don’t have linux
> on my comp right now, but 2.6 is intriguing me. I’m
> thinking about putting either redhat 9 or debian
> unstable back on my machine at some point so I can
> try out 2.6. Which would be easier to get this new
> kernel on? Any tips or pointers for compiling would be
> very helpful. Thanks in advance.
RedHat is always a good bet for compiling development kernels (as this is the distro that Linus uses at work). As for compiling, only compile the modules that you NEED, otherwise you might run into one of the few that still don’t build and will have left the computer running (compiling) and gone out for lunch in vane. Very frustrating!
“I reccomend a distro like linuxfromscratch or gentoo. They don’t come with any kernel so they make it easy for you to compile one on their own.”
What a fat load of hooey. Compiling a kernel is no easier/harder whether your distro comes with one precompiled or not.
To answer your question: if you run Debian Unstable you need only remember to install module-init-tools before booting off the new kernel. This package is needed for the handling of the new-style kernel modules in 2.6. I don’t know about Red Hat.
This link may also prove useful: http://www.osnews.com/story.php?news_id=2949/ – “The Very Verbose Guide to Updating and Compiling Your Debian Kernel”. Of course, nothing precludes you from doing it the non-Debian way if you so choose, but I personally find their method more convenient. 🙂
“RedHat is always a good bet for compiling development kernels (as this is the distro that Linus uses at work).”
Whoa! Correction. Your implication of favouritism towards Red Hat is completely unfounded. Linus has always been very careful to stress that he does not favour any one particular Linux vendor over any other. For example, from http://www.catb.org/~esr/faqs/linus/:
—-
13. Linus took a job with a non-Linux company?
Yes, he did. There is a rumor that he considered going to work for Red Hat, but did not want to show favoritism towards any one Linux vendor. Linus has declined to either confirm or deny this rumor. He uses Red Hat at work and SuSE at home.”
—-
Oh, yes… note that it also says that he uses SuSE at home. Does that mean that he’s biased towards SuSE *and* Red Hat, now? 😀
Linus also said, commenting on his leaving Transmeta to work for OSDL (http://news.com.com/2100-1016_3-1018057.html), that “OSDL is the perfect setting for vendor-independent and neutral Linux development.” Once again, the stress on neutrality.
For all I know (I haven’t tried it) the development kernels may compile better on Red Hat systems than certain other distros. However, if they do, it’s certainly nothing to do with the fact that Linus uses Red Hat at work. And I can personally attest that they compile on Debian Unstable, for example, just fine. 🙂
From the codemonkey document:
>>
– boot time root= parsing changed.
>>
Would that explain why I got stumped after building a new kernel last night? I kept getting the error:
“VFS: Cannot open root device “/” or 00:00. Please append a correct “root=” boot option. Kernel panic: VFS: Unable to mount root fs on 00:00″
If yes, Linus might have forgotten to update the kernel install script, because it too still does it the old way.
Kernel 2.6 requires a newer modutils package (Rusty’s). You can compile your own from his page, or on debian:
apt-get install module-init-tools (from testing/unstable).
I don’t think it has any new glibc requirements tho.
If yeah check Linus changelog everyday like i do, yeah notice that a large amount of applied patches are coming in through Andrew Morton, also he’s going to be the 2.6 maintainer so anything interesting that he has in his tree that isn’t in 2.6.0-pre series will no doubt end up in 2.6
please note I’m not a developer or a veteran sys admin, but heres my gut feeling. Insofar as 4.8 goes 2.6 will demolish it. No question. Everythings better, everythings been touched, replaced, or tweaked – even networking. Its more scalable, can handle load much more nicely, has many nice availibility features, etc. The new VM *rocks*, new scheduler, new i/o scheduler, block layer rewrite, threading, asyn i/o everywhere – even in networking, lots more.
5.x? Much closer call, at this point 2.5 kills FBSD 5.1 – but mostly because FBSD 5.x remains a work in progress while the groundwork for 2.6 is laid, the rest seems stabilization and drivers at this point. Once 5.x is finished I’d give the nod to Linux – but who knows, the BSD developers are very talented and many of them are vets.
> Whoa! Correction. Your implication of favouritism
> towards Red Hat is completely unfounded. Linus has
> always been very careful to stress that he does not
> favour any one particular Linux vendor over any other.
> For example, from http://www.catb.org/~esr/faqs/linus/:
I’m sorry, but my point was well founded. The fact that Linus SuSE at home is almost immaterial. Afterall, I would imagine he rarely compiles kernels at home (except when he has mates over for a few drinks, impress them perhaps?). RedHat is what he uses at work. This by implication means that Linux must compile at least at an acceptable level on RedHat. Linus would not sit still if it didn’t. Add to this the fact that RedHat is THE industry standard as far as Linux is concearned. Sorry to zealots stemming from other distros but it is widely recognised, not universally accepted I will admit.
For RH users the current modutils in rawhide works for both 2.4 and 2.6 kernels
“I’m sorry, but my point was well founded. The fact that Linus SuSE at home is almost immaterial. Afterall, I would imagine he rarely compiles kernels at home (except when he has mates over for a few drinks, impress them perhaps?).”
No, wrong again.
You’re completely missing out on one very important fact. Until very recently (the move to OSDL, still fresh in the news – link in my last post) Linus’s day job (at Transmeta) has had nothing to do with Linux kernel development.
Furthermore, as Linux began as his hobby and surely still remains close to his heart, yes, I’m sure he spends a lot of time on his home computer working on the kernel. In fact, read some of his interviews and it’s made abundantly clear; he mentions reasonably often what hardware he has at home, and the effect that this has on kernel development.
“Linux must compile at least at an acceptable level on RedHat.”
I’ll take your word for it; I haven’t tried it. My point was simply that Linus using a Red Hat machine at his place of work *in no way* implies that Red Hat has better support for development kernels than any other distro, for all the reasons illustrated above.
“Add to this the fact that RedHat is THE industry standard as far as Linux is concearned.”
Yeah; assuming that you’re saying that they’re therefore the most professional, that actually makes them *less* likely to be supporting beta code out of the box. Now that the kernel’s gone from 2.5.x to 2.6-pre, that will probably change rapidly, of course…
Besides, I think your caps overstate the case. Red Hat may well be the major player, true, but there are other contenders for the title: SuSE is very strong in Europe, especially Germany, for example.
Compiles and works well on RH9. You need updated mkinitrd and modutils – can be taken from rawhide. Also, http://people.redhat.com/arjanv/2.5/ has rpms for 2.5.75 for those who don’t like to compile.
“The fact that Linus SuSE at home is almost immaterial. ”
The last interview with Linus I read about his move to OSDL stated that he would be telecommuting from home.
pavel:
it seems the rpm need oprofile >= 0.5.3, where can i get it?
thanks in advance
From this link should I get the source rpms?
http://people.redhat.com/arjanv/2.5/
because I have a k6-2 and that is categorised as an i586
Yes Matthew Baulch because we all know Linus is the only one working on the kernel with a PC with a Linux distro on it.
> Yes Matthew Baulch because we all know Linus is
> the only one working on the kernel with a PC with a
> Linux distro on it.
I fail to see your point. Was sarchasm the intention or are you just out for a Sunday afternoon troll?
That’s an absurd claim.
Any issues you have compiling a kernel or supporting your chosen kernel are nothing to do with the direction Linus has taken Linux and certainly nothing to do with him using Red Hat and I wouldn’t even entertain the possibility of him favouring Red Hat.
To claim he does and that compiling is easier within Red Hat shows a fundamental misunderstanding of Linux as an entity and GNU/Linux as an operating system.
Red Hat is merely an organised collection of programs put together on top of GNU/Linux. The same is true of any distribution. Most distributions contain / use similar software in their collections. Any decent distro will let you substitute the default kernel with any other suitably configured variant of Linux. I’ve had no issues doing this with Gentoo, Red Hat or Mandrake and know people who happily do this with Slackware and Debian.
The comments that Linus would favour Red Hat as a target for his kernel are frankly ridiculous and smell of FUD. Somebody needs to do some research.
very very very very nice.. its the first 2.5 based kernel I’ve used… The performance difference between 2.6test1 and 2.4.22pre2 is just amazing.. I thought my computer was fast before with 2.4.22,, but now its just blazing.. you can’t even see the windows redraw when you drag them!
I can’t wait for the next versions![;)](https://www.osnews.com/images/emo/smile.gif)
I am currently using the NVidia drivers from their website, and the last time I tried to use a newer kernel (On Redhat) it complained about needing some kernel interfaces and stuff. I do not believe this should be necessary. Do it need new drivers, or interfaces for my geforce 2MX
Go to http://www.minion.de/ and you will find patches for the development kernel nvidia driver. Check nvidia forums for more info on how to patch if you can’t figure it out. The nvidia drivers work fine for me under the development kernel with those patches.
still unuseable with nforce2 here….too bad
RedHat is always a good bet for compiling development kernels (as this is the distro that Linus uses at work).
I don’t know why everyone is disagreeing with the initial statement. The original poster never stated it is easier to compile a kernel on Redhat; He never stated Linus favors Redhat (I don’t know where that came from after reading the original post above)
He simply said that Redhat is a good distro to try out compiling the new kernel, because the kernel is used on Redhat by Linus. Linus would probably catch problems with the kernel on a redhat install.
I’m not saying it is the truth, but he never said Linus endorses Redhat, or that new kernels are easier on Redhat.
I’m using Oracle9i on RH9.0 but it does not take advantage of the new threading library yet. The 2.6 kernel is awesome. I would like to eventually upgrade to SuSe and MySQL from Oracle and Java2…and devfinately pay for and support 2.6.
I’m going to throw money at Linux for no reason. It just makes me happy.
Thanks to those before who helpfully offered the links to the pages explaining the 2.6 features. However, for those who are not extremely tech-minded (like me), could someone give a fairly succinct explination of why 2.6 might be better for the average desktop/SOHO user than 2.4? (I’m not completely tech-illiterate, but I’m certainly no hacker, either.) Thanks!
Wasn’t this stable version planned to be called 3.0 sometime during the development?
And about FreeBSD 5.x vs. Linux 2.6 comparison. I think they have different goals and such comparison is difficult. But Linux is going to be quite responsive compared to 2.4 and maybe even more so than 5.x. Because FreeBSD 5.x is not based on pre-emptible kernel yet. John Baldwin is speculated to have some code but doesn’t work…
There are two major improvements that are supposed to help desktop performance: Preemetible kernel, and a new scheduler.
The preemtible thing simply means that, when used, the kernel will have reduced latency when responding to interactive events (such as everything you do on a desktop), even when it is under a heavy load. So, it means it will react better when you are using your desktop, even when you are doing several processor intensive things (e.g. playing a game, compiling something, a program in an infinite loop, you get the idea
.
Also, the compination of a new scheduler with the preemetible option will make things even more responsive and scalable.
Sound support is better now too, with the addition of alsa into the kernel. IIRC, the other, older, drivers are still in there too (for those people who prefer them).
The journaling filesystem xfs is now included in the kernel, rather than having to patch the kernel in 2.4.*
Will this also improve daemon server process responsiveness?
“However, for those who are not extremely tech-minded (like me), could someone give a fairly succinct explination of why 2.6 might be better for the average desktop/SOHO user than 2.4?” Sure. Here goes:
Better support for Bluetooth, Serial ATA, AGP 3.0, USB devices
Specifically:
Kernel build system. (Some consider the text mode menu of kernel building to be slow & confusing. Inlcuded is a graphically clean menu of choices on how to build the kernel)
~~~~~~~~~~~~~~~~~~~~
– The build system is much improved compared to 2.4. You should notice quicker builds, and less spontaneous rebuilds of files on subsequent builds from already built trees.
– There are new graphical config tools.
– Make menuconfig/oldconfig has no user-visible changes other than speed, whilst numerous improvements have been made.
Kernel preemption.
~~~~~~~~~~~~~~~~~~
– The much talked about preemption patches made it into 2.5. With this included you should notice much lower latencies especially in demanding multimedia applications.
Threading improvements.
~~~~~~~~~~~~~~~~~~~~~~~
– Users should notice a significant speedup in basic thread operations. This is true to a lesser extent even for old-threading userspace libraries such as LinuxThreads.
PnP layer.
~~~~~~~~~~
– Support for plug and play devices such as early ISAPnP cards has improved a lot in the 2.5 kernel. The new code behaves more closely to the code handling PCI devices (probe, remove etc callbacks), and also merges PnP BIOS access code.
ALSA. (The advanced linux sound architecture is the next gen sound support in linux. ALSA is comparable to streaming WDM or ASIO drivers for Windows.)
~~~~~
– The advanced linux sound architecture got merged into 2.5. This offers considerably improved functionality over the older OSS drivers, but requires new userspace tools.
The long term goal is to get everyone moved over to (the superior) ALSA.
v4l2
~~~~
– The video4linux API finally got its long awaited cleanup.
– xawtv, bttv and most other existing v4l tools are also compatible
with the new v4l2 layer. You should notice no loss in functionality.
– See http://bytesex.org/v4l/ for more information.
CD Recording.
~~~~~~~~~~~~~
– Jens Axboe added the ability to use DMA for writing CDs on ATAPI devices. Writing CDs should be much faster than it was in 2.4, and also less prone to buffer underruns and the like.
– Updated cdrecord in rpm and tar.gz can be found at *.kernel.org/pub/linux/kernel/people/axboe/tools/
– With the above tools, you also no longer need ide-scsi in order to use an IDE CD writer.
(some people had good luck with ide-scsi, while other had trouble with it. This new cleanup should make it easier.)
– Ripping audio tracks off of CDs now also uses DMA and should be notably faster. You can also find an updated cdda2wav at the same location.
Simple boot flag support.
~~~~~~~~~~~~~~~~~~~~~~~~~
The SBF specification is an x86 BIOS extension that allows improved system boot speeds. It does this by marking a CMOS field to say “I booted okay, skip extensive POST next reboot”.
CPU frequency scaling. (Typically used in laptops/handlhelds with “mobile” processors to extend battery life)
~~~~~~~~~~~~~~~~~~~~~~
Certain processors have the facility to scale their voltage/clockspeed. 2.5 introduces an interface to this feature, see Documentation/cpufreq for more information. This functionality also covers features like Intel’s speedstep, and the Powernow! feature present in mobile AMD Athlons. In addition to x86 variants, this framework also supports various ARM CPUs.
Yes, it will be faster. It will handle heavy loads better. In other words, you will be able to launch 1000 programs without noticing any performance degredation. Okay, I exagerated a little. A lot of bugs have been smashed. A lot of features have been tweaked. Playing games will be a wet dream. But you’ll have to configure the kernel manually and add the needed features. No, you don’t need a copy Red hat. No, you don’t need a copy of SUSE. I wish I can put it in simpler terms. It is just better and by next year you’ll see if it’s all hype or if it’s for real. People like me who compile programs from source code can do so happily while burning a CD on my 5th virtual desktop and watching DVD on my 9th virtual desktop. Enough said.
Yes, of course, server process responsiveness will be affected too. As the other message states, the new threading should be quite interesting too!
the new threading should be quite interesting too!
Yeah. But nowhere as interesting as FreeBSD’s KSE or NetBSD’s scheduler activations IMHO.
Well, some felt the progress since 2.4 to be so groundbreaking that the name should reflect this with a major version bump. Others disagreed, arguing a) that version number inflation is reminiscent of a glitzy, sensationalist marketing ploy or b) that the changes are simply not worthy of such a version bump. In the end, sanity won out over sensationalism, so we have 2.6 and not 3.0. 😉
You can check out the relevant lkml excerpts for yourself in the KernelTrap piece “2.6 vs 3.0; What’s In A Name?”, at http://kerneltrap.org/node.php?id=436.
for the timely and excellent responses. I was aware of the basic feature inclusions, but I wanted to know if it would actually affect me or if these would only be noticeable to the ultra-gamers and the video editors. Looks like I’ll actually see some improvement on my desktop. Thanks again, everyone!
“In the end, sanity won out over sensationalism, so we have 2.6 and not 3.0. 😉 ”
With 2.6 sounding so great and having many new features, I’m left wondering what would warrant a number change. Surely, the changes they introduce in the future won’t be able to compare to the leaps 2.6 will make. I’m thinking after 2.8, they’ll just go to version 2.10, and we’ll never see the 3.0 kernel.![;)](https://www.osnews.com/images/emo/smile.gif)
What do you mean by “unuseable”? I’m concerned ’cause I have a nForce2.![;)](https://www.osnews.com/images/emo/tongue.gif)
The nVidia graphics drivers have a kernel module component, and you *must* recompile the interface for that module. Most of the kernel module is binary, but you still have to compile a small portion against your existing kernel source tree for it to properly interface with your running kernel.
Everytime you update your kernel, unfortunately you have to do this.
When are we going to get decent ACPI support? I’m sick of the ACPI developers willfully refusing to be compatible with hardware.
What options do I have now in looking for a new Linux laptop? Are there any that have full support under the Linux kernel?
do you only have one computer or something?
Wrawrat:
1. Nforce2 driver can’t be compiled.
2. My ps/2 mouse doesn’t work.
(3).Assuming you use Nvidia VGA then u still need to patch Nvidia Graphics driver.
I love to dance in Miami:)