Kernel 2.6.23?

[barbro]

What kernel will Hardy ship with?
I hope that it will ship with a kernel that includes CFS?

[Quikee]

Most likely it will ship with 2.6.24 - maybe 2.6.25 but I doubt it.

[amano]

Since it is a LTS release, maybe they just switch to .23 to have 6 months of bug hunting. Who knows... But I guess that we will see a kernel with the new scheduler.

[MacUntu]

.23 fixes the annoying "PCI: Failed to allocate mem resource..." warning during boot up. That's enough for me to want it in! :D

[screaminj3sus]

It really needs a new kernel, the new scheduler is supposed to improve performance alot, and the current kernel is plagued by I/O bugs.

[FredB]

It really needs a new kernel, the new scheduler is supposed to improve performance alot, and the current kernel is plagued by I/O bugs.

Such as ?

I think hardy will have a 2.6.23.x (x>= 5) or a 2.6.24 kernel.

We'll see in time. Let's just have pleasure with gutsy right now.

[Zdravko]

I don't understand why people want a certain kernel.

[syxbit]

it's obvious, generally, later kernels have improved features, and bug fixes.
the reason ubuntu is popular is cause of the fast release cycle - which means an up to date system.

[Zdravko]

Hmm. Does a new kernel also offer the same stability? LTS means stability for me!!!

[Knowles]

I would imagine it would be .23, I can't see it being shipped with .24.

[23meg]

.23 fixes the annoying "PCI: Failed to allocate mem resource..." warning during boot up. That's enough for me to want it in! :D

Just due to cosmetics, or does that error have some important consequences? Asking because I've seen it on some configurations and haven't really researched it yet.

[FredB]

I would imagine it would be .23, I can't see it being shipped with .24.

Why not ? Too close from april 2008 release date ?

[Knowles]

Why not ? Too close from april 2008 release date ?

It's an LTS and if it took as long as it did to bang out .23 as it will for .24 then it might not make the freeze. Though this is my opinion and an opinion is all it is.

[hardyn]

it's obvious, generally, later kernels have improved features, and bug fixes.
the reason ubuntu is popular is cause of the fast release cycle - which means an up to date system.

ditto... i don't think the kernel version is something that really needs to be worried about.

[MacUntu]

Just due to cosmetics, or does that error have some important consequences? Asking because I've seen it on some configurations and haven't really researched it yet.

I didn't come across any problems.

Maybe it's just a cosmetic fix but from my pov an important one. An OS really shouldn't bug users with useless msgs. If I see text during a graphical boot process I think something is wrong.

[hardyn]

im sure its not useless... it probably means something to the developers or people who understand the nuts and bolts of the kernel... but it might be overly verbose.

my notebook does show the PCI error... but my older AMD based desktop does not.

[rubinstein]

Look at the Linux Weather Forecast at http://www.linux-foundation.org/en/Linux_Weather_Forecast

The 2.6.24 kernel should be available at the end of the year, so i suppose we will have this kernel version for Hardy.

[syxbit]

the linux kernel uses a 3 month release cycle, and usually they're close to that (.23 was late)
if that's the case, then 2.6.25 should come out close to Hardy's release.
2.6.24 should be well before the freeze.

ideally they'd take 8 months for their next LTS, like they did on Dapper. Dapper was their best release IMHO

[slavik]

I hope for 2.6.24 because by then the CFS should have group scheduling (group tasks and schedule them as a whole then have them fight it out inside their slice).

[tighem]

Initial reports show significant desktop speed improvements with a CFS enabled kernel. Here's hoping it makes it into Hardy.

[Arathorn]

ideally they'd take 8 months for their next LTS, like they did on Dapper. Dapper was their best release IMHO
Personally, I had far more problems with Dapper then the deliberately unpolished Edgy. I don't think lengthening the time before the next release would improve Hardy, but having an earlier freeze will. An earlier freeze means less time to introduce new features (and bugs), and more time for fixing them. I would like to see mostly bugfixes for Hardy, and off course an (alternative) install cd with KDE4 instead of 3.

[eremini]

Don't forget that the even series kernels are supposed to be the stable ones (.20, .22, .24). And they get .1, .2, etc updates for longer then the odd numbered kernels. So I am sure ubuntu devs will choose to go with .24

[David A Knight]

Don't forget that the even series kernels are supposed to be the stable ones (.20, .22, .24). And they get .1, .2, etc updates for longer then the odd numbered kernels. So I am sure ubuntu devs will choose to go with .24

No. The odd/even unstable/stable was for 2.4 = stable. 2.5 = unstable 2.6 = stable. The development cycle was then changed with 2.6 not to do this anymore and just stick with 2.6 unless major breaking changes forced a 2.7.

[FredB]

Initial reports show significant desktop speed improvements with a CFS enabled kernel. Here's hoping it makes it into Hardy.

We will see in time. Gutsy is anyway a great release.

[FredB]

No. The odd/even unstable/stable was for 2.4 = stable. 2.5 = unstable 2.6 = stable. The development cycle was then changed with 2.6 not to do this anymore and just stick with 2.6 unless major breaking changes forced a 2.7.

A 2.7 line ? What about a 3.0 line ;)

2.0 was released back in 1996, if I remember well ;)

11 years ago... Woo :)

[insane_alien]

3.0 kernel line will come the next time a complete rewrite of the kernel is required(not likely)

[syxbit]

An earlier freeze means less time to introduce new features (and bugs), and more time for fixing them. I would like to see mostly bugfixes for Hardy, and off course an (alternative) install cd with KDE4 instead of 3.

but that's precisely what happened with Dapper.
they kept the freeze time as per their 6 month cycle, and added 2 months after for bug fixing.

[hardyn]

it wasn't something they were planning to do... it was something that they felt they had to do.

I would be really suprised if a 8mo development cycle for hardy happend, mark and the development people were very appologetic about the 6 week delay for dapper... after all it does break from there mission statement of a release every 6months.

if they really wanted to polish it... they should not be adding features, and just finish and tweak what is in gutsy. add packages, change art. gusty is very featured, but needs the corners smoothed off a little.

[eremini]

No. The odd/even unstable/stable was for 2.4 = stable. 2.5 = unstable 2.6 = stable. The development cycle was then changed with 2.6 not to do this anymore and just stick with 2.6 unless major breaking changes forced a 2.7.

Not entirely true, check out how much .1, .2, etc releases even series usually have and how much odd. And some distros such as pclinuxos only build/distribute even kernels

[lazka]

Not entirely true, check out how much .1, .2, etc releases even series usually have and how much odd. And some distros such as pclinuxos only build/distribute even kernels

bugfix/security releases
2.6.11 -> 12
2.6.12 -> 6
2.6.13 -> 5
2.6.14 -> 7
2.6.15 -> 7
2.6.16 -> 55 (because still maintained by someone)
2.6.17 -> 14
2.6.18 -> 8
2.6.19 -> 7
2.6.20 -> 21
2.6.21 -> 7
2.6.22 -> 10

so.... i don't get your argument?

[Jeanpaul145]

A 2.7 line ? What about a 3.0 line ;)

2.0 was released back in 1996, if I remember well ;)

11 years ago... Woo :)

Linus *has* said something in an interview that comes down to that he doesn't see a Linux 3 appearing anytime soon, because he has no marketing motivation.
Ergo, no Giant headlines saying "get the new, improved Linux 3.x" :)

[bikeboy]

A quick review (Distrowatch page (http://distrowatch.com/table.php?distribution=ubuntu)) shows that over its history Ubuntu has generally jumped 2 kernel releases, except for Breezy -> Dapper and Edgy -> Feisty where it was 3.

Pretty sure I remember in the Edgy days that everyone was asking for a 2.6.18 kernel instead of the 2.6.17, until a dev/resourceful forum user pointed out how many backports the Ubuntu kernel team was shipping with their 2.6.17, it was essentially 2.6.18 anyway. In other words, the numbers don't matter, in the end it's the features, driver support and (lack of) bugs that matter.

[FredB]

Linus *has* said something in an interview that comes down to that he doesn't see a Linux 3 appearing anytime soon, because he has no marketing motivation.
Ergo, no Giant headlines saying "get the new, improved Linux 3.x" :)

Yes, I remember this interview. But after 4 years of 2.6.xx version, a 3.0 will be great advertisement ;)

[FredB]

A quick review (Distrowatch page (http://distrowatch.com/table.php?distribution=ubuntu)) shows that over its history Ubuntu has generally jumped 2 kernel releases, except for Breezy -> Dapper and Edgy -> Feisty where it was 3.

Pretty sure I remember in the Edgy days that everyone was asking for a 2.6.18 kernel instead of the 2.6.17, until a dev/resourceful forum user pointed out how many backports the Ubuntu kernel team was shipping with their 2.6.17, it was essentially 2.6.18 anyway. In other words, the numbers don't matter, in the end it's the features, driver support and (lack of) bugs that matter.

Erh...

Warty (4.10) = 2.6.8
Hoary (5.04) = 2.6.10
Breezy (5.10) = 2.6.12.x
Dapper (6.06.x LTS) = 2.6.15 - my first ubuntu ;)
Edgy (6.10) = 2.6.17
Feisty (7.04) = 2.6.20

So, a 2.6.24.x could be realistic, don't you think ?

[bikeboy]

So, a 2.6.24.x could be realistic, don't you think ?

Yep, definitely realistic and my guess is that 2.6.24 will be what we end up with, plus some drivers from 2.6.25 if they're really needed.

[Jeanpaul145]

Yep, definitely realistic and my guess is that 2.6.24 will be what we end up with, plus some drivers from 2.6.25 if they're really needed.

I think so too, although 2.6.24 won't be out until the end of the year, so in actuality I'm still guessing...

Yes, I remember this interview. But after 4 years of 2.6.xx version, a 3.0 will be great advertisement ;)

No argument here:guitar:

[frenchcr]

Just due to cosmetics, or does that error have some important consequences? Asking because I've seen it on some configurations and haven't really researched it yet.

Its on my new laptop too, ive tried to work it out for a while, its an unknown still

[Jeanpaul145]

Its on my new laptop too, ive tried to work it out for a while, its an unknown still

I've got it too, and it annoys me a bit, but I've yet to discover any relation between that message and any problems I'm having with Kubuntu Gutsy:popcorn:

[Kow]

2.6.24 seems almost a guarantee for hardy, however I would still like to see a 2.6.23 middle-ground. This way the kernel development is not flat on its butt for the next couple of months and come Jan-Feb-March bug elimination wont be as hardcore (instead of having to eliminate bugs introduced in 2.6.23 as well as those introduced in 2.6.24). But, then again we (ubuntu) can leave it up to the kernel.org dev team to eliminate the 2.6.23 bugs and then take the 2.6.24 RC's as the base for hardy development.

Regardless, the kernel dev team is probably quite bored (and happy) right now and probably will be for the next month or so. :P

[Kow]

2.6.24 seems almost a guarantee for hardy, however I would still like to see a 2.6.23 middle-ground. This way the kernel development is not flat on its butt for the next couple of months and come Jan-Feb-March bug elimination wont be as hardcore (instead of having to eliminate bugs introduced in 2.6.23 as well as those introduced in 2.6.24). But, then again we (ubuntu) can leave it up to the kernel.org dev team to eliminate the 2.6.23 bugs and then take the 2.6.24 RC's as the base for hardy development.

Regardless, the kernel dev team is probably quite bored (and happy) right now and probably will be for the next month or so. :P

I quickly retract my statement after trying to compile ubuntu-modules against 2.6.23.1. It looks like the reckless implementation of CFS into the mainstream kernel has broke MANY of these packages, most of which are not maintained by the creator anymore. This is without mentioning the deprecation/removal of the old-school ACPI implementation (breaking most of the acpi packages)... Good luck ubuntu kernel/driver devs. :)

[volanin]

Since it is a LTS release, maybe they just switch to .23 to have 6 months of bug hunting. Who knows... But I guess that we will see a kernel with the new scheduler.

I am using Gutsy with the kernel 2.6.23.1 right now.
I can tell you first hand that the CFS improvements aren't ground-breaking.
You barely feel anything...

... but on the other hand.
The I/O improvements are the bomb.
In the original Gutsy kernel, if you were making any disk-intensive activity, like copying many files, the system would slow to a crawl.
Heh... try copying a whole kernel tree and opening a simple gnome-terminal, and you'll see what I am talking about.

Linus, if I am not mistaken, has already told that he hated the .21 and .22 kernels... In 2.6.23, you can do I/O easily, and the system is as responsive as always.
And suspend/resume works a lot better as well.

Anyway, be it CFS, I/O or any other thing, once you use it for a while, you won't want to go back to .22 for sure!
:)

[bikeboy]

I was reading last night that the CFS as it stands now is sort of a compromise between server fairness and the real-time/low-latency scheduling that desktop users prefer. The kernel guys (in general) preferred a single, not completely ideal solution that's easier to maintain than having multiple schedulers for specific environments and greater complexity to maintain (understandably). There will be further improvements to desktop scheduling with changes in the works for CFS. Something about group scheduling iirc. Can't remember when it is due, but .24 or .25 ring a bell.

[syxbit]

i compiled 2.6.23 a while back (when in rc6) but wireless didn't work, and i didn't have time to fiddle. i know you're supposed to include the iwlwifi.
i just don't have the time anymore.
a while ago I loved fixing and braking linux. Now, I'm just tired and busy. I may just stick with what ubuntu has as default

[Kow]

Yea all I can say is linux-ubuntu-modules and linux-restricted-modules will pretty much need to be recreated from scratch. Most of the ubuntu-modules do not compile against 2.6.23.1 and neither does the nvidia drivers (at least the legacy does not.) Madwifi compiles fine if you use the latest stable version (0.9.3.3). I finally got linux-ubuntu-modules to compile after a lot of source updating (either upstream, or did it myself). linux-restricted drivers just is not going to happen specifically because the nvidia drivers are not open source.

[volanin]

Yea all I can say is linux-ubuntu-modules and linux-restricted-modules will pretty much need to be recreated from scratch. Most of the ubuntu-modules do not compile against 2.6.23.1 and neither does the nvidia drivers (at least the legacy does not.) Madwifi compiles fine if you use the latest stable version (0.9.3.3). I finally got linux-ubuntu-modules to compile after a lot of source updating (either upstream, or did it myself). linux-restricted drivers just is not going to happen specifically because the nvidia drivers are not open source.

Yes, you are right. It a major headache.
So major, I gave up on them and just uninstalled everything kernel related.
Then I recompiled only the modules my computer needs. No problems until now.
But you must know your hardware inside-out!
:)

[ThrobbingBrain66]

I didn't come across any problems.

Maybe it's just a cosmetic fix but from my pov an important one. An OS really shouldn't bug users with useless msgs. If I see text during a graphical boot process I think something is wrong.

I hope it's not just cosmetic. The PCI address given in that message on my machine refers to the card reader. Coincidently (or not) my card reader doesn't work in Gutsy...

[Jeanpaul145]

I hope it's not just cosmetic. The PCI address given in that message on my machine refers to the card reader on my laptop. Coincidently (or not) my card reader doesn't work in Gutsy...

I also have a card reader on my laptop, but haven't had the need to use it...ever.
Maybe I should try it out then:lolflag:

[stmiller]

I hope Hardy will have the 2.6.24 kernel. It boasts plans for wifi and HD tv card improvements, among other things that are good for desktop users.

http://lkml.org/lkml/2007/10/24/7

The 2.6.23 kernel will be old news by April 2008...

[Kow]

I hope Hardy will have the 2.6.24 kernel. It boasts plans for wifi and HD tv card improvements, among other things that are good for desktop users.

http://lkml.org/lkml/2007/10/24/7

The 2.6.23 kernel will be old news by April 2008...

And with huge changes come more quantities of bugs... and oh there will be. Even the best of the best miss things and the kernel development team is no exception. Hopefully 2.6.24 is out long enough for the ubuntu kernel dev team to stabilize it by March/April. All of these wireless changes will likely break a lot of out-of-kernel wireless drivers (and we'll have to wait for the closed source wireless drivers to be fixed.)

[volanin]

Hopefully 2.6.24 is out long enough for the ubuntu kernel dev team to stabilize it by March/April.

Just for reference, the time it took from 2.6.23-rc1 (first testing release) to 2.6.23 (official release) was 3 months.
And this kernel version took a little longer than usual.

2.6.24-rc1 was released yesterday.
The final version will probably be out before mid-January.
And then, using the same 3-month spam, 2.6.25 will be out by April.

So probably Hardy will settle for 2.6.24.X, with the usual debian/ubuntu patches.
:)

[oobuntoo]

.23 fixes the annoying "PCI: Failed to allocate mem resource..." warning during boot up. That's enough for me to want it in! :D

I'm glad to hear that this problem is gone with the kernel 2.6.23. I'm wondering how Dell laptops that were sold with Feisty or Gutsy deal with this annoying error message.

[gtdaqua]

correct me if i am wrong.

2.6.23 is out. But Gutsy currently is using (by default) 2.6.22, right?

Because my kubuntu gutsy 64-bit is "2.6.22-14-generic". Am i missing something here? I want to try and use .23 because of the new CFS .

[Jeanpaul145]

correct me if i am wrong.

2.6.23 is out. But Gutsy currently is using (by default) 2.6.22, right?

Because my kubuntu gutsy 64-bit is "2.6.22-14-generic". Am i missing something here? I want to try and use .23 because of the new CFS .

For a kernel to be accepted into a Ubuntu release it needs to be added BEFORE the feature freeze (this because of testing purposes, for both the kernel and software dependent on it). For Gutsy, this time has come and gone a long time ago.
:popcorn:

[gtdaqua]

For a kernel to be accepted into a Ubuntu release it needs to be added BEFORE the feature freeze (this because of testing purposes, for both the kernel and software dependent on it). For Gutsy, this time has come and gone a long time ago.
:popcorn:

so are you saying that i should be having .23 now?

[Arathorn]

No. You should be on .22.

[gtdaqua]

Honestly, i dont understand. Jeanpaul145 says the time has come and gone for Gutsy. What is that supposed to mean? Gutsy should be in updated kernel right?
Sorry if all this so stupid but doesnt make sense to me.

[volanin]

What he meant is that .23, to be included, should have been released BEFORE the Gutsy freeze... but it DIDN'T.
So Gutsy stood with the kernel .22
:)

Since the kernel version is NEVER upgraded in a release...
... Gutsy will always use the .22 version.

To try a new kernel you have three options:
- Compile a new version yourself.
- Get a package from someone who compiled.
- Wait for Hardy, which will probably come with .24

[gtdaqua]

What he meant is that .23, to be included, should have been released BEFORE the Gutsy freeze... but it DIDN'T.
So Gutsy stood with the kernel .22
:)

Since the kernel version is NEVER upgraded in a release...
... Gutsy will always use the .22 version.

To try a new kernel you have three options:
- Compile a new version yourself.
- Get a package from someone who compiled.
- Wait for Hardy, which will probably come with .24

Its clear now! I got it the other way around the first time!

Thank you!!

[Jeanpaul145]

Honestly, i dont understand. Jeanpaul145 says the time has come and gone for Gutsy. What is that supposed to mean? Gutsy should be in updated kernel right?
Sorry if all this so stupid but doesnt make sense to me.

As Volanin was so kind to point out:), indeed a kernel version is never updated after a Ubuntu feature freeze.
"The time has come and gone for Gutsy" was referring to the period GUtsy was in BEFORE it was feature frozen:popcorn:

[mendieta]

I wonder, anyone know if it would be possible to request the backports team, either a patched 2.6.22 with CFS, or (less likely) a 2.6.23 kernel ?

Cheers!

[RAOF]

I wonder, anyone know if it would be possible to request the backports team, either a patched 2.6.22 with CFS, or (less likely) a 2.6.23 kernel ?

Cheers!

No. Even more so than interpreters/runtimes (see mono/python from backport wikipage (https://help.ubuntu.com/community/UbuntuBackports)) a new kernel has the opportunity to break everything.

[Jeanpaul145]

I wonder, anyone know if it would be possible to request the backports team, either a patched 2.6.22 with CFS, or (less likely) a 2.6.23 kernel ?

Cheers!

You can always ask, as it couldn't hurt:popcorn:

[Kow]

No. Even more so than interpreters/runtimes (see mono/python from backport wikipage (https://help.ubuntu.com/community/UbuntuBackports)) a new kernel has the opportunity to break everything.

It also has the opportunity to awesome-ize everything. :)

[mendieta]

No. Even more so than interpreters/runtimes (see mono/python from backport wikipage (https://help.ubuntu.com/community/UbuntuBackports)) a new kernel has the opportunity to break everything.

I agree (for a new Kernel). But there is a clean patchset just for the CFS. A patched 2.6.22 doesn't seem like a huge risk, this is just the scheduler. It would be installed side-by-side with the regular kerne. If it doesn't work well you reboot and use the official kernel. Besides, the backports are unofficial AFAIK. One should in principle stay away from this repository in a mission critical box.

Many people in this forum have tried a CFS enabled kernel and reported nice responsiveness enhancements in desktop usage (overall throughput is unaffected AFAIK).

Cheers!

[23meg]

It's 2.6.24.

https://blueprints.launchpad.net/ubuntu/+spec/hardy-kernel-schedule

[barbro]

I read a while ago that there was some talk about Kernel Graphics Subsystem.
Is this already in the current kernel or will they merge it or simply put it in the trash (due to stability issues)?

[RAOF]

I read a while ago that there was some talk about Kernel Graphics Subsystem.
Is this already in the current kernel or will they merge it or simply put it in the trash (due to stability issues)?

I'm not sure what you're talking about here. Do you mean moving a bunch of video driver infrastructure from X to the kernel - modesetting, etc? That is planned (and is happening right now with the nouveau driver - you need the kernel module to run the X driver).

[barbro]

I'm not sure what you're talking about here. Do you mean moving a bunch of video driver infrastructure from X to the kernel - modesetting, etc? That is planned (and is happening right now with the nouveau driver - you need the kernel module to run the X driver).

This is what im talking about: http://kerneltrap.org/node/8242 dont know if its the same thing

[RAOF]

Ah, yes. That was what I was talking about, too. That is in no way finished :).

My current understanding is that the nouveau driver is pretty much the only driver that does anything like that. Maybe there are experimental Intel drivers doing that, too, but I'm not involved there.

So, in short: that's where video-support-on-linux is heading, but we're not anywhere near finished! I wouldn't expect (but am in no way authorititive) Hardy to have any drivers that depend on kernel-module-modesetting.

[vatavr]

Hi,

could someone please post *very* briefly how to compile 2.6.23 for Gutsy?

Using make-kpkg I got the kernel_image.deb, but modules_image won't work, I get some "script"-output and no .deb

[Jeanpaul145]

Hi,

could someone please post *very* briefly how to compile 2.6.23 for Gutsy?

Using make-kpkg I got the kernel_image.deb, but modules_image won't work, I get some "script"-output and no .deb

My kernel -compiling is a little rusty, but I think it goes like this:

Downoad the sources, run xconfig / menuconfig, configure the modules you need, exit xconfig / menuconfig, run make.

Yay compiling your own kernel!:guitar:

However, don't be surprised *if* some things break with 2.6.23...

[bollix47]

Hi,

could someone please post *very* briefly how to compile 2.6.23 for Gutsy?

Using make-kpkg I got the kernel_image.deb, but modules_image won't work, I get some "script"-output and no .deb

Have a look here (http://ubuntuforums.org/showthread.php?t=311158).

I used it for .23 and it worked fine except for fglrx (ati). It worked but extra work is necessary to get it to work well. You won't have restricted drivers and that might cause you problems in other areas.