Linux 2.6 Kernel Stability Freeze

An anonymous reader writes "Linux Creator Linus Torvalds released the 2.6.0-test7 Linux development kernel today and declared a "stability freeze". It has been made quite clear that from this point only "strictly necessary stuff" will be accepted, clearing the way for an official 2.6.0 release sooner than later... possibly at the end of this month."

Comments

Time to upgrade!

[caluml]

Great! That means it's really stable now. I shall upgrade the fw at work to this tomorrow. DNS and mailserver as well.

Re:Time to upgrade!

[Short Circuit]

Whoops! Hold on a second.

The stability freeze only means that no new features will be added. There are still lots of bugs to be worked out. Else we'd have a 2.6.0 release instead of a freeze.

Re:Time to upgrade!

[caluml]

Duh! Don't you know anything? Linus 2.6 is the most stable OS ever. A man at the pub told me, and he works in IT selling computers, and I believe him.

Re:Time to upgrade!

[RealAlaskan]

Great! That means it's really stable now.

BZZZZZZZZZT! WRONG! It's a "stability freeze". That means that the stabilizers are frozen, and can't function. So, it will be unstable until they get the stabilizers repaired.

Sheesh. Goofy kids these days...

Re:Time to upgrade!

[Berzelius]

You haven't paying attention at physics class, even the so called 'stable' elements show a certain amount of 'decay'. p.s. I guess the fall-out here is the other OSs ;-)

That's good

[ixt]

2.5 has been largely successful, and a lot of end users were able to compile it. 2.3? That's another story. I remember not being able to compile 2.3 once.

Good job to all the kernel hackers.

This month will certainly go down as

[Maskirovka]

The October of cool new toys:
Sony PSX
Panther (Mac OS 10.3)
2.6 kernal
Half LIfe 2
Ow! Ouch! Sorry!

Re:This month will certainly go down as

[sharkey]

The October of cool new toys:
2.6 kernal

Still waiting on a spellchecker, though.

Re:Features?

[Anonymous Coward]

Oh, I dunno, SCO registration form on the first boot-up?

My Module

[blackmonday]

I wrote a speaker bracelet module. Alas, it's been rejected because I turned it in too late. It was really cool though.

Re:My Module

[Lxy]

It's OK. I got my truck candle module to compile against it and I can release a patch.

Stability freeze: In related news..

[adeyadey]

..Microsoft, after the latest virus attack, have declared an instability melt..

Re:Stability freeze: In related news..

[gl4ss]

because it's partly funny, and got modded as so probably first.

theres a thin line between troll and funny.. and if something is modded funny it's much more likely that it will get modded funny instead of troll by the next guy modding as well.

though i'm pretty sure you wont bother to check back to actually read this answer since you're an ac.

Reiser 4

[agrippa_cash]

So it looks like we'll have to wait a while longer for Reiser4, or were some of the Reiser4 implimentation problems due to the shifting kernel patches? Anyone? Anyone?

Re:Reiser 4

[caluml]

More importantly, is XFS in there by default? I haven't tried it since about 2.5.59. It's annoying when patches made for vanilla 2.4 don't apply on 2.4 + XFS. If the vanilla kernel came with XFS, those patches would be made against that, and would apply.

Re:Reiser 4

[caluml]

Then the answer is yes. Reiserfs for lots of small files, XFS for lots of big files or its nice ACLs, and ext2 for that /boot partition. Ext3 over my dead slow body.

Re:Reiser 4

[Skeme]

Good question. It will appear in someone's (I believe AA's) tree in a couple months or weeks. From there it will stabilize and get added to 2.6. Here's the latest status update from Hans:

The filesystem is getting reasonably stable.
This weekend we hit a bug in space reservation, which we can't reproduce yet but probably isn't too hard to find by code inspection. There is some thought that the assertion not the space reservation is buggy, in any case we'll release a snapshot after it is fixed.

Our performance is generally wonderful and getting better.
It has the following weakpoints:

* We allocate a "jnode" per unformatted node in the filesystem. The traversing of these jnodes consumes more CPU than performing the memcpy from user space to kernel space when doing large writes. I don't yet really understand on an intuitive level why this is so, which is a reflection on my ignorance as it is consistent with stories I have heard from other implementors of filesystems who found that eliminating per page structures was an important part of optimizing large writes. We will fix this by creating a new structure called an extent-node that will exist on a per extent basis, and this will probably cure the problem. This will greatly simplify parts of our code for reasons I won't go into, and it will also take us 6 weeks to do it. I don't think users should wait for it, and so we will ship without it.

* Our dbench performance was poor, has improved due to coding changes, and we need to test and analyze again. Perhaps more fixes will be needed, we can't say yet.

* Our fsync performance is poor. We will pay attention to this next year, frankly, after we have fully implemented the transactions API. At that point we will say something like, if you care about fsync performance you should be using the transactions API and/or sponsoring us to tune for NVRAM, users will say back "but our legacy apps on hardware without NVRAM matter!", and we will grudgingly but effectively tune for this because we care about real users too.;-)

Nikita recently invented and implemented a clever bit of code that keeps track of the highest node in the tree that spans a directory, and then performs repeat lookups within the same directory starting from there rather than the root. This is a nice answer to those who keep asking me, wouldn't it be faster to have separate trees for each directory? Now I have better answer for them --- nice work Nikita. It also has the nice side effect of reducing spin lock contention on the root node for 4-way SMP.

I am hoping to move my laptop to SuSE 9.0 running reiser4 sometime this week, and I am hoping we will ask for more outside testers to help us find bugs at that time. While I have mentioned only the performance flaws in this email, our overall performance seems to leave little doubt that the filesystem as-is is far better than V3, and even though it will get much faster with another year or so of tuning, if now we are the fastest available on Linux, we should be shipping now (assuming we find no new bugs in the last round of internal testing).

Benchmarks can be found at www.namesys.com/benchmarks.html

As you can see in those benchmarks, in V4 tails IMPROVE performance due to saving IO transfer time. This is a great improvement over V3, and generally speaking V4 stomps all over V3 performance. It also scales better, has plugins, and improves semantics a little bit (big semantic improvements will be in the next major release not V4).
You'll also notice that we increased the size of the fileset to be more fair to ext3, and we tested some ext3 configurations Andrew Morton suggested testing.

--
Hans

Not "possibly at the end of this month"

[JoeBuck]

Linus wrote: In other words, this should calm things down so that by the end of October we can look at the state of 2.6.0 without having a lot of noise from 'not strictly necessary' stuff."

That is, at the end of October he will "look at the state of 2.6.0". That's quite different from shipping it.

Who cares.

[BoomerSooner]

Mac OS is on 10.3, that's like 7.7 better. And no fair skipping like MS does. Windows 95 my ass, more like Windows 3.11b

Yes, but

[Anonymous Coward]

Is he an MCSE?
I wouldn't trust anyone else's opinions.

Re:Yes, but

[sharkey]

Is he an MCSE?

I was there. He said his name was Darl Mc-something, and he had a lot of his code in the Linux 26 thingy, and that's why it works.

Stability?

[Wooky_linuxer]

When the 2.4 series came out, it was much criticisd for not having anything near the stability of the old 2.2 series (I'd say it haven't catch up yet,but since I use it in a desktop machine 2.2 is not an option)... What can we wait from the brave new world the 2.6 kernel will bring?

Re:Stability?

[efti]

I've been using the 2.5 series since 2.5.66 or so. The main reasons I recommend 2.6 are:

• Greatly improved responsiveness under heavy load -- I no longer notice cpu-intensive tasks like a kernel recompile or the slocate database rebuild cron-job happening in the background. And X isn't even running with higher priority.
• Built-in ALSA (Advanced Linux Sound Architecture) -- much improved audio, especially audio recording
• Improved ACPI power management and CPU frequency scaling (my main machine is a laptop)
• Software suspend (just like hybernate on Windows), again handy for laptop users, or those who like to sleep without listening to the whine of their super mega cooler CPU fan / vacuum cleaner attachment.
• Built-in IPSEC support. This is mostly useful for those who need to set up VPN tunnels. I imagine it is more efficient to handle IPSEC inside the TCP/IP stack itself

These are the ones I can think of off the top of my head. I haven't used the built-in IPSEC yet, and software suspend still doesn't work properly on my laptop, but it's not far off. 2.6 will be a pretty sweet series.

Re:Stability?

[wfberg]

More hypocrisy, like before. Linux can put out several filesystem corrupting kernel releases and major showstoppers as in the 2.4.x series, but if a user-transmitted e-mail virus makes the rounds, it's a "Microsoft hole."

Linux doesn't put out Linux releases, Linux is Linux. And Linux is used in several distributions - you can get a five nines Linux distribution if you like. Bugridden open source software does get flack -- distributions don't incorporate a kernel they don't feel comfortable with (RedHat's kernels are heavily patched for instance), no-one will touch wuftpd with a 50 ft pole, people wil nag authors with patches, fork or start competing projects (qmail, postfix vs. sendmail) etc.

Speaking about qmail and open source software getting flack, ever read DJB's comments on BIND and sendmail? Or ANY holy war? (BSD vs. linux, EMACS vs. VI,.. )

User-transmitted e-mail virusses? That's called a trojan horse. Recent worms -- exploiting holes on Microsoft's e-mail client running on Microsoft's operating system and Microsoft's browser -- depend on bugs and design in Microsoft's software and that's squarely their responsibility (e.g. why is RPC even listening to anything but localhost by default? If you needed it to listen to the entire internet, you'd know and could change the default).

Besides, those crappy kernels you mention haven't affected me one bit. Whereas I've spent quite some time getting people to install patches, firewalls, and remove those darned worms.

Some people may have a certain amount of unfounded (or at least, not founded in technical fact) animosity towards Microsoft, but let's face it, most mature open source software we rely on is much, much more secure, stable and well-designed than MS Outlook and its ilk. And that most certainly includes the Linux kernel. Comparing apples and oranges, maybe (the 2k/XP kernel isn't half bad either) but that doesn't mean that Microsoft should get away with crappy products that aren't kernels.

Re:Stability?

[Pharmboy]

Remember the 98a defrag corruptions?

Remember how Microsoft told us to never "Always trust content from Microsoft"?

Did you know that "Linux" is about 35MB of kernel source code only? Its not an operating system, its a freaking kernel. Mandrake, RedHat, SuSe and even Caldera are "Linux distributions". I installed several distributions in the 2.4 series, and at least a dozen different kernel releases, and never saw, experienced or heard about any file system corruption due to the kernel. Then again, most intellegent people don't load the freshest kernel on a production machine.

As a matter of fact, you don't change the kernel unless it has an errata that actually affects you, you need the new features, or you are foolish. At least with Linux, you have a choice of kernels. You show me a smart Linux admin, and I will show you someone using a kernel that is at least a couple versions back.

There are Linux viruses, btw. The difference is when a Linux box gets infected, it is designed with permissions that make it impossible to send it out to every other Linux box out there. Its by design.

Re:Stability?

[Overly Critical Guy]

I installed several distributions in the 2.4 series, and at least a dozen different kernel releases, and never saw, experienced or heard about any file system corruption due to the kernel. Then again, most intellegent people don't load the freshest kernel on a production machine.

Then you weren't paying attention. 2.4.x was a complete wreck, and everyone complained. I still remember the infamous Thanksgiving "turkey" kernel that randomly corrupted ext3 partitions.

The base definition of an operating system is the kernel running it, dummy. Linux is the system operating my devices and letting me operate my computer.

Windows also has permissions. Every single Windows network I've ever run or worked with operated the same way.

Another "GNU/Linux" weenie. Here's the part where someone mods me down instead of posting in disagreement.

Re:Stability?

[Anonymous Coward]

Face it. All the Linux users will freely ignore and disregard corrupting kernel releases. If Microsoft even dared do something like corrupting everyone's NTFS installations, they'd be eaten alive by hundreds of fanatical posts here.

We can ignore the kernel releases that have had major problems because the only people who run them are the developers and the geekiest of the geeky Linux users already. The problem is then fixed before anyone uses, (hopefully), the kernel on a production system. The big difference here is that Linux is just the kernel, if Redhat released a distro/OS that caused massive filesystem corruption and major problems then people would indeed be posting here flaming RH for not testing an operating system and allowing major flaws of that nature. A kernel release is quite a bit different than an entire OS, you have certain expectations when you purchase/run an OS. Compiling the latest kernel on your system should not come with the same expectation of quality that you would expect from a full fledged and tested OS. But as we have seen with Windows 98a those expectations are not always met.

And those Windows problems have never, ever affected me.

Well you must be living in a different world than me. The bandwidth that all of these worms are creating affects my servers every day. I drop thousands and thousands of packets targeting windows specific flaws every day. I have to filter out the "install this patch!" emails that are propagated by millions of idiot Windows users. Thousands of those emails. When the blaster worm shut down the part of the US govt that was reponsible for Visas it cost tax money out of my pocket to fix it. Same goes for all the other damage done by these worms and exploits, (regardless of platform), it's money out of my pocket.

Might I also mention taking a look at my sig and seeing that OSS is just as full of bugs as Microsoft? Need I mention the "Linux Most Attacked OS" article Slashdot posted that linked to a study showing Linux as the most breached server on the net? Need I mention ssh/ssl? Bind? Sendmail? I could go on and on...

You know that that article was a bunch of crap, people have been responding to that every time that you bring up the issue, the group who did the study has been repeatedly discredited. How do you define breached? Did it count all of the thousands of Windows servers that were breached with blaster, viruses, and the other various worms that make the rounds? As for the vulnerabilities in ssh/ssl, bind, sendmail... exactly how many million computers were taken down by exploits for those vulns? How much damage was done? What percentage of people had their boxes owned thanks to those exploits?

Re:Stability?

[broeman]

also by the side of IPSEC, cryptoapi/loop is shipped with the kernel, making it easier to encrypt your disks (eventhough it needs a newer util-linux, which has less options, like weird ways to choose different bits).

Re:Can you help me reinstall iLife on my iBook?

[Gherald]

Do the internet a favor and click the "Shutdown" icon.

2.6 seems unimportant for me

[MarcQuadra]

I've tried 2.6-testX and it doesn't seem to do all that much more for me than 2.4 does. I remember moving from 2.2 to 2.4 and there was a LOT more that I could use, USB and ReiserFS and quite a speedup. 2.6 seems to perform about the same as 2.4 on my boxes though.

Maybe I'll have to wait until I get a TCQ-enabled drive and see if that makes a difference.

Re:2.6 seems unimportant for me

[hey]

What about O(1) switching.
This should make the desktop smooother.

Re:2.6 seems unimportant for me

[Gherald]

> 2.6 seems to perform about the same as 2.4 on my boxes

Yes 2.4 and 2.6 are very similar, but 2.6 does have a couple advantages. Asside from the exta .2 you also get increased bragging rights, i.e. you get to laugh at people still using that backward 2.4.22 ;)

Re:2.6 seems unimportant for me

[superchkn]

That probably helps too, but had you tried say test4 and then applied Con's O1 patches, that's one hell of a difference.

With mozilla compiling in the background and vanilla test4, it's very hard to position my mouse pointer accurately on the screen while Mozilla is parsing a new page. Adding Con's patch fixes things to the point that if things didn't take longer to load, I wouldn't even notice that the compile was occurring.

This is on a pretty low-end system though ->K6-2 400 w/384MB

In pratice

[rf0]

I've been using 2.6.0-test4-mm4 daily without problems. No glitches. The 2.6.0 kernel has real improvments in the shape of Alsa being mainstream. Also the I/O schedular + interactivity is much better under load than the 2.4 kernel.

Of course however I won't be putting 2.6 into production use until at least 2.6.8 or there abouts to make sure there are no nasty surprises in there

Rus

Re:Blah blah

[Anonymous Coward]

Well, first, increase your expectations ;-)

I've not had such luck with XP as it lumps most everything onto one IRQ and then stutters as 6 devices fight over the same IRQ. The MS provided NEC USB 2.0 PCI card driver likes to BSOD too. I expect the OS to be able to handle a machine stuffed full of cards (AGP, 4 PCI, 1 ISA).

I'll admit though that from test3 to test6, the 2.6 kernel no longer acknowledges my BIOS setting that's supposed to keep PCI cards off IRQ 5 (for my ISA soundblaster), but I can live without sound (not to mention this is a beta and my problem is logged). I can't live with an OS that constantly stutters about and crashes all the time, which is what XP does.

Re:Blah blah

[Anonymous Coward]

Sure, but wait until you turn the computer on!

Re:Is it faster?

[miketang16]

just for kicks... what filesystem might you have been using? Crazy idea... I know...

So what's cool about 2.6

[Alien Conspiracy]

I'm still on kernel 2.2 with debian/stable. My servers have been running 2 years without a reboot.

Is there anything really cool in 2.6 to convince me to upgrade?

Re:So what's cool about 2.6

[Pierre]

I can't even keep electricity that long.

Modular source code?

[pjack76]

How difficult is it to only download those kernel modules I actually want to compile? As time goes on and new stuff keeps getting added to the kernel the source just gets out of hand. Someone should set up a little webby clicky thing that's like "make menuconfig" but then assembles a tarball only containing your precise configuration and those modules you've selected. Just a thought.

Re:Modular source code?

[MBCook]

While a neat idea, I think that it would be very hard to implement, and the tradeoff isn't a good one. What tradeoff?

Well, the patch for -test7 is just over 650k if it's bzipp2ed. While it would save space/time/whatever to only download a i386 version (no PPC code, etc), or a version without all the sound code (I'm making a server, I don't need that...) etc, you run into a problem. Just what are you going to do? Let's say you have 4 architectures, the sound option, an ISDN/amature radio option, some PPP/dialup stuff in an option, and two others. That's 4x2x2x2x2x2, or 128 possible combinations of kernels. Now there are more than four architectures, and there are probably some more things that could be put into downloadable options.

So, are you going to do everything through CGI scrips that fetch the right one? Are you going to use those scripts to package and test each of the possible sets of options to make sure no important code gets left out? Is it simply going to be a single choice: x86 or "other"? Do you want to move things like sound to external patches like ALSA has been for such a long time?

Really, I don't think that this is worth it. It's a neat idea, and if you want to make some things available on your personal website or something like that, go ahead. More power to you. It's a good idea. But for it to be "official", just seems like too much of a hasle.

Re:Modular source code?

[Chops]

This is an LKML FAQ. The short answer is, "Yes, it would be cool, but it's much harder than you think. We're busy doing kernel code; you're welcome to do it yourself if you wish. People would thank you profusely."

Re:Is it faster?

[Tribbin]

The 2.4.23 was not out by then, it's not even out right now.

www.kernel.org

2.4.23-pre6 is the latest.

Get your facts straight

Strictly necessary stuff?

[bennomatic]

Wait! Wait! Wait! I've got a million lines of SCO code I want to insert!

Re:Strictly necessary stuff?

[GlynDavies]

Bad luck, looks like your too late. Still, try contacting IBM or RedHat - they may have some alternative suggestions on where to insert it. ;-)

Re:Is it faster?

[Anonymous Coward]

with 2.4.22 and ReiserFS it takes me:

real 0m0.244s
user 0m0.010s
sys 0m0.110s

to copy a 15 MB file across partitions (same physical drive). Takes me 24 seconds to copy it across network to another computer.

You are spinning your hard drives with your hand or what? Try harder!

Re:Is it faster?

[Anonymous Coward]

Can it copy it in less than 2 minutes now like Windows Server 2003 on the othyer partition of my computer?

That's really not saying much. Usually when a person says, "in under x minutes" they really mean, nearly. So from this we can deduce that this computer that you allegedly own transfers at about 0.14 MB/sec. Wow, I think my IDE ZIP can do better than that! Not to mention my P233MMX w/64MB and an ancient 1.0 GB Quantum Fireball running in MDMA2 on Linux 2.4.18 outperforms your computer running Windows 2003 with exceedingly more powerful hardware.

Damn, I'm staying away from that Microsoft bloat, holy hell!! ;-)

Radeon FB fixed?

[fishbowl]

Is the Radeon FrameBuffer Console fixed?

It's been horribly broken in the 2.6 test kernels I've tried.

Re:Radeon FB fixed?

[efti]

Nope, sorry, no changes to drivers/video/radeonfb.c. Are either of these bugs the problem you're having? If not, then you should report it here.

Re:You can boot into Reiserfs

[Theatetus]

Because you rarely write to the boot partition, you shouldn't mount it, and it's a bitch to tell GRUB to ignore the length of the journal.

I miss 'make dep'

[Theatetus]

It's not the same thing without 'make dep && make clean bzImage modules modules_install'

Now it's just 'make menuconfig && make'

Linux has gotten soft... time to migrate to BSD. I would if I could get my laptop's touchpad to work. Sigh...

Re:I miss 'make dep'

[anthonyrcalgary]

Sadly, BSD is getting soft too. Everything on my OpenBSD system keeps working without me having to get my hands dirty. FreeBSD is worse. Don't even get me started on OS X...

NetBSD still has promise though. It's practically unusable. It doesn't even have sudo installed by default. If you're in a masochistic mood, try it out.

ataraid

[thrift24]

Does anyone know if ataraid is in the kernel yet, or what exactly they plan on doing with that? In 2.6-test6 there wasn't a trace of ataraid around. This is bad news for anyone wanting to upgrade to 2.6 who use highpoint or promise raids. Wanted to install gentoo w/ 2.6 on the girlfriends computer a couple days ago when i found this out, now she's running a heavily patched 2.4 kernel and ataraid is buggy...It would really suck to not see a working ataraid driver in the 2.6 kernel

Re:Is it faster?

[efti]

Don't you know that Windows does fake copying and actually copies in the background? It is really annoying on removable drives, and its dangerous.

That's called buffered I/O and is a standard feature of modern operating systems. Where it gets dangerous is that Windows doesn't force you to manually unmount removable disks before pulling them out, which can easily result in data loss. But that's what the little light is for next to those drives. If the light is on, don't take the disk out or you will lose data.

Oh, and always stop the hardware before removing USB or firewire storage devices as on Windows that's the only way to be sure that all the data has been written to them.

Re:BSOD module

[Pharmboy]

But then again, Windows doesn't have a DLL for kernel panic either. I am not sure if its because the Windows kernel is apathetic and simply doesn't care or what.

On a lighter note, back in the windows 3.1/Lantastic days, I used to mess around with a program called "The Draw" (i ran a bbs, figure it out or google it) which could turn an ANSI screen into a .COM file. I would create a false BSOD that would say something along "Windows has detected a dumbass on the wireless end of the keyboard. Please use a pencil and paper instead" and place this in autoexec.bat, just before a "pause >nul"

The funny thing is half of them would tell me they have a "blue screen thingy" without reading it, giving me the opportunity to ask them "what does it say?". Its much more fun to hear them actually read it out loud over the phone intercom.

Slackers

[DaveAtFraud]

Slackers:

[dave@bend ~]# cat /proc/version
Linux version 2.6.0-test7 (dave@bend.local.davenjudy.org) (gcc version 3.2.2 20030222 (Red Hat Linux 3.2.2-5)) #1 SMP Wed Oct 8 19:09:28 MDT 2003
[dave@bend ~]# uptime
19:37:24 up 18 min, 8 users, load average: 0.62, 0.20, 0.13

So why haven't *YOU* built and booted with 2.6.0-test7 yet?

Re:You can boot into Reiserfs

[Chuck Bucket]

from my /etc/fstab: /dev/hda1 /boot reiserfs noauto,noatime,notail 1 1

Yeah, that was a bitch.

P

Re:BSOD module

[some guy I know]

Does linux have a BSOD module in the kernel?

I heard that BSOD was dying...

CD-RW / DVD Packet Writing?

[PastaAnta]

Does someone know if packet writing is in?

In kernelnewbies status list it is listed as pre-2.6.0 stuff, and the patch has been around for ages. I very much hope we will finally be able to use CD-RW's instead of the antique floppy drive. It is frustrating and somewhat embarrasing Linux still does not support this feature. I assume DVD-RAM/-RW/+RW etc. also depend on this?

Pretty, pretty, pretty, Please!

Re:You can boot into Reiserfs

[stor]

> Compareded to a friendly wizard? Yeah, it was.

If I had a friendly wizard I'd have no need for a computer.

Cheers
Stor

kernel freeze stable?

[Tukla]

I suppose so. My computer has always been very stable after the kernel froze.