Slashdot Mirror
linux 2.2.9 Released
rohirrim was the first
to let us know that the hit hot single 2.2.9 has made its
debut on the usual FTP sites. So if you're the type
that desires a newer stable kernel, get to it.
← Back to Stories (view on slashdot.org)
A kernel a day
keeps sanity away
by time you have untarred one
another has come
has anything changed?
is linus deranged?
i state with much glee
i use freebed!
Just think for one moment - if there was no interest in having this on Slashdot, then how come there are so many comments and discussions about it then?
;)
;)
Malda! Please give us a poll on this to get it sorted once and for all
thanks
According to what Alax Cox said, they don't really care about this Microsoft-driven test. Mindcraft is just trying to find some kind of credibility after all that they have done (if you remember their tests on Novell vs. NT, you may understand how they're dumb and market-driven). i don't think neither Linux or Alan or any other serious Linux advocate could care about a company which have no grade of independency.
Marco Ermini
linux@markoer.org
This is typical - the last stable 2.0.x was 2.0.37 wasn't it... come on people. Obviously you haven't been using Linux for long enough to criticize the kernel revision process.
This is what makes Linux BETTER than 'doze... things get updated quickly. A few days is much better than Microsoft's response to NT service patch 4.
There once was a man from nantucket
who thought "one more compile won't muck it"
He said with sadistic gaul,
as he typed "make install"
"If this breaks an inode, just fsck it".
There were some changer to buffer write back .
( fs/buffer.c ) in the 2.2.8 , which were
reverted back in 2.2.9
I believe there was a good reason for that.
I suggest staying with 2.2.7 or upgrading to 2.2.9.
Just to be on the safe side
stein
Yes, Don't run 2.2.8 or 2.3.0!! Or at least,
don't do any heavy disk IO while running them.
Hopefully 2.2.9 is a fix for the bug in the
disk buffering code.
Lots of people download the kernel. It's much more efficient to just have tarballs for people to download. :)
Too free, or too much of a free thing? I too don't really understand these people. I guess they are just conditioned from Windows type OSes where there are NEVER fixes to anything (games sometimes put out patches quickly, but stop after say 2 months of the game being released :)). Do people really equate never releasing a fix as, "This software is completely stable and perfect?"
It's one thing to want Linux perfect, but holding back fixes isn't in our nature...
I was just wondering, I remember seeing 2.0.37 prepatches for a while, but a final 37 never came out. Have they officially retired the 2.0 line, or are people still working on making it stabler?
Comments like that one aren't needed... you can use freebsd with pride cause it's a good system. Is it the same as linux? no, and I'm glad. Because varity is good.
:)
Do I care about caps, spelling, or anything else at this point? No, cause I'm on vacation.
David
This sig left intentionally blank.
Linux kernels are:
2 = Major version
2 = Minor version
8 = Patch level
and optionally
-x = Pre-patch level
Some programs are:
Major.Minor.Build
But I've never seen a program refer to the third number as the "micro" version.
--
The world is neither black nor white nor good nor evil, only many shades of CowboyNeal.
Slackware doesn't have this problem, thankfully. It depmod's /lib/modules/[version loaded number here] every time, so there's no conflicts.
I belive I have under /lib/modules:
... but no default!
2.2.6
2.2.7
2.2.8
2.2.9
---
Spammed? Click here for free slack on how to fight it!
--
# Canmephians for a better Linux Kernel
$Stalag99{"URL"}="http://stalag99.net";
Whichever way you decide to go, there's no need to wait... (see my previous post above). :)
Er... excuse me? 2.0.36 hasn't been out for any 436 days...
Earlier this week, I had to reboot one of my machines (because of a failing hard disk). It had been up for 273 days, running 2.0.35, because that was the latest stable kernel when I brought it up...
The only problem I ran into was in smbfs. For anyone out there who hasn't got even the modicum of C it takes to figure it out, the solution is to add a semicolon at the end of line 91, in linux/fs/smbfs/inode.c ("insert_inode_hash(result);") (I was going to include a patch, but Slashdot mangles it).
Yup. I grabbed both of them, for different machines. There were already three pre-patches out towards 2.3.1, anyway... the 2.3.1 final patch presumably would contain all of that stuff, plus the vital bit that rushed 2.2.9 out.
I suppose I could diff the two patches and see how much overlap there really is, but I'm too lazy to figure out the switches that would give me some useful information...
But my kernel's done cooking... reboot time...
According to "finger @linux.kernel.org", 2.3.1 is out as well. I suspect that whatever problem (I heard something about filesystem corruption?) that caused the quick release of 2.2.9 encouraged a similar release on the 2.3 tree... 2.2.8 and 2.3.0 are identical, after all...
And anyone who's got a problem with letting people know that there's a new dev kernel out can flame away. I've got a thick skin.
AFAIK, the "preferred" link is made in the startup scripts. I got to where I didn't like it much and just make sure I have a module directory that matches my kernel version. I do a 'depmod -a' then modprobe my modules and all is well.
Codifex Maximus ~ In search of... a shorter sig.
Posted by Myrdraal:
Kind of unlikely. The only change was one variable for char to signed char... Actually, I probably shouldn't have even noted that change because I believe chars default to being signed.
-Myrdraal
I have a motherboard with the ALi Aladdin V chipset that includes the M1543 IDE controller, which has always worked fine for me (I was running 2.1.something when I bought it). So what does this mean? Is there functionality that I wasn't getting out of my M1543 before that I will now if I upgrade to 2.3.1? Anyone know?
I hope not. The last thing we need is Microsoft driving the development of Linux. Since there are no shareholders involved, Linus has the option of just ignoring them, and working on things that are the most important, instead of putting out fires in the press.
TedC
New features come out all the time for older hardware that make it run faster. MTRR is one, and sound fixes have been most useful.
I've been in the upgrade cycle for a while waiting for better functionality (it works but not like I think it should) of NFS.
Improving memory management is another thing constantly speeding up the kernels. And for the late 2.0.x kernels improving PCI support helped out alot too.
So the moral should be don't upgrade unless you know what your getting? Or maybe upgrade unless you have better things to do.
for me I always keep kernel source patched to the most up to date possible. But I don't compile/reboot unless I have to or find a good reason too. IT also saves time in downloading an entire source tree five or ten revisions down the line.
^~~^~^^~~^~^~^~^^~^^~^~^~~^^^~^^~~^~~~^~~^~
I think it means the inducement to the sheriff not to close down the brothel :)
If you have a multi processor machine then:
:)
:)))
1) You're a lucky bugger
2) I seem to remember a post about Linus sorting something out that affected multiprocessor machines in the 2.3.1 pre patches and i assume this is included in the 2.3.1
If you have a uni processor machine then just sit back and enjoy the apparent speed gains whether they are real or not.
>nothing wrong with a bit of 'healthy' competition
Whoever said anything with MircoSoft involment was 'healthy'? Certainly not anyone who's familar with MicroSoft's history and it's behavor in the DOJ trial in presenting rigged tests there also....
I have no clue abt coding, but would the recent release of the quake3 test have anything to do with this? yummy...
-----
If Bill Gates had a nickel for every time Windows crashed...
Stop reading them, stop posting on them. Skip and move on. Create an account and explicitely remove "Linux" from your topic list.
If you can't seem to handle that, at the very least STOP WHINING.
According to Mydraal on his comment at Cutting Edge Linux, 2.3.1 has compile problems. No te jodas la vida, and go the stable way.
In Soviet Russia, Jesus asks: "What Would You Do?"
You know, if things like this bother you, Rob added a filtering mechanism just for stuff like this. I wanted to know. It fixed some problems I was having. I read slashdot, because I can count on it getting news out quickly.
Don't like it? Filter it out.
----------------- "I have a bone to pick, and a few to break." - Refused -------------------
I'm completely confused by the negative comments here. What is the problem? If you're not interested in a particular revision of the kernel, avoid downloading it.
Unless you're completely anal retentive, it shouldn't matter if you miss a kernel revision. Just wait until you see something worth upgrading to, and be happy that those who had problems (Oracle, hd buffer stuff) had it fixed before they even had time to complain.
----------------- "I have a bone to pick, and a few to break." - Refused -------------------
This release may fix the filesystem corruption problem that some people had with 2.2.8. It may be worthwhile looking at it if you installed the previous version.
I'm just going to keep trying to upgrade until I find a kernel that'll even compile on my machine; the last one that even built was 2.2.6, and I only managed that once.
GPL is one thing, but I still think RMS is going to throw a fit when he hears about this one.
Nonetheless, I like it. I hope there'll be some nicer-looking clients out soon, but the code's open enough that doing that is a possibility.
That should be your philosophy with EVERY kernel. Why upgrade if you're not having difficulty? I'm still running 2.0.35 on a couple of my boxen, with no problems whatsoever. And I'm sure there are thousands of users out there with earlier kernels still chugging happily. The only reason to upgrade with each kernel, besides having problems with old ones, is if you're a developer (in which case you should be keeping up with the pre-patches..)
Just because there's a new kernel is no reason to upgrade.
Log
we're running some machines with 2.0.36 too :)
hany
More Kernls to Compile Downloads take awhile Did I spell that right? A kernel haiku(sp?)
Consider the whole Mindcraft deal. Now consider that, according to ESR's CandB paper that part of the motivation for creating open source software is satisfaction of ego. Now imagine you're kernel hacker and MS has, either by design or accident, thrown down the gauntlet. Note to large, Redmond based software companies: Mindcraft is a loss even if it's a win. Let it lay....all you're going to do is make it worse.
Many thanks to all the kernel hackers out there!
Keep up the great work and don't let all this bickering
about version numbers and too frequent releases get you down!
Accelerated-X from Xi Graphics is the choice
for G200 development. I've been running my
G200 since they came out and Xi has been the best
manager for this card.
Regards
do what I do (for same rev kernels, like 2.2.x -> 2.2.(x+y)): .old, resymlink so that _at next boot_ the new kernel comes up, while leaving a lilo entry for the .old kernel
* run a stable kernel
* build a new kernel, mv orig kernel to
* keep rebuilding the new kernel, while not rebooting unless the system crashes.
Just don't forget to lilo and you should be fine. Now stop whining and enjoy the fruits of much labor!
I think it's because they -have- been conditioned by Microsoft and a lot of other software entities. When commercial software gets released to a new version, there is usually a marked difference (notice I didn't say improvement) in the new product.
So when a 'new version' of anything Linux is released, the first reaction by people like Splatty is that of a sheep: "Baaa, must get new software, it's a new numbaaaa, it must be better. Wait. This is a minor patch! Whaaa!! You guys suck! Stop making me download this stuff!"
And they never consider the possibility of reading the changelogs FIRST, or learning that the third number in a Linux kernel version indicates a MINOR patch.
How some of these people ever managed to learn how to use Linux is a mystery to me.
Hokey statistics and ancient misconceptions are no match for a good thought in your head, kid!
Hey Rob & Hemos,
:-)
How about posting the Linux Kernel news under a different heading than "Linux" (perhaps "Linux Kernel")? That way anyone who feels that kernel updates are not newsworthy can filter them out and not bother complaining about it to those of us who find it interesting.
My own opinion: I happen to enjoy seeing the new Kernels posted here. It often prompts some good discussion, even though there are quite a few "who cares about new kernels" threads. Also, frequent kernel updates are one of the beauties of Linux. Quick features and bugfixes...gotta love it! Besides, nobody says you MUST download newer versions anyway. If it works for you, stick with it unless you really want to try it out.
Cheers,
Vic
Perhaps he means hardware 3D accelleration support.
It seems to me like there are often interesting discussions regarding the changes in the newly released kernels, here on slashdot.
In that sense, I think Slashdot as a forum works well. As long as there are people who post with interesting content, there is obviously some interest in the event.
Does anyone know what changed in kernel 2.3.1 from the 2.3.0 version. I just compiled the kernel and the machine feels significantly faster. It does not use any Swap on light load.
Help fight continental drift.
Not unless the 2.3 series has some mind-blowingly fundamental changes to the kernel. Fixes, enhancements, and minor features would in all likelihood lead to a 2.4.0 -- and since Linus has said that he doesn't intend for the 2.3 series to run nearly as long chronologically as did 2.1, odds are there won't be time for anything that spectacular to make its way in.
Learn to spell: nickel, missile, lose, solely, amendment, speech, kernel, probably, ridiculous, deity, hierarchy, versus
Use /sbin/ipfwadm-wrapper and you don't have to change a thing. No, you won't get the fancy IP chains, but everything will work at least and you can have a life instead of reading more documentation...
-Mike
You would have to make Linux maybe 5x faster to make Mindcraft benchmarks work - essentially a Complete Waste Of Time.
And it's dollars to doughnuts that the other benchmarks quoted by MS on their page were against SaMBa set up for security (default) and then benchmarked for speed.
One benchmark graph in particular showed what looked like a benchmark of CGI versus server-module; in other words, it wasn't benchmarking like against like. Surprise.
Do these benchmarks also verify that the received data is actually correct? The thought I have is that if not, "we" could do the same thing that MS appear to be doing, i.e. set everything up to be as fast as possible and to hell with the risk of corruption, and show them what a _real_ server OS can do. Full speed ahead, and damn the checksums! (-:
Got time? Spend some of it coding or testing
Contrary to popular belief, upgrading to each minor kernel revision is not required by law!
The Kernel Gestapo is not going to bust down your door and demand to see your papers and ask why have you not upgraded to zee newest release?
it seems like you spend more time keeping up with new releases to be able to ENJOY it!
~centurion
Just goes to show that when there is a fix for something, or something new to add, it's out there as soon as possible. Not like Microsoft, who waits until they feel good and ready to release something, if at all. Remember the y2k fix for Win95 they held? That won't happen with linux.
And like upgrading a kernel is all that big of a deal? Grab the source patch, recompile, and reboot. Compile the option to use different module versions, and you don't even have to re-compile those. How easy is that?
The patch is teeny. As someone else already
posted, it should have been a pre-patch.
Most of it's m68k and arm updates... there's a
few other tiny patchlets for tcp.h, fs/buffer.c,
aha152x.c, ide.c, ide-dma.c, and ip_masq_quake.c
inside. That's about it.
Unless you're having trouble with 2.2.8, triggered
by anything listed above, don't bother downloading
it. Wait for the next one.
- Preferences: Solaris 10 (servers), Ubuntu (desktops), Solaris 11 (personal servers) -
Don't you remember 2.0.36 and 2.1.132, etc? It wasn't that long ago....in fact, neither was 1.2.13 (the one that i started with ;)
"The value of a man resides in what he gives,
and not in what he is capable of receiving."
--Albert Einstein
gee wizz.. I am behind.. I am still using 2.0.36.. where is my RH6.0 cdrom and I hope it doesn't have to many bugs... I'll have lots of patches to apply, and kernels to build.. good thing is that it is only 6 minutes a build on my machine...
Only 'flamers' flame!
Splatty, I think I know a solution to your problem. Don't get it unless someone sticks a gun to your head and forces you to. (That would be one really desperate for attention kernel hacker.)
I'm working off 2.2.5 on one system and 2.2.7 on another. And if I'm going to upgrade either of those, I'm going to learn how to use patches instead.
Alternate solution to your problem. Give each release a week or two. You won't be "l33t bl33d1ng 3dg3" (man, I hate that style of writing), but you'll be more likely to have a truly stable kernel.
Digital Wokan, Tribal mage of the electronics age
You mean free as in beer or as in speech?
----------
In a real emergency, we would have all fled in terror, and you would not have been notified.
You meant to say "some people had problems with Oracle on Linux after upgrading to 2.2.8", and not "2.2.9" as you originally said, right?
----------
In a real emergency, we would have all fled in terror, and you would not have been notified.
Try looking in the depmod manpage.
just an idea.
New things are always on the horizon
Agreed.
I think the Mindcraft (and other tests showing relativly poor linux performance) is really great for pushing development ahead.
Among the FUD on the MS page in response to the Mindcraft (and other) benchmarks there are some valid points that I hope can be addressed with the newer kernels like filesystem and memory shortcommings.
------ Live free or Die!
In the changes file for 2.2.7 there was something about this:
"DHCP clients for 2.0 do not work with the new networking code in the 2.2 kernel. You will need to upgrade your dhcpcd / dhcpclient."
It's usually a good idea to look through all that stuff in the Documentation directory that comes with the kernel.
Someone's been watching too much southpark
Lab test show that use of micro$oft causes deadly cancer in lab animals.
Yea, I rememeber that... It was like a new kernel every 2 days... So is 2.2 now about 1 every week?
Lab test show that use of micro$oft causes deadly cancer in lab animals.
Mike
--
Mike
--
"Wi nøt trei a høliday in Sweden this yër?"
(licks the finger and touches the kernel with it) Hshhhh! Piping hot! Not surprising, considering the speed it arrived at... At this rate, the kernels will soon start burning through the bottom of my hardrive!
Seriously , though: Could it be that the kernel folks are trying to fix some issues in time for the 3rd Mindcraft test?.. Is that why the upgrades are coming at such speed?
--
--
Victor Danilchenko
Because Linus doesn't like it (that's the official response).
If you check out http://www.bitmover.com/bitkeeper/ you will find the program that _will_ be used (was meant to be for 2.3 but I guess they started that a little early)
Matt
Uh, what does the GPL have to do with anything?
;-)
XFree86, among a ton of other common programs, are not GPL either. Quick! You better delete them before anybody finds out you're using non-GPL'd programs!
Hang on a minute, your computer hardware was not released under the GPL either! Better put it back in the box, take it back to where you bought it, and tell the salesperson that you're too stupid to own a computer.
Matt
they should start limiting the releases of new kernels to once a month so I can keep atleast a 30 day uptime average :)
It's just Crap.
I disagree. I check here for news, any news. I don't always run over to Linux Today. If it bothers you that much, get off your lazy butt and set filters...oops, I forgot...you're too lazy to even get an account, much less set up filtering.
Of course, the problem with this is it would allow people to make optimizations to look good for the benchmark in question. There was a well-known commercial compiler several years ago (can't remember its name) that contained several "optimizations" that would effectively detect when it was having a known benchmark run on it (a calculation of a large number of primes), and rather than actually doing the math, would simply produce the known total of the series as output. The company actually used this "statistic" in their advertising for some time, claiming they produced code that was n times smaller and faster than the competitors.
Lo! Improved network drivers.
Ever so stable.
'make menuconfig'
Shows me myriad options
Like springtime flowers.
Well, the Mindcraft tests did expose some problems with Linux (or at least raised the priority of some known issues).
For example, Linux did have suboptimal scheduling behavior when many processes are waiting on a single event. This happens when a web server such as Apache forks itself many times to provide ready servers for incoming connections.
The scheduler would awaken all the processes even though only one would actually get the event. The rest would just go back to sleep. So the solution was to just wake one up. Hopefully this speeds up the Linux/Apache web server.
I'm starting to feel the rush..
we'll have 3.0 by the end of the month at this rate.. on a side note, ever since 2.2.8 came out, my dhcp died and cable modem died..bleh prolly something i did though
laters
------ Poo-tee-weet?
Bitkeeper is not GPL'ed, so are they really going to use it?
I just last night installed Red Hat 6.0 on my 300Mhz K6-2 with a G200 AGP. It runs, although I'm sure most of the acceleration features that could be found in a windoze driver doesn't exist in the XF86_SVGA server.
I'm sure "SlashdotMedia" will improve on all the wonders that Dice Holdings blessed us all with
I'm just waiting for full TB Montego support. I wish they would speed up development on PnP sound cards and such....oh well, who needs sound anyway, right?
I'll venture out on a limb and say you configured the kernel slightly different that your previous one. Perhaps some certain code was compiled in the kernel, rather than a module, or perhaps you omitted an unused feature this time, who knows. I've done that a few times, and I was like, "Wow, whatever I did differently this time seemed to work pretty darn good". I dunno. I haven't looked at the 2.3.x stuff yet, but from what I heard, it should be a (close-to) verbatim copy of 2.2.8/9.
Really? I've tried their drivers, but on mine, the playback of audio files are really really really fast. Take the pronunciation (sp?) of Linux by Linus (english.au). Doing: /dev/audio
cat english.au >
Sounds like Alvin the Chipmunk on speed. Perhaps they have a new driver out? I downloaded mine only about a week ago (the eval version). I'm running Redhat 5.2 w/ kernel 2.2.7 on a P3-500. I've got the original TB Montego (not the Montego II).
nothing wrong with a bit of 'healthy' competition
The reason Rob has worked hard is to keep people like you happy, by being able to filter out stuff such as this. Now I know this won't help those wanting linux, but not kernels update, but as I saw someone else say, these updates will slow quite a bit in a couple months. Relax. I'm sure you will be able to find happiness in life. What rob does w/ slashdot won't deprive you.
http://www.somethingpositive.net Funny + bitter = comedy gold
I think it would be nice to have the kernels up on CVS, but I don't think they should break the style of the current releases, ie linus puts them out when neccesary. Really, the advantage would just be for ppl who really dig or for some odd reason, only have access to a CVS tree.
But ya know what? I'd bet cash that someone already has this setup for those who really need it. I doubt there'd be demand enough from everyone else.
http://www.somethingpositive.net Funny + bitter = comedy gold
wouldn't the next stable version after 2.2.x be 3.0?
What is everyone complaining about? I think this simply has to do with the fact that the 2.2 series added a whole slew of stuff that was in the dev series. This means that now all of the sudden a bunch more people are using all of this previously devel code, and most likely finding bugs and other issues.
I would much rather have 10 kernel updates come out in 3 1/2 months, then say three come out in 10 months. If you don't need the new stuff or don't like screwing around with rebuilding your kernel, then don't worry about the new release unless you know you need it. Enough said.
I think I would be in heaven if only I could find a Linux job.
I doubt this would be the case, unless the kernel folks didn't read
Mindcraft's publication on the matter.. (which is also doubtful, as
you'd think that if they were releasing the patches for it, they'd at
least have found out what the nitty gritty details..) (hmm, nice
circular piece of logic on my part there, but I hope you catch my
drift...)
Specifically, the Mindcraft paper stated that the Linux team would be
unable to use any patches not publicly available before April 20.
Is it that hard for you people to look at a release?You don't have to get it, say if you had a certain type of cdrom and the last kernel had a bug which wouldn't let it be mounted and the new one fixed it, I'm sure you'd be appreciative if you want a system that doesn't fix there mistakes goto windows
cd /lib/modules; rm default; ln -s 2.2.9 default
If I remember right, I've seen parts of Redhat's init scripts that look like they'd be related to fixing this symlink, but I've never really looked into it.
Shall I apply the 2.2.9 patch and follow the stable path, or wait for 2.3.1 and lead an exciting but dangerous life?
It'll be 2.2.10, just like 2.3.x might be leading to 3.0.0 instead of 2.4.0.
2 = Major version
2 = Minor version
8 = Micro version
At least where I grew up.
I don't understand it. There are a lot of new kernel releases coming out lately, one might say that they are being released one after another in rapid-fire succession, however, wouldn't it be much easier to use CVS (GNU), perhaps along with CVSup (which is essentially 'cvs update' on steroids - including an optional X UI and protocol compression) to keep its users up-to-date on the kernel sources? I believe I understand the Linus-final-word structure, but a one-man CVS operation would still be worth it if one can download only the latest updated sources from the tree.
This is, as some of you may know, the way FreeBSD keeps its whole OS tree (since it is, after all, one integrated operating system), as well as splitting the source tree (kernel and userland) into CURRENT, STABLE, and RELEASE (essentially a frozen -STABLE from a specific time), and I think it would be very valuble for Linux to do the same thing.
Comments? Flames?
Interesting note: FreeBSD mirrors use CVSup to update a CVS tree, so that users may use those mirrors which have mirrored the entire CVS tree on their system, to download either just the latest sources, or to download the actual CVS tree, on which one can make the usual CVS operations (useful for extensive work on things, such as the FAQ or Handbook, which I happen to work on)
Some people have had problems with Oracle on Linux after upgrading to 2.2.9. Matthew (mattshouse.com, an Oracle / Linux) emailed Alan Cox with some details, and received this response, which seems germain to this discussion ...
(Matthew wrote)
>I just received this e-mail from Alan Cox:
>>2.2.8 has an fs deadlock and an exploitable remote network crash problem.
>>2.2.9 will be appearing rather shortly
(end quoted email)
Oracle users should definitely avoid 2.2.8, and it would seem that the problems Alan notes could affect non-Oracle things as well, hmmm?
Best Regards,
BillyG.