Slashdot Mirror
Linux Kernel v2.6.23 Released
diegocgteleline.es writes "After 3 months, Linus has released Linux 2.6.23. This version includes the new and shiny CFS process scheduler, a simpler read-ahead mechanism, the lguest 'Linux-on-Linux' paravirtualization hypervisor, XEN guest support, KVM smp guest support, and variable process argument length. SLUB is now the default slab allocator, there's SELinux protection for exploiting null dereferences using mmap, XFS and ext4 improvements, PPP over L2TP support. Also the 'lumpy' reclaim algorithm, a userspace driver framework, the O_CLOEXEC file descriptor flag, splice improvements, a new fallocate() syscall, lock statistics, support for multiqueue network devices, various new drivers, and many other minor features and fixes. See the changelog for details."
overlord. welcome. yay.
On a more serious note, are these improvements dramatic, or is story featured just because it's the newest Lolnus kernel?
Unfortunately, Netcraft could not be reached for comment.
I'm so excited, I wish I could have stayed up until midnight in a huge line for it! HOW AM I SUPPOSED TO SLEEP NOW?!
Google's Super Secret Search Algorithm: SELECT @search_results FROM internet WHERE @search_results = 'good'
yeah, but does it run... Oh. Nevermind.
I think I'll take the opportunity to upgrade to 2.2.26; I don't waste my time with unproven technology.
After 3 months, Linus has released Linux 2.6.23. This version includes the new and shiny CFS process scheduler, a simpler read-ahead mechanism, the lguest 'Linux-on-Linux' paravirtualization hypervisor, XEN guest support
Yes, what they don't mention is that the XEN "guest support" is in the form of a crowbar.
The theory of relativity doesn't work right in Arkansas.
I RTFA and it didn't mention whether or not it was released under GPL v2 or v3. Does anyone know?
Using openSUSE instead of Windows since 9th of October, 2007 and liking it.
phew. I few more unstable linus kernel uot there eh!
Linux will never be GPL3. Got that? NEVER!
So has anyone done any "real" benchmarks yet? Hmm? Hmm? What would the robot do!
Obligatory blog plug: http://www.caseybanner.ca/
Unless you're actively building Linux from scratch, this type of news is like a Tennessee strip club. A lot of flashy lights and naked chicks but no alcohol.
Won't someone think of the ISOs?
thinkpad-acpi: enable more hotkeys, add input device support to hotkey subdriver
Woot!
Money is the root of all evil?
...the extra flavor that makes this release a little bit more headline-worthy than usual is probably the whole controversy involving the Completely Fair Scheduler. Between Con Kolivas leaving kernel development, the Really Fair Scheduler flamewar and almost ten release candidates, the whole 2.6.23 development was some kind of geek soap opera.
I'm sure all of this stuff is really cool. Slashdot is a technical community so my comment may not be well received. I've played with Linux a lot and would like to say, it never seems to be about the user experience. Usability should be a top concern for Linux to increase it penetration in the mainstream market. I know there are distributions like Ubuntu which are making that a reality by leaps and bounds. But graphic UI's are the future of computing and I think it's high time for a distribution to make it HARD to find the shell in an OS. Let the Linux community do what Apple (NeXT) did for Unix (I'm preparing to be grilled for this comment), at the end of the day all most users care about is getting their work done. Please Linux developers, unify the OS and create something that at least 90% of the computing population can accomplish something on, not just the brainy and overwhelmingly patient. Thanks for reading my opinions. Dylan
first?
But does it run OS X?
Yes, this is a good thing. However, they seem to have missed some: sockets and pipes. Sockets are not close-on-exec by default, so you may pass a sensitive socket to a child.
Windows NT has the same problem: sockets are inheritable by default until you call SetHandleInformation to disable inheritance. Other handles' inheritability is selected at open/create time.
Luckily, there is a workaround for it, if not pretty: use a reader/writer lock with opening handles as writers and forks as readers.
By the way, the linked changelog on kernelnewbies.org has a bad link for the "recommended LWN article".
For the SELinux thing against null pointer attacks, won't that break DOSemu?
"Screw Sun, cross-platform will never work. Let's move on and steal the Java language." - Visual J++ Product Manager
fallocate() is a new system call which will allow applications to preallocate space to any file(s) in a file system. Applications can get a guarantee of space for particular file(s) - even if later the system becomes full
I was about to go and make fun of Linux for creating a feature that's been around in Windows for quite a while - take your pick of SetFilePointer or sparse files. Yes, yes, I understand that reserving space for a file is not the same as growing it and not using that space. Twas meant to be a troll....But, it turns out that a bit of googling reveals that sparse files under Windows are not all that they are cracked up to be:
http://www.flexhex.com/docs/articles/sparse-files.phtml
This is my sig.
I thought XEN was the guest host when Leno goes on vaction
Great!
In other news, RIP linux for the desktop. With the loss of the two main developers for Thunderbird, it looks like linux is going on without a single decent desktop (non-terminal) email app. Kmail, the one time I tried it, promptly erased all my email folders, and in any case is not very flexible. Thunderbird had plenty of faults, but it was the best one of the crop. Now, without a single great email app, how can an IT manager decide to go with linux over the desktop? What would be the advantage to Mac?
I wish Linus could give an edict and convert some kernel developers to app developers and to app integration, but it doesn't work that way, technical people will always enjoy more working on the kernel than on apps, and linux over the desktop will always be poor.
I am now switching all my group (10 people?) from linux to Mac, and I have only good things to say about it.
Pick out any of the top 5-10 distros at distrowatch and chances are unless you have some exotic hardware/extremely brand new and bleeding edge, or special needs, you won't have to touch the command line to use it. This level of functionality has existed for at least a couple of years now with the major distros, and with applications at your fingertips, it is light years beyond what redmond offers. There is really no comparison what you get out of the box with any major linux distro once you see how much variety and functionality you get and any windows OS, even the "professional" vista stuff. There's just not. Some peripherals, etc obviously function better on windows from driver issues, but just a modicum of homework and you can build a hardware system with some distro that will "just work" for the most part to the level of which you are looking for (most likely, have to guess for your needs here), and never have to touch the command line at all. And with live CDs you can try before full installation, it is ridiculously easy to test them and see what might work on your system you have right now. And the price is right, download and burn for free, or send in a few bucks to one of the clone shippers. I'm not a dev or programmer, and I use linux exclusively, from the GUI all the time now, because I have found there's no real reason to use the CLI, the gui tools are plenty good enough now. I'm just a computer user, I like other geeky stuff,but not into programming at all. I know just enough BASH to know I like running the mouse better, I can navigate any random gui menu tree a lot faster than I can memorize some arcane commands where one single missed or wrong keystroke can bork your reality. If linux wasn't good enough or easy enough, I wouldn't use it. Some years ago it wasn't, but now it mostly is.
...Netcraft confirms it.
Just take OpenBSD and re-release it under the GPLv3!
You would have problem with getting a fix for your issue with any software company. First, you don't know what happened. Second, you cannot give steps to reproduce the problem. So, the developers are left with,
"whaa whaaa whaa.!!!! Software broke! Erased stuff! Fix it!!! Fix it or I switch!"
Good luck with the switch. Be this commercial or free software, you are likely to get the same type of support if you are unwilling to help with the debugging of your problem.
Posts like this make me feel like I'm not a nerd. Just like going to a political rally in Berkeley makes me feel like a centrist.
I just upgraded, and I'm already seeing a 122% increase in throughput for my entire server farm.
i was looking for some masturbation material, and now i have found it!
To create a sparse file in Linux, you open a file, and seek to some arbitrary size.
fallocate, is something else.
What exactly does this have to do with Windows 2000.
Windows 2000 is no longer sold. The latest version of Windows 2000 is ReactOS.
An exploit with feature-complete proof of concept was released for x86_64 linux kernel ia32syscall emulation by cliph at isec in Poland. Exploit code was wildly popular on milw0rm, indicating that this local exploit has lots of potential.
http://tinyurl.com/4ny52
the lguest 'Linux-on-Linux' paravirtualization hypervisor
Linux on linux, that's so hot!
Is anyone else noticing the almost exponential rise in the rate at which new features are being added to the kernel? Linux major release anouncements would dwarf similar anouncements by 'competing' operating systems.
I don't think it can be entirely attributed to the linux kernel merely catching-up with other operating systems.
oh well, is this the year of desktop linux yet?
"The Most Fun Possible on 4 wheels" is at SunBuggy in Las Vegas
Telling readers that links are meant to be clicked is so 1995!
So how many Microsoft patents does the new version infringe upon? It wouldn't be worth doing a release if it doesn't infringe upon at least an extra patent or two.
Not to nitpick, but the milw0rm main page says '2007-09-27' beside that exploit. I'd hardly call that today's Linux news.
N/T
Patents Drive Free Software as Hurricanes Drive Construction Industry
[This is the part where I say something enough on-topic to justify the stupid joke in the subject line. Use your imagination. Thanks.]
This version includes the new and shiny CFS process scheduler,
I know what I'm doing!
a simpler read-ahead mechanism,
And I knew it before you did!
the lguest 'Linux-on-Linux' paravirtualization hypervisor,
LeVar Burton would not approve of that kind of self-abuse!
XEN guest support,
Crabs are NOT invited to my head!
KVM smp guest support,
The sump reverses the M^V>K process?
and variable process argument length.
Oh great, now you'll never get them to stop bickering.
SLUB is now the default slab allocator,
Your build has now lost the battle of the bulge.
there's SELinux protection for exploiting null dereferences using mmap,
If you're going in those kind of holes, you better be wearing something slinky!
XFS and ext4 improvements, PPP over L2TP support.
No no, you put the TP over the PP 2 L8!
Also the 'lumpy' reclaim algorithm,
Mike Rowe? Have I got a job for you...
a userspace driver framework,
Better make it a rollcage.
the O_CLOEXEC file descriptor flag,
I thought only birds have those.
splice improvements, a new fallocate() syscall,
Your Linux will do anything for love but they WON'T do THAT!
lock statistics, support for multiqueue network devices, various new drivers, and many other minor features and fixes. See the changelog for details.
Whew. Time for my cigarette.
1000000000000000000000000000th time I've seen this comment. Every time some expectant little shit like you comes along and posts this, I care slightly less about devoting effort to attracting more of your kind.
You dickheads have Ubuntu, leave the rest of us the fuck alone.
New kernel releases are only for people with big smelly dual-cores these days. Us PIII users have to set aside entire days if we want to compile a new kernel.
Need to learn about cross-compilation, I guess.
Those new quad core AMDs are hot off the presses. Has anyone done any testing with this in the kernel to see if it works at all?
And dare I ask if anyone's tried the per-core voltage/speed throttling?
How about CPU errata bugfixes?
Going over the list, there is little there I wanted and nothing I needed. I think the Linux kernel is getting more and more bloated.
Also added in this release is the new b43 driver. I've been waiting for a long time for this, since my BCM4318 doesn't work with the bcm43xx driver. Unfortunately, unless the driver can be backported to .22, I'm stuck with ndiswrapper (I'm on CentOS, and the Fedora 2.6.23 package wants a bunch of updated core packages like nash).
"Beware of he who would deny you access to information, for in his heart he dreams himself your master."
...it IS missing quite a few things:
1. Games. Other than Urban Terror, pretty much any game I have tried in Linux is crap, or runs kind of crappily under wine. There are some emulators, so you can have decent games, but most of them don't work very well out of box, and are ludicrously difficult to set up. Plus: Until Nintendo figures out a smart/easy way for users to license content, emulation is inevitably going to be a piracy only option. Also, joystick support under wine is pretty awful (though getting better), and I have yet to figure out how to do
2. 3d audio. I'm assuming there's a way to do it. Good luck figuring it out, oh, and this isn't Linux's fault, but many Creative cards don't work at all because Creative are douches about making
3. drivers. There are plenty of missing drivers out there, and again this is not the fault of Linux, but it is an unfortunate part of the reality of where we are today. M$ has bullied many manufacturers into NOT supporting Linux, and the market has not brow beat them into sense about it (yet).
4. Ease of Use. I'm not saying Ubuntu hasn't made strides, it has, the new version even more so, but there are still plenty of common tasks that are very difficult and shouldn't be.
5. Videoconferencing. I've read that it's possible in ekiga, or by running VLC in some kind of special mode, but I have yet to be able to do it AT ALL, and I have thrown several hours away on it.
6. Recording Video. You know, like off of a TV card. Again, I've heard it was possible, and even read about MythTV and Mythbuntu, and others. If you can get it to work, kudos, you must be a fucking genius. I've tried like 8 approaches with zero success.
7. Microsoft Office. Linux does have the superior Open Office, but a LOT of students and professionals require M$ office, because they currently enjoy
8. a hegemony. Linux is going to have to be better, faster, easier, and backwards compatible before "The Year of Desktop Linux" happens. But this is a problem that feeds on itself since MS has a hegemony.
Linux is KICKING ASS right now, but it is still losing the war, and I'm nervous that it may not be ABLE to win, given some of the problems I've run into using Ubuntu in particular, and dealing with the Ubuntu community specifically. There is no bar for entry into the FOSS community, and it would not surprise me if M$ and the like were inserting enemy combatants left and right. I have to assume they are, given the ludicrously easy things that Linux as an OS (and yes I know the parent article is about the KERNEL, and this is all OT) fails to do.
I really hope the problems can be addressed, but I'm skeptical if Linux will ever break out of the server farms based on most of my recent experience, especially in the game department!
rhY
I hold very few opinions. I hold information based on observation and fact. If you wish to disagree, please use facts.
Installing a Linux distro on compatible hardware is fairly easy.
Keeping it running gets interesting.
I am not a Linux guru, but I've been writing how-to articles on Linux for the last 3 years.
I have a fairly standard sort of setup, a Biostar GeForce 6100 AM2 integrated motherboard with Nvidia chipset and Athlon 64x2/4200 and 2G DDR2.
The normal procedure for installing a new nvidia video driver is:
# aptitude remove nvidia
# aptitude install nvidia[version compatible with kernel version]
Easy enough.
I had to do extensive research to find workarounds that would permit me to install the nvidia driver on the last three kernel upgrades
Last time around, I found out that the new kernel upgrade was compiled on a different gcc version than the version of gcc which had been pushed out via automatic update about a month before.
Before that, I found out based on a web search on the error message that the kernel developers decided to make a kernel call relating to paravirtualization unavailable to non-GPL proprietary drivers, some digging found me a patched kernel with the fix.
Would you like to talk a MCSE or your grandmother through what I just described?
I don't take the assertion that "Linux is ready for the masses" seriously yet and neither should anyone else. This delusion is bad for the Linux community as a whole, as it reduces the pressure on developers to fix the remaining problems.
Getting there? Certainly. I'd be far more surprised than not if Linux is to the point where a member of the general public can use it without having a Linux guru available to provide hands-on help by this time next year. But that time is not now. Do you want 20 or 30 million people running into trouble they can't handle, reformatting their boxes for XP, and telling their friends that Linux is shit? I certainly don't.
Tech Public Policy stuff
And I just finished compiling 2.6.22.9 :( Fuck!
Tried compiling this on a HP DL360 G5 - Debian Etch today, seems to be an issue with the make modules not actually making the modules. Have found a web reference for the same issue with v2.6.22. Ended up backtracking kernel versions to something a little more stable.
Upgrade to 2.6.23 right now? Are you out of your mind? Everyone knows you're supposed to wait for the SP1 release before upgrading to a new operating system!
More code, more bells and whistles, more bugs in the kernel. Thanks Linus!
The new kernel also includes ALSA support for the Dreamcast sound device (for the first time - an out of mainline OSS driver did/does exist for 2.4).
More dreamcast support is on the way - expect some more stuff in 2.6.24 and 2.6.25 and I (the author of the code) would love to hear from willing testers, etc
Pirate: My witty statements always become memes
Guybrush Threepwood: Oh really? To me that just sound like clichés.
Alternative reply: Too bad no one's ever heard of YOU at all.
Bad reply: I am rubber, you are glue.
In other words, there is now a key component of Xen in mainline, but still nothing even remotely close to a usably complete Xen system.
I don't get this at all. Xen was the very first virtualization system for Linux, it's been around for years, and has always been open source. So why isn't it fully merged into the kernel yet?
New virtualization systems are appearing at a rate of knots. KVM and Iguest didn't seem to get held up in being merged, so what's up with Xen?
vendors of video display cards? Yes, that would be nvidia and ATI. So it's reasonable to expect support. Yes, I know the legal / technical / political issues involved, it's hardly possible to write for money about Linux without being aware of them. Would "the masses" know to research whether or not a video chipset is supported? More to the point, I verified that there was Linux support for nvidia before buying the motherboard.
Debian packagers do supply packages for nvidia and rebuilt as debs... as you might have noticed from my post that said the usual method for upgrading nvidia is # aptitude install nvidia.
The problem here is that the nvidia packages weren't fully maintained up to current version and there were problems with them besides that. All a Debian end user is responsible for is to check to see if Debian maintains driver packages compatible with the product one is considering buying. An end user isn't supposed to have to cope with situations where the compiler that's current with a distro (gcc 4.2) that one uses to build a nvidia binary (as I said, the Debian nvidia wasn't ready) is not the same as the one the kernel was built in (gcc 4.1). One of the points behind automated upgrades is to prevent that situation from happening.
If the driver packages exist in Debian repositories, that's "supported" as far as a Debian user is concerned. It's the developer/packager problem to make sure that they work on the chipsets they are claimed to work with.
Part of the Debian learning curve is finding out that non-free and contrib sections of the repositories have to be added to sources.list so one can get access to proprietary drivers. I think they deal with this better in Ubuntu.
Tech Public Policy stuff
Karma whore who misstates a week old (and already fixed) flaw as beeing new. Step up against such karma whoring and mod -1 troll. Thank you.
"paravirtualization hypervisor" is going to be my new tech-babble phrase for a while :p
I think the Linux kernel is getting more and more bloated.
I'm starting to think so too. One of the things that made linux popular was that it ran like the wind on rescued hardware, and dumpster salvage. While the response of FC7 on a 500MHz machine with 128M of RAM is quite snappy in a geological sense, on a human scale it runs like a drunk pig. There's so much stuff running in the background, it's hard to know what's needed and what's not, or even what the hardware supports. Can a PentiumIII even handle paravirtualization hypervising?
When our name is on the back of your car, we're behind you all the way!
yes, with default policy the only thing allowed to override this at the moment is X which means other users of vm page 0 will have trouble. Custom SELinux policy would have to be built for DOSemu (or disable the protection in /proc/sys/vm/mmap_min_addr)
I'm sure you'd find plenty of people on the selinux mailing lists willing to help if you run into trouble!
Where can I find more information on the differences between the old and new process scheduler?
Yep, yet another new version. It's really a shame how they couldn't be bothered to code it right the first time around.
As little as I like the "there-is-no-stable" mentality of the current kernel development methodology, it's quite clear that it has enabled the kernel devs to incorporate new features into the mainline, and get them tested by a very large user base, in an exceedingly short period of time. Of course, the result is a decline in stability, and as a result, many still pine for the days of the old stable/unstable split. But, it seems the tradeoff is considered worth it by most.
I'm also willing to bet the move to a more distributed source control system has had a large effect. While I'm not convinced of it's utility for corporate development, a distributed source control system, and it's effect on development, seems to have been a very positive change for the kernel developers, and has greatly reduced the burden on Linus, as it makes it very natural for individual subsystem maintainers to groom, test, and incorporate downstream patches before sending the final mods upstream to Linus for inclusion into the kernel proper.
That's the step where it goes wrong. Person B is still bound by the terms of the BSD license.
Let's simplify it a bit, and assume just one file. Person A obtains a file consisting solely of one person's work, licensed under the BSD terms. Person A makes significant changes to it, and releases a derived work to Person B, under the GPL license. Person B now needs two licenses to modify and distribute what Person A gave them:
The key point is that Person A can only enforce copyright on his GPL'ed work to the extent that the alleged infringing material is not based on the original, BSD-licensed file. For example, if B takes a function from A's GPLed file and puts it into a non-GPL program, A can only complain if that function didn't appear in the original BSD-licensed work.
Are you adequate?
A good Jedi never has enough time to rewrite a memory deallocation process.
An excelent Jedi, leaves it for the hackers.
?
the kernel I got was compiled with gcc 4.1 and the gcc version that's current and pushed out via automated upgrade is 4.2? Remember, the team responsible for distributing updated Debian-packaged kernels is also the one who decide what compiler goes into the current distro upgrade. I think this problem is going to be common to any driver that has to be compiled into the kernel, I'm merely glad that for me, nvidia is the only problem.
.deb Debian repackage is also supposed to be up to date with the current kernel for the same reasons.
More to the point, the nvidia
Tech Public Policy stuff
the problem is that the kernel will only tolerate the insertion of a module compiled on the same compiler version as the kernel.
Tech Public Policy stuff
I think the mod might have had a *WHOOSH* moment.
You don't change the license terms of the code you're based upon, you release a new work under a new license. The original code remains under its original license, you can't take that away.
How could I change terms? A license, roughly speaking, is a conditional promise not to sue. I can't take away the original promise, because it was not me who has made it. Nor can I sue you, because the copyright is not mine. But if I create a derivative work based on some BSD-licensed code, I can add my own promise not to sue you over use of the derivative, with different conditions. The original code that is a part of the derivative continues to be licensed under its original BSD license. (There's nothing wrong with a whole and a part having different copyright holders and a different licenses.)