Slashdot Mirror

• Released October 27, 2005 changelog

• Numa-aware slab allocator: It creates slabs on multiple nodes and manages slabs in such a way that locality of allocations is optimized. Each node has its own list of partial, free and full slabs. All object allocations for a node occur from node specific slab lists (commit - benchmarks)

• Lazy page table copies in fork() for VMAs without anonymous pages (the ones with anonymous pages are still copied): Defer copying of ptes until fault time when it is possible to reconstruct the pte from backing store, speeding up fork() greatly specially for processes using lots of shared memory (commit)

• Add /proc/$PID/smaps: This file will shows how much memory is resident in each mapping. Useful for people who want to perform memory consumption analysis (commit)

• Add /proc/$PID/numa_maps: This file will show on which nodes pages reside (commit)

• Lock-free file descriptor look-up (commit) - (commit)

• Four-level page table support for the ppc64 architecture: extends the usable user address range to 44 bits (16T). (commit)

• Support hotplug cpu on 32-bit SMP powermacs: When a cpu is off-lined, it is put into sleep mode with interrupts disabled. It can be on-lined again by asserting its soft-reset pin, which is connected to a GPIO pin (commit)

• Add TASK_NONINTERACTIVE task state bit to the cpu scheduler: It can be used by blocking points to mark the task's wait as "non-interactive". This does not mean the task will be considered a CPU-hog - the wait will simply not have an effect on the waiting task's priority - positive or negative alike (commit)

• PPTP (Point-to-Point Tunneling Protocol) support: RFC 2637. Used to implement VPN products (notably, Microsoft in all the Windows versions). Wikipedia article (commit)

• DCCP: "Datagram Congestion Control Protocol". Datagram protocol (like UDP), but with a congestion control mechanism. (LWN article) Currently a RFC draft (commit)

• Implement SKB fast cloning: Protocols that make extensive use of SKB cloning, for example TCP, eat at least 2 allocations per packet sent as a result. To

• Released October 27, 2005 changelog

• Numa-aware slab allocator: It creates slabs on multiple nodes and manages slabs in such a way that locality of allocations is optimized. Each node has its own list of partial, free and full slabs. All object allocations for a node occur from node specific slab lists (commit - benchmarks)

• Lazy page table copies in fork() for VMAs without anonymous pages (the ones with anonymous pages are still copied): Defer copying of ptes until fault time when it is possible to reconstruct the pte from backing store, speeding up fork() greatly specially for processes using lots of shared memory (commit)

• Add /proc/$PID/smaps: This file will shows how much memory is resident in each mapping. Useful for people who want to perform memory consumption analysis (commit)

• Add /proc/$PID/numa_maps: This file will show on which nodes pages reside (commit)

• Lock-free file descriptor look-up (commit) - (commit)

• Four-level page table support for the ppc64 architecture: extends the usable user address range to 44 bits (16T). (commit)

• Support hotplug cpu on 32-bit SMP powermacs: When a cpu is off-lined, it is put into sleep mode with interrupts disabled. It can be on-lined again by asserting its soft-reset pin, which is connected to a GPIO pin (commit)

• Add TASK_NONINTERACTIVE task state bit to the cpu scheduler: It can be used by blocking points to mark the task's wait as "non-interactive". This does not mean the task will be considered a CPU-hog - the wait will simply not have an effect on the waiting task's priority - positive or negative alike (commit)

• PPTP (Point-to-Point Tunneling Protocol) support: RFC 2637. Used to implement VPN products (notably, Microsoft in all the Windows versions). Wikipedia article (commit)

• DCCP: "Datagram Congestion Control Protocol". Datagram protocol (like UDP), but with a congestion control mechanism. (LWN article) Currently a RFC draft (commit)

• Implement SKB fast cloning: Protocols that make extensive use of SKB cloning, for example TCP, eat at least 2 allocations per packet sent as a result. To

• Released October 27, 2005 changelog

• Numa-aware slab allocator: It creates slabs on multiple nodes and manages slabs in such a way that locality of allocations is optimized. Each node has its own list of partial, free and full slabs. All object allocations for a node occur from node specific slab lists (commit - benchmarks)

• Lazy page table copies in fork() for VMAs without anonymous pages (the ones with anonymous pages are still copied): Defer copying of ptes until fault time when it is possible to reconstruct the pte from backing store, speeding up fork() greatly specially for processes using lots of shared memory (commit)

• Add /proc/$PID/smaps: This file will shows how much memory is resident in each mapping. Useful for people who want to perform memory consumption analysis (commit)

• Add /proc/$PID/numa_maps: This file will show on which nodes pages reside (commit)

• Lock-free file descriptor look-up (commit) - (commit)

• Four-level page table support for the ppc64 architecture: extends the usable user address range to 44 bits (16T). (commit)

• Support hotplug cpu on 32-bit SMP powermacs: When a cpu is off-lined, it is put into sleep mode with interrupts disabled. It can be on-lined again by asserting its soft-reset pin, which is connected to a GPIO pin (commit)

• Add TASK_NONINTERACTIVE task state bit to the cpu scheduler: It can be used by blocking points to mark the task's wait as "non-interactive". This does not mean the task will be considered a CPU-hog - the wait will simply not have an effect on the waiting task's priority - positive or negative alike (commit)

• PPTP (Point-to-Point Tunneling Protocol) support: RFC 2637. Used to implement VPN products (notably, Microsoft in all the Windows versions). Wikipedia article (commit)

• DCCP: "Datagram Congestion Control Protocol". Datagram protocol (like UDP), but with a congestion control mechanism. (LWN article) Currently a RFC draft (commit)

• Implement SKB fast cloning: Protocols that make extensive use of SKB cloning, for example TCP, eat at least 2 allocations per packet sent as a result. To

• Released October 27, 2005 changelog

• Numa-aware slab allocator: It creates slabs on multiple nodes and manages slabs in such a way that locality of allocations is optimized. Each node has its own list of partial, free and full slabs. All object allocations for a node occur from node specific slab lists (commit - benchmarks)

• Lazy page table copies in fork() for VMAs without anonymous pages (the ones with anonymous pages are still copied): Defer copying of ptes until fault time when it is possible to reconstruct the pte from backing store, speeding up fork() greatly specially for processes using lots of shared memory (commit)

• Add /proc/$PID/smaps: This file will shows how much memory is resident in each mapping. Useful for people who want to perform memory consumption analysis (commit)

• Add /proc/$PID/numa_maps: This file will show on which nodes pages reside (commit)

• Lock-free file descriptor look-up (commit) - (commit)

• Four-level page table support for the ppc64 architecture: extends the usable user address range to 44 bits (16T). (commit)

• Support hotplug cpu on 32-bit SMP powermacs: When a cpu is off-lined, it is put into sleep mode with interrupts disabled. It can be on-lined again by asserting its soft-reset pin, which is connected to a GPIO pin (commit)

• Add TASK_NONINTERACTIVE task state bit to the cpu scheduler: It can be used by blocking points to mark the task's wait as "non-interactive". This does not mean the task will be considered a CPU-hog - the wait will simply not have an effect on the waiting task's priority - positive or negative alike (commit)

• PPTP (Point-to-Point Tunneling Protocol) support: RFC 2637. Used to implement VPN products (notably, Microsoft in all the Windows versions). Wikipedia article (commit)

• DCCP: "Datagram Congestion Control Protocol". Datagram protocol (like UDP), but with a congestion control mechanism. (LWN article) Currently a RFC draft (commit)

• Implement SKB fast cloning: Protocols that make extensive use of SKB cloning, for example TCP, eat at least 2 allocations per packet sent as a result. To

• Released October 27, 2005 changelog

• Numa-aware slab allocator: It creates slabs on multiple nodes and manages slabs in such a way that locality of allocations is optimized. Each node has its own list of partial, free and full slabs. All object allocations for a node occur from node specific slab lists (commit - benchmarks)

• Lazy page table copies in fork() for VMAs without anonymous pages (the ones with anonymous pages are still copied): Defer copying of ptes until fault time when it is possible to reconstruct the pte from backing store, speeding up fork() greatly specially for processes using lots of shared memory (commit)

• Add /proc/$PID/smaps: This file will shows how much memory is resident in each mapping. Useful for people who want to perform memory consumption analysis (commit)

• Add /proc/$PID/numa_maps: This file will show on which nodes pages reside (commit)

• Lock-free file descriptor look-up (commit) - (commit)

• Four-level page table support for the ppc64 architecture: extends the usable user address range to 44 bits (16T). (commit)

• Support hotplug cpu on 32-bit SMP powermacs: When a cpu is off-lined, it is put into sleep mode with interrupts disabled. It can be on-lined again by asserting its soft-reset pin, which is connected to a GPIO pin (commit)

• Add TASK_NONINTERACTIVE task state bit to the cpu scheduler: It can be used by blocking points to mark the task's wait as "non-interactive". This does not mean the task will be considered a CPU-hog - the wait will simply not have an effect on the waiting task's priority - positive or negative alike (commit)

• PPTP (Point-to-Point Tunneling Protocol) support: RFC 2637. Used to implement VPN products (notably, Microsoft in all the Windows versions). Wikipedia article (commit)

• DCCP: "Datagram Congestion Control Protocol". Datagram protocol (like UDP), but with a congestion control mechanism. (LWN article) Currently a RFC draft (commit)

• Implement SKB fast cloning: Protocols that make extensive use of SKB cloning, for example TCP, eat at least 2 allocations per packet sent as a result. To

• Released October 27, 2005 changelog

• Numa-aware slab allocator: It creates slabs on multiple nodes and manages slabs in such a way that locality of allocations is optimized. Each node has its own list of partial, free and full slabs. All object allocations for a node occur from node specific slab lists (commit - benchmarks)

• Lazy page table copies in fork() for VMAs without anonymous pages (the ones with anonymous pages are still copied): Defer copying of ptes until fault time when it is possible to reconstruct the pte from backing store, speeding up fork() greatly specially for processes using lots of shared memory (commit)

• Add /proc/$PID/smaps: This file will shows how much memory is resident in each mapping. Useful for people who want to perform memory consumption analysis (commit)

• Add /proc/$PID/numa_maps: This file will show on which nodes pages reside (commit)

• Lock-free file descriptor look-up (commit) - (commit)

• Four-level page table support for the ppc64 architecture: extends the usable user address range to 44 bits (16T). (commit)

• Support hotplug cpu on 32-bit SMP powermacs: When a cpu is off-lined, it is put into sleep mode with interrupts disabled. It can be on-lined again by asserting its soft-reset pin, which is connected to a GPIO pin (commit)

• Add TASK_NONINTERACTIVE task state bit to the cpu scheduler: It can be used by blocking points to mark the task's wait as "non-interactive". This does not mean the task will be considered a CPU-hog - the wait will simply not have an effect on the waiting task's priority - positive or negative alike (commit)

• PPTP (Point-to-Point Tunneling Protocol) support: RFC 2637. Used to implement VPN products (notably, Microsoft in all the Windows versions). Wikipedia article (commit)

• DCCP: "Datagram Congestion Control Protocol". Datagram protocol (like UDP), but with a congestion control mechanism. (LWN article) Currently a RFC draft (commit)

• Implement SKB fast cloning: Protocols that make extensive use of SKB cloning, for example TCP, eat at least 2 allocations per packet sent as a result. To

• Released October 27, 2005 changelog

• Numa-aware slab allocator: It creates slabs on multiple nodes and manages slabs in such a way that locality of allocations is optimized. Each node has its own list of partial, free and full slabs. All object allocations for a node occur from node specific slab lists (commit - benchmarks)

• Lazy page table copies in fork() for VMAs without anonymous pages (the ones with anonymous pages are still copied): Defer copying of ptes until fault time when it is possible to reconstruct the pte from backing store, speeding up fork() greatly specially for processes using lots of shared memory (commit)

• Add /proc/$PID/smaps: This file will shows how much memory is resident in each mapping. Useful for people who want to perform memory consumption analysis (commit)

• Add /proc/$PID/numa_maps: This file will show on which nodes pages reside (commit)

• Lock-free file descriptor look-up (commit) - (commit)

• Four-level page table support for the ppc64 architecture: extends the usable user address range to 44 bits (16T). (commit)

• Support hotplug cpu on 32-bit SMP powermacs: When a cpu is off-lined, it is put into sleep mode with interrupts disabled. It can be on-lined again by asserting its soft-reset pin, which is connected to a GPIO pin (commit)

• Add TASK_NONINTERACTIVE task state bit to the cpu scheduler: It can be used by blocking points to mark the task's wait as "non-interactive". This does not mean the task will be considered a CPU-hog - the wait will simply not have an effect on the waiting task's priority - positive or negative alike (commit)

• PPTP (Point-to-Point Tunneling Protocol) support: RFC 2637. Used to implement VPN products (notably, Microsoft in all the Windows versions). Wikipedia article (commit)

• DCCP: "Datagram Congestion Control Protocol". Datagram protocol (like UDP), but with a congestion control mechanism. (LWN article) Currently a RFC draft (commit)

• Implement SKB fast cloning: Protocols that make extensive use of SKB cloning, for example TCP, eat at least 2 allocations per packet sent as a result. To

There's just one problem. This perception of the software-as-services model is a jaundiced misrepresentation of the way that on-demand applications actually work. No on-demand customer pays simply for the privilege of accessing the software. They pay because the software delivers business results. And that simple distinction exposes once and for all the clay feet, the emperor's new clothes, of the traditional applications software industry. Their products don't actually work until they've been tweaked and customized by customers or partners, and therefore the licence of itself has no out-of-the-box value to the end user. Asking people to pay for the privilege of using the software isn't offering a service, it's taking a liberty. It's as much of a nonsense as asking a punter to pay a performance fee for whistling a copyrighted tune. If I'm paying a fee to watch a movie, listen to a song, or use an application, I expect to experience a professional, finished execution.

True on-demand application vendors understand this. Conventional software vendors seem to think the world still owes them a living, just for bothering to write some software.


This article sounds as if the guy was jaded from the start. His complaints are similar to those people who first scoffed at the notion of leasing a car instead of buying it. Some may consider it foolish, but some also see the benefits. In my experience you can lease a car for 12 months, have the "owner" of the car (or software) continually maintain it when it needs it.

Don't read too deeply in on that analogy, please.

But BOTHERING to write some software? By us Bothering to write some software you have some of the best software out there that's been used to secure most of the IT infrastructure the world runs on. Apache, The Linux Kernel, The Various BSD's, SQL Databases, Iptables, SNORT IDS software, OpenSSL, and many many more!

This guy is just trolling. The article is slanted because he believes that once written, any bugs, flaws (as in it doesn't do this the _way_ it should for ME) should all be done for free simply because he or general consumers are greedy. To a point, bug fixes should be fixed like glaring security flaws that could be used to take over your computer (ala windows in general, yes I'm biased) or damage your information etc.

But get real. If you paid ONCE for your anti-virus software and expected it to work flawlessly and capture all viruses, worms etc without having to pay extra every year to maintain that reliability you're just out of your mind. There is no incentive to keep something up for free especially in an evolving industry. One that evolves and almost 2-5 times the normal rate of other industries.

Think of it this way. You pay a subscription service similar to that of an anti-virus vendor. Receive continual updates, bug fixes, serious flaws get fixed for an annual price. This ensures the developers can work and continue to live as well. Why not? If you don't pay for the next years license, you simply don't get major version upgrades (maybe a serious bug fix or service pack) or new "features".

I'm not keen on the idea of keeping your apps on a server/central location, unless it's on my home network and I have the option to install it centrally or on each workstation. It's just foolish to do it that way. But this guy's "it's mine, I want it all forever" after a simple purchase doesn't cut it. Want that new fender or tires? They're better quality than the current tires you have, then pay for them. Don't expect it for free buddy.

This guy really pissed me off. And I have a football game to watch.

We've already /.'ed Kerrnel.org?
I think a mirror is at http://kernel.org./

And they're coming to Linux in 2.6.14, the port of the 9P protocol has been included. FUSE (a alternative filesystem-in-userspace approach) has also been included

And they're coming to Linux in 2.6.14, the port of the 9P protocol has been included. FUSE (a alternative filesystem-in-userspace approach) has also been included

Seriously, though - can you imagine how screwed the rest of the world would be without their Internet able to access US sites? Bye-bye, Google, Yahoo! and Lycos. Bye-bye, Ebay. Bye-bye Slashdot. And (perhaps more importantly, now that more and more of Europe is switching to Linux) bye-bye http://www.kernel.org/ (for those of you who don't know it's the Linux kernel website where you download Linus and his team's code). There's probably quite a few other sites I'm forgetting here.

Plus, let's not forget that if the EU built their own Internet, that would mean that they wouldn't be able to access the rest of the world's Internet - unless they connected to our Internet. But that would still require them to register domain names with us, if they wanted the two to be able to communicate with each other. Otherwise I could register netraven5000.com on the US Internet, and someone in Europe could register it on the EU Internet. Which one will people get when they try to go there - if they want my site, they might get his instead, and vice versa.

Highly detailled specs: Specs

FAQ: FAQ

Support: Support

How to reply on Slashdot: Here

I don't understand why people still say that Linux is not user-friendly. Complaints

Your kidding right? The Xeon suffers from a slow FSB meaning the opteron kills it on most performance benchmarks. You need a much higher clocked Xeon to keep up with an Opteron and even then it depends on the benchmark.

Were the uniprocessor benchmarks done on the same hardware? If youve discovered some problems under load then I'd suggest posing them to linux-kernel@vger.kernel.org

OpenBSD, from what I've heard, is good, but most of its security is based upon correct implementation. This is good, but the OpenBSD team can only audit and control the base system, meaning that applications and libraries added to the system can often degrade the security of the system as a whole.

Judging from the technologies and companies mentioned in the summary, this attempt at Linux security is based on providing better access controls and privilege models in the Linux kernel. By better, I mean that these mechanisms can:

1) Provide finer grain privileges so that fewer programs can be exploited to escalate privilege, and
2) Isolate unrelated programs and users from each other (e.g. an exploit in a DNS server is restricted to only accessing DNS files but is not able to manipulate web server pages).

These two techniques basically reduce the number of avenues an attacker can use to exploit a system. It is less likely that a piece of exploitable software will have sufficient access to whatever it is the attacker wants to get to. Granted, it is not a complete solution, but it's a handy thing to have in one's security toolbox.

I believe that the OpenBSD/OpenSSH teams are beginning to do similar things (e.g. OpenSSH privilege separation), but I don't think they've taken the leap to providing more sophisticated access controls in the kernel.

If you're interested, examples of trusted operating systems/access controls can be found at the following places:

Linux Capabilities:
http://ftp.kernel.org/pub/linux/libs/security/linu x-privs/kernel-2.4/capfaq-0.2.txt

Trusted BSD:
http://www.trustedbsd.org/docs.html

Argus Systems Group (go to the Support section and take a look at the docs for PitBull LX and Foundation; they give a rather complete description of the mechanisms):
http://www.argus-systems.com/

Trusted Computer Solutions (mentioned in the article):
http://www.trustedcs.com/index.html

Disclaimer: I used to work for Argus Systems Group, and I know a few of the TCS employees (as they are also ex-Argus employees).

Here in Japan, I have 55 megabit fiber DSL.

That's all? I'm sitting here in Yokohama with an ONU in my kitchen, and wget doing things like this:

--15:16:09-- http://www.jp.kernel.org/pub/linux/kernel/v2.6/lin ux-2.6.13.tar.bz2
[snip]
15:16:13 (8.67 MB/s) - `linux-2.6.13.tar.bz2' saved [38372729/38372729]

It doesn't. The homepage says that 2.6.13.2 is "the latest stable version". The old "2.x.y is stable if x is even, otherwise it's a dev tree" hasn't been correct anymore for quite some time. So, 2.6.13.2 is considered stable, but that does not mean that every 2.6.x.y is considered stable. Plus, Linux now leaves final stabilization to the distros, because that is what people run in production. There's been a lot of discussion going on for months, it's all on kerneltrap.

Well, here it is (odd second#) ftp://ftp.kernel.org/pub/linux/kernel/v1.1/v1.1.0. tar.bz2 for you to feel kewl and believe me, that it was still fresh in 94

It sounds like Windows is thrown together with practically no organization what so ever. Yet all the computer nerds that only code free time can make a highly stable, flexible, and organized OS.

Always pronounced ``"lin-nucks" ("lin" rhymes with "chin")''?

Maybe always only very recently.

I'd never heard it pronounced "lin-nucks" for the first 5 or 6 years I was using it. One evening after hearing Captain Picard say "privacy" with the first syllable sounding exactly like the first syllable in "privilege" I asked a Brit acquaintance how they pronounced the name "Linus". Result: "lin-nus". Coupling that with Linus himself in the two sound files I was perhaps a bit too vague in referencing earlier ( http://www.mx.kernel.org/pub/linux/kernel/SillySou nds/ ), and with experience with people who natively speak various European languages, the whole thing gets rigidly cemented into place.

I may sound like a displaced Kentuckian here in Hoosierland when I say "line-ucks", but I'm consistent; as are arguably most other-language-speakers from around the globe, with saying Linux and Linus using the same vowel sounds.

I'd say Linus as "lin-nus" just to be able to conform with the rest of you would-be Brits (when saying "Linux"), but I feel utterly stupid saying it that way. I'd bet you would/do too.

In terms of vowel sounds, if you had a buddy named Linus, however you'd pronounce his name (wherever in the world you lived) would be the same way you should pronounce Linux.

Here's a thought. Do an informal "man on the street" series of interviews right there in Tennessee. One where you show folks the written word "Linux" and ask them to pronounce it. And I'm not talking about your IT friends; just your run-of-the-mill folk on the street. See how many you'd have to "correct", step back, and re-evaluate your position for consistency.

So...the "BSD tcp/ip stack...and the ftp/telnet apps, and SFU (MKS utils and Interix subsystems)" are "trivial" as well?

How about a preembtible kernel? A full office suite?.

Jeez. I'd love to see some of the stuff you've coded if this stuff is trivial.

Actually both v9fs and FUSE made it into 2.6.14-rc1, also DCCP is in too - can't have too much of a good thing.

No ReiserFS though.

http://www.kernel.org/pub/linux/kernel/v2.6/testin g/ChangeLog-2.6.14-rc1

We (my company) deliver servers to printing houses in europe, asia and even to the US. Okay: We have had some trouble with reiserfs in the early days (this was with Linux 2.2.14).

I have never experienced serious trouble with systems from 2.4.18 and up that was not caused by a hardware failure. Even than: With reiserfsck --rebuildtree volume it was sometimes possible to recover from some of the damage caused by disk read failures.

Of course I've to admit that 80% of our customers run SuSE Linux and the SuSE kernels always differed somehow from the official http://kernel.org/ kernels.

The performance gain of reiserfs compared to ext3 makes a big difference in our application area. I'm looking forward to find some time evaluating reiser4 and I like the work of Hans Reiser.

(soon to be merged in mainline)

Oh well, it's already there and it seems to have been there for ages: http://kernel.org/git/?p=linux/kernel/git/torvalds /linux-2.6.git;a=blob;h=f3ea492ab44dfcfc05e2fad402 7e303f9b69d9dd;hb=caf39e87cc1182f7dae84eefc43ca14d 54c78ef9;f=kernel/kprobes.c

There are two choices for upgrading the kernel on Slackware. You can either find a 2.6 kernel package already built in /testing on your install CD's (I don't remember which disc). This would be the easiest way, but you might also get an older kernel like 2.6.9 instead of the latest 2.6.13. To install it, just use installpkg. It should install the kernel in /boot, and then you can specify it in /etc/lilo.conf. (I'm a bit fuzzy on the details, I haven't used Slackware in a while.)

The other way is to compile your own kernel. Despite what everyone says, it's pretty simple once you've done it once. The hard part is making sure you have all the right drivers in it. To start, go to kernel.org and download the latest full kernel tarball (it will be named something like linux-2.6.12.tar.bz2). Put it somewhere, open a terminal, cd to that directory, and type tar -xvjf linux-2.6.x.tar.bz2 (replacing x with whichever version you get). This will unpack the tarball into a subdirectory named linux-2.6.x . cd into that subdirectory and type make menuconfig. This will bring up a menu of configuration options. It's a bit complicated, I admit, but you should go through it step by step and answer "Y" to all the hardware you have. In particular, you need to check off IDE DISK support (assuming you have an IDE hard drive--most are these days) and check off whichever filesystem you use (probably ext2, ext3, or [my personal favorite] reiserfs). Make sure you have your network card, your graphics card, drivers for any ports (serial, parallel, USB, firewire, etc...) and other devices you use. If you need help, you can ask in #slackware on irc.freenode.net, or email me at thinkinginbinary(at)gmail(dot)com. You should probably go through the config, picking out what you can, and writing down a list of questions to get answered.

Once you've finished configuring the kernel, exit the configuration screen (answer "Yes" to the save prompt). Then, type make -j2. This will build the kernel. (The -j2 tells it to compile two files at once, which can be slightly more efficient.) Once that finishes, type make modules_install and make install. This should install your kernel in /boot and symlink vmlinuz to the new kernel. Make sure you keep a copy of your old kernel in /boot as well, because there is a decent chance that your first kernel will be missing something important, and you should always have a backup one to use.

Reboot, and see if it works. If it does, congratulations. If it doesn't, email me and I'll try to help.

Code signing can be used to determine where the code came from (or, at least, some place that claims to have provided the code). If that's really what you want to know, then it's fine. It's a good mechanism for giving some information to people who click on a program and select "properties".

On the other hand, this information is not useful for deciding whether to trust some code, because the signature only provides proof of origin, not proof of correctness. Anything that only permits signed code to run is fundamentally flawed, because it makes the assumption that there are signers who only sign good code, which is manifestly false.

On the other hand, mere origin is useful in some cases, although, lacking any security implications to forged signatures, there's little need for cryptography, which is why I've signed a bunch of my code. (It's worked reasonably well; people sometimes realize when I'm the right person to debug something, and sometimes don't.)

There is some potential for validating that things haven't been changed using signed code. If the OS knows that you're trying to run Microsoft Word, and it's not signed by Microsoft, it can tell you that. And some OSes could actually retain the public keys with sufficient security to actually catch some problems that way (particularly on applications installed by users). But this is an unusual situation and only protects things that viruses don't generally do these days (modify programs without compromising the system).

1. wget ftp://ftp.kernel.org/pub/linux/kernel/v2.4/linux-2 .6.x.tar.bz2
2. tar jxvf linux-2.6.x.tar.bz2 && cd linux-2.6.x
3. less README
4. cd Documentation && less post-halloween-2.6.txt

1. wget ftp://ftp.kernel.org/pub/linux/kernel/v2.4/linux-2 .6.x.tar.bz2
2. tar jxvf linux-2.6.x.tar.bz2 && cd linux-2.6.x
3. less README
4. cd Documentation && less post-halloween-2.6.txt

2.4 will give you a more stable kernel due to being tested better. I am still sticking with 2.4 on my servers, but just because I'm lazy. :)

I've been running 2.6 on all desktop & laptop systems without problems since 2.6.9 (about a year). I certainly would not want to give up the better interactivity, better MM performance, wide hardware support, and features like udev.

2.6.x will have have hickups now and again because that is where the development occurs. That's why a few kernel hackers (Chris Wright & Greg Kroah-Hartman) started maintaining stable point releases (2.6.12.1, 2.6.12.2, etc) to deal with bug fixes only. If you are concerned about stability on 2.6 you can use these releases. You can get them here: http://kernel.org/pub/linux/kernel/v2.6/

Considering we don't have boot results for the released 2.6.13 kernel on a non-NUMA i386 ... maybe we should wait for a bit before installing it on our own x86 systems?

Maybe you mean the internal kernel API - which affects to modules, NVIDIA & friends etc. That API is unstable on purpose, as explained here: Apologia for constant gratuitous compatibility breaking changes

I have one carefully considered word to describe my reaction to the referenced apologia: UNCONVINCING. I recognize that it's not a piece of cake, but it's not rocket science either.

Depending on the version of the C compiler you use, different kernel data structures will contain different alignment of structures.

Maybe if this were the dark ages and we did not have compiler packing pragmas to handle exactly this situation, then it would be a problem. Anyone with a modicum of experience knows how this problem is customarily turned into a non-problem.

Depending on what kernel build options you select, a wide range of different things can be assumed by the kernel:
- different structures can contain different fields

The way to handle this is via embedding versioning/option information in the structures. This is not rocket science.

Some functions may not be implemented at all, (i.e. some locks compile away to nothing for non-SMP builds.)

A non-problem using the same techniques above mentioned.

Parameter passing of variables from function to function can be done in different ways (the CONFIG_REGPARM option controls this.)

No rationale being offered, you're going to have to cut me some slack if I say this sounds like a gratuitous obstruction.

etc. etc.

Pretty much all the offered rationales are excuses for not designing properly up front. I don't mean designing perfectly; that's impossible; but designing with a view to accomodating future changes in a way that preserves compatibility with carefully written modules.

Bottom line: the linux kernel design is an extremely impressive piece of work; a magnificent gift to the community, but it does have major conceptual flaws in that no one ever bit the bullet in terms of maintaining module compatibility across versions; particularly important if you have any interest in supporting binary-only modules in any real sense. If it's a religious decision that we don't want to support binary-only modules because they are inherently evil, then fine; there's no point in arguing religion; but if it's that we don't want to support binary-only modules because it's too hard and we don't want to exercise the discipline necessary to support them effectively, that's another thing entirely.

The PCMCIA subsystem has been substantially re-written. The good news is that the lame support that was there before has hopefully been fixed. The bad news is that people who had something running with the old, lame support may find out that 2.6.13 breaks it. Support in Fedora is *probably* coming but don't expect it right away:

> Date: Mon, 29 Aug 2005 01:23:22 -0400
> From: Dave Jones
> Reply-To: For users of Fedora Core releases
> To: For users of Fedora Core releases
> Subject: Re: 2.6.13 Kernel
>
> On Sun, Aug 28, 2005 at 08:36:30PM -0600, David G. Miller (aka DaveAtFraud)
wrote:
>> The 2.6.13 version of the kernel is now available from
>> http://www.kernel.org/ as well as the usual mirrors. Anyone have any
>> thoughts as to plans by Fedora to move FC4 to the 2.6.13 kernel?
>
>'soon'. But not probably not in the next week or two.
>
>> I'm normally not a "new kernel junkie" but PCMCIA support gets
>> significant fixes in 2.6.13.
>
> read as: almost complete rewrite. It needs completely different userspace,
> and is almost guaranteed to break existing configurations.
> We're still trying to make it work in rawhide.
>
> I'm not sure how this is going to play out in FC4 yet.
> It may even come to the extreme of reverting chunks of it so that
> the existing cardmgr style in FC4 continues to work.

Unfortunately, by the time 2.6.13 finished building on my laptop (HP Pavilion zv6015) last night, it was too late to do much besides see if it would boot (it did). Next step is get ndiswrapper working with 2.6.13 (haven't even recompiled it yet) and then see if a PCMCIA card I insert is at least recognized (it would be a nice start).

Linux DOES has a stable ABI, this is, the syscall interface. It hasn't been changed in years...I know people who is running binaries compiled for linux 1.0 in 2.6 kernels. If your app breaks or works bad when changing the kernel version (ej: openoffice when the semantics of yield() where changed in 2.5) is probably because your app was broken in first place. Now, regression and bugs can happen too, but those aren't on purpose

Maybe you mean the internal kernel API - which affects to modules, NVIDIA & friends etc. That API is unstable on purpose, as explained here: http://kernel.org/git/?p=linux/kernel/git/torvalds /linux-2.6.git;a=blob;h=f39c9d714db3d6bf2f6440d2f6 cf9353057eeae5;hb=02b3e4e2d71b6058ec11cc01c72ac651 eb3ded2b;f=Documentation/stable_api_nonsense.txt

Or maybe you mean "compatibility" WRT gtk & friends, if GTK breaks compatibily thats their broblem

it's kernel.org. they mirror other people's stuff.

I dare you to compare:

http://www.kernel.org/pub/linux/kernel/v2.6/linux- 2.6.12.5.tar.gz
(keep in mind that most of this is driver code)

http://ftp.gnu.org/gnu/glibc/glibc-2.3.4.tar.gz
http://ftp.gnu.org/gnu/coreutils/coreutils-5.2.0.t ar.gz
http://ftp.gnu.org/gnu/bash/bash-3.0.tar.gz
http://ftp.gnu.org/gnu/grep/grep-2.5.1.tar.gz
http://ftp.gnu.org/gnu/tar/tar-1.15.tar.gz
http://ftp.gnu.org/gnu/gzip/gzip-1.2.4.tar.gz
http://ftp.gnu.org/gnu/sed/sed-4.1.4.tar.gz
http://ftp.gnu.org/gnu/findutils/findutils-4.2.23. tar.gz
http://ftp.gnu.org/gnu/gcc/gcc-4.0.1/gcc-4.0.1.tar .bz2
http://ftp.gnu.org/gnu/binutils/binutils-2.16.tar. gz
http://ftp.gnu.org/gnu/make/make-3.80.tar.gz
http://ftp.gnu.org/gnu/patch/patch-2.5.4.tar.gz
http://ftp.gnu.org/gnu/diffutils/diffutils-2.8.1.t ar.gz
http://ftp.gnu.org/gnu/gawk/gawk-3.1.5.tar.gz
http://ftp.gnu.org/gnu/bc/bc-1.06.tar.gz

One CD-ROM vendor found that in their "Linux distribution", GNU software was the largest single contingent, around 28% of the total source code, and this included some of the essential major components without which there could be no system. Linux itself was about 3%. So if you were going to pick a name for the system based on who wrote the programs in the system, the most appropriate single choice would be "GNU".
But we don't think that is the right way to consider the question. The GNU Project was not, is not, a project to develop specific software packages. It was not a project to develop a C compiler, although we did that. It was not a project to develop a text editor, although we developed one. The GNU Project's aim was to develop a complete free Unix-like system: GNU.

Many people have made major contributions to the free software in the system, and they all deserve credit. But the reason it is an integrated system--and not just a collection of useful programs--is because the GNU Project set out to make it one. We made a list of the programs needed to make a complete free system, and we systematically found, wrote, or found people to write everything on the list. We wrote essential but unexciting components because you can't have a system without them. Some of our system components, the programming tools, became popular on their own among programmers, but we wrote many components that are not tools.(...)
When Linus Torvalds wrote Linux, he filled the last major gap. People could then put Linux together with the GNU system to make a complete free system: a Linux-based version of the GNU system; the GNU/Linux system, for short. The earliest Linux release notes recognized that Linux was a kernel, used with parts of GNU: "Most of the tools used with linux are GNU software and are under the GNU copyleft. These tools aren't in the distribution - ask me (or GNU) for more info.

you can switch to Linux, which will always be free, always be DRMless ...

Well, not really. Someone could write the drivers to use Digital Restriction Management hardware and authentication routines on Linux machines.

Oh wait, that's already happened.

Of course, your personal version of Linux may not be encumbered with DRM, but that does not mean that Linux as a whole will be free of it.

How does that justify the protection racket that has been set up around the trademark. The term Linux was supposed to be public domain until a US company buggered it up for everyone. Now the people that set out to reclaim the name are now the perps at the top of a slippery slope with the best of intentions.

There are a lot of other alternatives:

• Ignore the scum, and have a laugh when MS brings out a product called LinuxSomething. good luck to them. It wont stop you or I from downloading it from here, and calling it Linux.
• Delegate licensing down to the LUGs, where participation is a good thing. Linux companies most likely already have staff attending the LUG meetings. New Linux companies would do well to have to chat to the local group who can keep tabs on the products to be sure that the use of the trademark is suitable. For-profit licensing would still be required to keep the trademark protection coffers full, but that is much more palatable than the current scheme.
• Ignore the laywers, and organise rallies and boycotts when people overstep the line. Companies that blatantly abuse the good will of the people should not be quietly accused in court -- public hangings are required.
• If nothing else works, rename the kernel regularily to avoid confusion with improper uses of the old name; after all, it appears to have worked for Firefox.

Not quite. Can you take the Open Solaris source code and release "Geek Solaris 4.2"? Nope. Sun would kill you over the Solaris trademark. Yet anyone can take Linux and make "Geek Linux 4.2 (The best Linux yet!)".

Hmm. Check this and then this out.

Oh, I patched it alright

# What is this about GPLONLY symbols?

        * (REG) By default, symbols are exported using EXPORT_SYMBOL, so they can be used by loadable modules. During the 2.4 series, a new export directive EXPORT_SYMBOL_GPL was added. This is almost the same thing, except that the symbol can only be accessed by modules which have a GPL compatible licence (note that this includes dual-licenced BSD/GPL code). This new directive was added for these reasons:
                    o To clarify the ambiguous legal ground on which non-GPL (particularly proprietary) modules lie. A strict reading of the GPL prohibits loading proprietary modules into the kernel. While Linus has consistently stated that proprietary modules are allowed (i.e. he has granted an explicit exemption), it is not clear that he is able to speak for all developers who have contributed to the Linux kernel. While many think Linus' edict means that all contributed code falls under this exemption granted by Linus, not everyone agrees that this is a legally sound argument. The new EXPORT_SYMBOL_GPL directive makes the licence conditions explicit, and thus removes the legal ambiguity.
                    o To allow choice for developers who wish, for their own reasons, to contribute code which cannot be used by proprietary modules. Just as a developer has the right to distribute code under a proprietary licence, so too may a developer distribute code under an anti-proprietary licence (i.e. strict GPL).
            Note that Linus has stated that existing symbols will not be switched to GPL-only. Developers of proprietary modules for Linux need not fear. Furthermore, it is quite unlikely that Linus will look favourably upon the introduction of new core driver APIs which are restricted to GPL-only modules. This would not be in the best interests of Linux. Linus has forwarded me a message he sent to someone else to clarify his views.

However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.

Though many may reply "SCO 5ux0rz and Linux 0wnz" there is a lot of crap in this article. To back up his security claim he cits " In CNET's, May 27, 2005 article entitled "OS Makers Slow to Fix Flaw ". As any bugzilla will show Linux is patched frequently and quickly. Check google news if you don't think Linux is secure Darl. Point one for Darl, 1770 for Linux. Darl references (though gives no link) a study done by the MI2G group. This group is famous for FUD and being special interest lackeys. Great sources.

Next Darl takes Linux to task for disorganization.
Linux will likely continue to face challenges about its development methodologies and roadmaps as long as it continues to be a loosely organized set of volunteers who develop what they want, when they want.. Has he not heard of Novell, RedHat, Mandriva, or Ubuntu? What about the OSTG?!? Are these "loosely organized volunteers?" NO! These are firms, supporting and developing Linux, firms that are pounding SCO into non-existence.

He claims The grand promise of Linux was that it wouldn't fork or fragment into multiple Linux operating systems. . Never have I heard that. The grand promise of Linux is that it is open. Free as in freedom. Unlike the "Open Server" SCO sells, which is neither open nor free.

Next he asks the following.
Who is checking for compatibility across thousands of applications, drivers, hardware and peripherals? Who is verifying backward compatibility? Well if you are using Debian, it is the Debian team. If you are using SuSE it is Novell. Et cetera et cetera. Darl betrays extraordinary ingorance in thinking that all operating systems built on GNU/Linux are the same. Gentoo != Mandriva != Slackware != Knoppix. Ye the media (and Darl, who shouldn't be able to plea ignoracne) continue to ignorantly blanket statement all Linux distros as "Linux".

Frankly this is crap. He admits to being biased, but doesn't have the balls to point out where his bias is. That is because it is everywhere, throughout this ridiculous article.

And who the heck has ever heard of "Steve the Linux Super Villain Guy?" And why would a "popular internet cartoon" lend credence to a serious business claim??

Though I am going to burn Karma for this, the holy Slashdot would be a lot more interesting if it didn't post Media/FUD as news.

One reason could be that Red Hat makes the SRPMS for their RHEL line of operating systems available to anyone. This has allowed projects like CentOS to become so popular. Some people like having the option of using a RHEL-like operating system on servers that they know they can't commit big $$ to. You can install CentOS where it makes sense and pay for RHEL when appropriate. With SUSE Linux Enterprise you can use the operating system on several computers, I think, but you can't use the updates they provide. Which would be fine if the SRPMS were made available but.........

there are ISOs get them here:

ftp://mirrors.kernel.org/suse/i386/9.3/iso/

both CD and DVD, so quit yer bitchin

Being able to download the fully installable ISO images will be great.

Huh. The ISO images are available now. When did that happen ?
ftp://mirrors.kernel.org/suse/i386/9.3/iso/

Now I CAN tell people to use something better than Fedora Core.

OS X has an Infinium driver == evil.

Linus has an Infinum driver == good.

I see. I must be new here.

This is terrible. There is another OS kernel with a TPM driver now as well. We'd all better stop using this 'Linux' thing as well.

Yes, because he hasn't had reason to proclaim such a statement yet.

there is dra administrator or data recovery agent (not in xp sp2)
wich is automaticcaly created so go to ftp://ftp.kernel.org/pub/dist/planb/ and download plan b

start-off from cd reset admin account to *

reboot login and there you go

local admin rights + you can decrypt efs encrypted files(except for admins encrypted files)

doesn't look that hard too me

Most OS X users probably don't even know they're running a flavor of Linux, but they can slowly get more involved with that world while as they feel comfortable.

Since no one explicitly why OS X is not Linux: OS X is based on the Mach Microkernel and FreeBSD (a BSD Unix variant, not Linux (see section "What is Linux?").