Slashdot Mirror

Screw the CPU scheduler at this point. The kernel folks are missing the obvious and utter brokenness of the IO scheduling. These bugs have been outstanding about a year now!! And it's not just AMD64 anymore either. Quoth the kernel bug report:

"Now, as far as this bug being AMD64 only. We develop a portable data analysis
tool and we run it on Intel Core Mobile systems (Sony UX series, Panasonic
Toughbook series) and see this bug or one almost exactly like it on those
platforms as well.
"

http://bugzilla.kernel.org/show_bug.cgi?id=7372
http://bugzilla.kernel.org/show_bug.cgi?id=8636
http://www.nabble.com/IO-activity-brings-my-deskto p-to-its-knees-(2.6.22.1-ck1)-t4192136.html
http://forums.gentoo.org/viewtopic-t-482731-start- 500.html

At first, deadline IO was touted as an answer, but that doesn't completely fix things.
Some say Native Command Queueing is broken. One person claims deadline + NCQ disabled helps.
Some say the kernel's vfs_cache_pressure settings help, while others refute it (compare kernel bug report versus page 21 of the gentoo forum thread). But no one understands what's really broken in the kernel.

Can we please get Ingo working on IO scheduling? PLEASE?

There are two big penguins then a small fleet of little penguins underneath them with 'SPE' tattooed on their chests in red letters. Like so http://www.kernel.org/pub/linux/kernel/people/geof f/cell/debian-penguin-shot.png/

ESX server does not under Windows. They have several products, most of which do, but ESX Server is (sold as) an operating system unto itself, and does not install or run under any other OS. It does however use a Linux kernel to boot, which then hands control over to the "VMkernel". The question is whether this "VMkernel" is itself derived from the Linux kernel, or otherwise so co-dependent on it (as in, a kernel module) that it cannot function without it.

The COPYING file clearly states:

Damn I can't think of any major science/tech developments it would be good to have back in 2004... :s/P

A lot of posters are complaining that the author has been to clever, that it won't help convince people to read the documentation, that people shouldn't have to search for it. These posters need to look more closely at the situation.

Let's take a look for this super hidden documentation. Here's the opening, cleverly hidden as a comment as the very first thing in the file named lguest.c. That seems a pretty freaking obvious place to put an introduction to the system. For all the article's spin, all the author has done is place his documentation in comments with the code; perhaps the most obvious place for it. The "make Preparation!" and friends simply extract and collect the documentation scattered around the code into a single place, sort of like Doxygen, JavaDoc, or any of a pile of other documentation systems.

That the documentation contains some mild humor is irrelevant. Does it convince people to read it when they wouldn't have otherwise? Probably not. But it can lighten the mood for both author and reader, inspiring people to read on a little more. It can help relax the reader and make them more receptive to the ideas presented. I'd be suspicious of any non-trivial code base completely free of humor; suggests someone operating on the mistaken belief that programming must be Serious Work, the sort of person who ignores basic human behavior and thinks people can behave like robots.

There's this little project of which you may have heard: http://www.kernel.org/

• Make latencies deterministic
• Reduce latencies if used correctly
• Add slight over head to overall througput

The official kernel site says 2.6.23 is only on release candidate 1.

I've killed some time on this since it's a pretty interesting idea. It turns out there are plenty outside the D and F range. It does seem to like pages with a single Flash object and not much else, so that's bad. It also makes some pretty arbitrary decisions which don't mean squat to many sites. There are some sites that get enough traffic that speed is a factor but not so much that a content delivery network is really necessary, for example.

I skipped the actual link and score on sites that are pretty much just representative of the sites around them. I wanted to include them by name, though, to show where they fall. I've stuck mostly to main index pages, and I've noted where I've gone deeper.

A: Google (99%), Altavista main page (98%), Altavista Babelfish (90%) (including upon doing a translation from English to French), Craigslist (96%), Pricewatch (93%), Slackware Linux, OpenBSD, Led Zeppelin site at Atlantic (100%), supremecommander.com, w3m web browser site (96%)

B: Apache.org (87%), the lighttpd web server (84%), Google Maps, which also got a C once (84% in most cases), Perlmonks (84%), Dragonfly BSD (85%), Butthole Surfers band page (81%), 37 Signals

C: One Laptop Per Child,, ESR's homepage, the Open Source Initiative (78%), Google News (73%), Lucid CMS (74%), Perl.org (75%), lucasfilm.com, Charred Dirt game

D: gnu.org, The Register, A9 (66%), kernel.org, Akamai (64%), kuro5hin.org, freshmeat.net, linuxcd.org, Movable Type (61%), Postnuke, blogster.com, Joel on Software (67%), Fog Creek Software, metallica.com, gaspowered.com, Scorched 3D (68%), id software (64%), ISBN.nu book search

F: MS IIS (49%), microsoft.com, msn.com, linux.com, fsf.org, discovery.com, newegg.com, rackspace.com, the Simtel archive (26%), CNet Download (29%), Adobe (58%), savvis.com, mtv.com, sun.com, pclinuxos.com, freebsd.org, phpnuke.org, use.perl.org, ruby-lang.org, python.org, java.com, Rolling Stones band page (56%), powellsbooks.com, amazon.com, barnesandnoble.com, getfirefox.com

My site for my company (96%) gets an A (no, I'm not going to get it slashdotted) which is pretty simple but has a pic and some Javascript on it. Several sites I have done or have helped design with someone else get C or D ratings.

2.6.22-8? Ubuntu is cutting edge considering the latest is 2.6.22-1

http://www.kernel.org/pub/linux/kernel/v2.6/

Their CPU's and chipsets are adequate, just don't try to use their USB 2.0 controller in linux. http://bugzilla.kernel.org/show_bug.cgi?id=6374

As an amd64 owner who's been taken aback by the very long-standing disk IO crisis:
http://bugzilla.kernel.org/show_bug.cgi?id=7372

I'm curious how this new scheduler may help/hurt me?
Deadline appears to help the problem, albeit not fix the root cause.

And is there any kind of disk IO scheduling work going on to help this in the future?
I'm really not sure why amd64 would be so broken while x86 is not. (IOMMU?)

It would be helpful were you to actually read all of the attached links completely instead of seeing some bogus reports in the Gentoo area and dismissing the whole thing based on that subset.

I'd suggest http://bugzilla.kernel.org/show_bug.cgi?id=7372#c1 08 and http://bugzilla.kernel.org/show_bug.cgi?id=7372#c1 12 as the best summary of the kind of problem people are running into. There are no optical devices involved.

It would be helpful were you to actually read all of the attached links completely instead of seeing some bogus reports in the Gentoo area and dismissing the whole thing based on that subset.

I'd suggest http://bugzilla.kernel.org/show_bug.cgi?id=7372#c1 08 and http://bugzilla.kernel.org/show_bug.cgi?id=7372#c1 12 as the best summary of the kind of problem people are running into. There are no optical devices involved.

Dude, you haven't read the links have you?

http://forums.gentoo.org/viewtopic-t-482731-start- 450.html
"... And of course all along I've been experiencing the slowdowns with the SATA (now back to IDE) disk access mentioned at the beginning of this thread."

http://bugzilla.kernel.org/show_bug.cgi?id=7372
"... The only thing related to libata I can think of is NCQ interacting badly with io scheduler..."
"...Yes, and this means that the problem is getting worse with TCQ/NCQ enabled, but
it is not the root cause."

This issue really is about disk IO performance in general, not specifically CD burning! Please don't make light of what is a very serious problem. It was at a point today where I had a hard time even starting "top" today during some DV video playback. Unacceptable.

So I have never ever checked source code from Linux, and I don't know C or C++, but I decided to look at some changes. The first code I see has a goto in it. I always thought goto was considered harmful?

http://git.kernel.org/?p=linux/kernel/git/torvalds /linux-2.6.git;a=blobdiff;f=fs/utimes.c;h=b3c88952 465fa28cce7e0bb213fceaf59873fdf9;hp=480f7c8c29da13 ee10941f5cf5e560faffbde0a6;hb=1e5de2837c166535f9bb 4232bfe97ea1f9fc7a1c;hpb=4e99325b462ba180757685826 21af74a6b79d2a5

And there isn't even a direct download link the full file in question...

Ummm, I guess now is the opportune time to wish the kernel.org sysops some good luck?

For anyone in the dark, disk IO has been broken sometime after 2.6.17 on amd64.
I thought I was going crazy, being on 2.6.18 and discovering that any disk activity slows down the whole system, let alone accesses to any other disk.

Then I found a 19-page thread on the gentoo forums that says I'm not alone and it's not unique to a particular chipset:
http://forums.gentoo.org/viewtopic-t-482731-start- 450.html
(with evidence that the deadline scheduler may alleviate _some_ of the problem but not the root cause)

And more importantly the kernel bug report here:
http://bugzilla.kernel.org/show_bug.cgi?id=7372

So I'm happy people aren't ignoring the problem. ...Or should I be worried that something so utterly fundamental has been lost in the shuffle across so many kernels in the past year? Amid all the eagerness to add new features since then (virtualization for example, and now complete rewrites of firewire?!?!).

Why can't we have a 2.7 kernel for this stuff?

You probably already downloaded and built today's patch, but just for grins go download the whole file.

What, you didn't download the patch yet? What kind of geek are you? :)

This is Linus' canonical justication for 80-column lines. Full version (worth reading!) at:
http://git.kernel.org/?p=linux/kernel/git/torvalds /linux-2.6.git;a=blob;f=Documentation/CodingStyle; hb=HEAD

-------
Linux kernel coding style

[...]
Chapter 1: Indentation

Tabs are 8 characters, and thus indentations are also 8 characters. There are heretic movements that try to make indentations 4 (or even 2!) characters deep, and that is akin to trying to define the value of PI to be 3.

Rationale: The whole idea behind indentation is to clearly define where a block of control starts and ends. Especially when you've been looking at your screen for 20 straight hours, you'll find it a lot easier to see how the indentation works if you have large indentations.

Now, some people will claim that having 8-character indentations makes the code move too far to the right, and makes it hard to read on a 80-character terminal screen. The answer to that is that if you need more than 3 levels of indentation, you're screwed anyway, and should fix your program.

In short, 8-char indents make things easier to read, and have the added benefit of warning you when you're nesting your functions too deep. Heed that warning.

Don't put multiple statements on a single line unless you have something to hide:

if (condition) do_this;
do_something_everytime;

Don't put multiple assignments on a single line either. Kernel coding style is super simple. Avoid tricky expressions.

Outside of comments, documentation and except in Kconfig, spaces are never used for indentation, and the above example is deliberately broken.

Get a decent editor and don't leave whitespace at the end of lines.

Chapter 2: Breaking long lines and strings

Coding style is all about readability and maintainability using commonly available tools.

The limit on the length of lines is 80 columns and this is a hard limit.

Statements longer than 80 columns will be broken into sensible chunks. Descendants are always substantially shorter than the parent and are placed substantially to the right. The same applies to function headers with a long argument list. Long strings are as well broken into shorter strings.

[...]
Chapter 4: Naming

C is a Spartan language, and so should your naming be. Unlike Modula-2 and Pascal programmers, C programmers do not use cute names like ThisVariableIsATemporaryCounter. A C programmer would call that variable "tmp", which is much easier to write, and not the least more difficult to understand.

HOWEVER, while mixed-case names are frowned upon, descriptive names for global variables are a must. To call a global function "foo" is a shooting offense.

GLOBAL variables (to be used only if you _really_ need them) need to have descriptive names, as do global functions. If you have a function that counts the number of active users, you should call that "count_active_users()" or similar, you should _not_ call it "cntusr()".

Encoding the type of a function into the name (so-called Hungarian notation) is brain damaged - the compiler knows the types anyway and can check those, and it only confuses the programmer. No wonder MicroSoft
makes buggy programs.

LOCAL variable names should be short, and to the point. If you have some random integer loop counter, it should probably be called "i". Calling it "loop_counter" is non-productive, if there is no chance of it
being mis-understood. Similarly, "tmp" can be just about any type of variable that is used to hold a temporary value.

If you are afraid t

>> Item 8 states that the majority of the traffic on P2P is pirated material .... While no one is going to argue the amount of pirated content available on P2P networks, given (a) that many Linux releases use BT as a distribution medium (Fedora, Ubuntu, CentOS, OpenSUSE, etc.) with images up to a DVD-ROM's worth of data (4.5 Gb), (b) the amount of video-based material (movies and television) that is out there, whose files are no doubt larger than audio MP3 rips, is it fair to assume that the music industry's concerns are a relatively small portion of the overall P2P traffic? Going to a popular tracker site such as http://mininova.org/, the largest BT swarms are typically found for the prior night's TV shows. Item 9 (sales decline directly related to pirated content) has been contested for some time; the industry has reduced its number of releases, the majority sellers are now the big-box stores who carry little in the way of back-catalog material in favor of chart-topping new releases, and the rise of sales in DVDs (sell-through DVD prices comparable to new-release CDs appearing as a better value).

What you describe isn't unusual within Linux at all. If you have a problem with a module, you can often visit a developer's page, and fairly often get a later version of the driver which is still compatible with the kernel version you're running (e.g. IVTV). Then you either patch your kernel tree or compile the modules outside the kernel tree (for instance the pcmcia subsystem) and load it in. As for competing subsystems, think about oss vs alsa, or ipfwadm vs ipchains vs iptables, or udev vs devfs. It's just that all this is usually left to distro maintainers, because most people don't care. Maybe what you think you want is rock-steady kernel interfaces so a given driver is compatible with more versions of the kernel - but then you can kiss progress goodbye. I think Linux has it right: support the current way of doing things, plus the outmoded way of doing the same thing for a year or two, then get rid of it. Otherwise you're going to replicate the 20 year transition period it took Microsoft to get from DOS to XP.

People-ready business

> It may feel good, but if it doesn't do what I want and how I want it to, then I'm not buying it.

I hear you, man. Here's how you go about getting apps like that. You just need a few simple tools:

vi or emacs: you can get from http://www.vim.org/ or http://www.gnu.org/software/emacs/, respectively

C compiler: very good one @ http://gcc.gnu.org/

OS: kernelspace of a good one is @ http://www.kernel.org/ but there are others.

And then after you install this stuff, you write a piece of software that does exactly what you want, how you want it done. It might be good to take an OSS app that does /almost/ exactly what you want, how you want it done as your base, because that way you'll have to write less code, but do whatever you feel is easiest from a software engineering perspective.

Unless you at least partially write your own software, you will /NEVER/ get exactly what you want from it, whether you use Windows, Linux, or Plan 9 from Bell Labs. Everyone wants something slightly different and everyone has different tastes. At least with open source, you can start with a good base to build on. With closed-source you're stuck with what they give you. With closed-source, if you want to make it right, you'll have to start from scratch.

What, you're not willing or not able to put forth this effort? Well, OSS owes you nothing; they're trying to make the software THEY want, after all. How about shutting up and letting us get back to that? k thx.

Linux does NOT use the "standard GLP" it makes a few changes for example he removed the "and later versions" part.

He has added a note clarifying that userland programs making system calls are not derivative works. That is clarifying his interpretation of the license. At the most, it adds an exemption, which is common practice.

The "and later versions" clause is not part of the GPL, it lies outside it. In effect, it makes the work automatically multiple licensed as and when new versions of the GPL comes out: so if you distribute something under GPL v2 with that clause, when v3 comes you it will become dual licensed under 2 and 3, when v 4 comes out it will be triple licensed etc. This allows people to redistribute under v2 or any later version, with or without the "and later versions" clause.

The Linux kernel is a monolithic kernel with various sections that can be enabled, disabled, or compiled as a module that can be added into the running kernel later.
The standard kernel includes support for a number of devices, file systems, and other functions.

The IPTables code allows the user control over his/her network traffic by passing the traffic through "tables" of rules.

The latest version of the kernel source may be obtained from http://www.kernel.org/

The product is called Linux. The fact that GNU contributed some technology that they had lying around to Linus' product doesn't give them naming rights.

http://git.or.cz/index.html has all you need to get started with Git. It is suggested you start with this tutorial http://www.kernel.org/pub/software/scm/git/docs/tu torial.html ; it goes over init, add, commit, branching, merging and I don't remember what else, but it is all you need to get your work done.

anybody have a good tutorial? (not the crappy one which comes with it

I think the current tutorial is pretty good, actually. (But I'm biased--I wrote most of it. Suggestions welcomed....)

See also the user manual for a more thorough treatment.

anybody have a good tutorial? (not the crappy one which comes with it

I think the current tutorial is pretty good, actually. (But I'm biased--I wrote most of it. Suggestions welcomed....)

See also the user manual for a more thorough treatment.

It takes a while for the mirrors to sync. Same thing happens when Microsoft releases new updates too every patch Tuesday. 2.7G files can't be copied instantaneously.

http://mirrors.kernel.org/fedora/releases/7/Fedora /i386/iso/

Has the DVD iso for x86. I'd feel bad about posting a url to a mirror site if this one didn't have 2gbit/sec bandwidth!

It takes a while for the mirrors to sync. Same thing happens when Microsoft releases new updates too every patch Tuesday. 2.7G files can't be copied instantaneously.

http://mirrors.kernel.org/fedora/releases/7/Fedora /i386/iso/

Has the DVD iso for x86. I'd feel bad about posting a url to a mirror site if this one didn't have 2gbit/sec bandwidth!

...there is a need for ability to configure a modern Linux system from scratch..

My browser and other browsers are downloading this exact same page.. look it's parallel programming and no one had to do anything special.

I'm gonna bet there are a lot of groups and developers that completely disagree with that statement. I'm thinking of the Apache Group, MySQL developers, Perl developers, Slashcode developers, Linux developers, etc...

Say that to entire highly parallel capable OS's written in C. Also, C has a nice, simple threading model, it's just not quite as nice as some of the others out there (python).

Are you aware of the reasons for the lack of a stable binary API? If not, I suggest you read the following:

http://git.kernel.org/?p=linux/kernel/git/torvalds /linux-2.6.git;a=blob;f=Documentation/stable_api_n onsense.txt;h=a2afca3b2bab6fb923fb9eda102073606d15 c278;hb=HEAD

The best way to get something supported by linux is to get it into the mainline tree as early as possible. Linux has support for hardware with a single unit in existance, so its not as if its really hard to get drivers accepted into the tree.

As for not wanting to release hardware specs....do you really think that their competitors don't already have the specs on the hardware? I had a roommate who worked for a company that did nothing but reverse-engineer chips and write specs on how they worked.

And to imply that Richard Stallman "politely asks" people to call it GNU/Linux is inaccurate. He rants and raves about it, and basically calls anyone who refers it by the name Linus gave it (ie. "Linux") bad people.

The product is called Linux.

The fact that GNU contributed some technology that they had lying around to Linus' product doesn't give them naming rights.

1. not some, GNU project's software represents significant portions of any GNU/Linux variant. In fact, you can't even separate one from the other ever since it's 0.0.1 first release. Even Busybox/Linux variants require the GNU system in order to be built...
2. not past, as just suggested, even today GNU project's software is a very significant part of any GNU/Linux variant you have around the neighbourhood
3. not naming, the FSF and Richard Stallman are not taking credit after Linus' work. Linus' taking credit after *their* work. When Stallman asks people to say GNU/Linux, he's not renaming Linux. It is your wrong perception of what Linux is that leads you to think otherwise. All he's asking is for people to recognize the role of the GNU operating system, and to remember the freedom they have, instead of merely focusing on a popular but factually incorrect version of history. Even Linus said that Linux wasn't meant to be something big and professional like GNU . So there you go. At least then Linus didn't try to misappropriate the GNU project.

In fact, that's the WHOLE POINT of open source software--

If Stallman had wanted to keep marketing control over his invention, he shouldn't have open-sourced it.

Anyone can take GPL software and create a product and call it whatever they want, no matter what Stallman says or thinks. No one is forced to prepend "GNU/" to their product name.

Thus, calling it GNU/Linux without written permission is diluting the Linux trademark and implying a business relationship with the Free Software Foundation, isn't it?

1. A Trademark doesn't grant you omnipresent control on the word, and GNU/Linux is is way far from the controlled by trademark law parts
2. It's not diluting anything since he's not calling Linux by nother name, he's asking people to recognize the GNU project and not just merely the kernel of this particular variant.
3. For there to be a business relationship you have to have contracted something.

The product is called Linux.

The fact that GNU contributed some technology that they had lying around to Linus' product doesn't give them naming rights.

1. not some, GNU project's software represents significant portions of any GNU/Linux variant. In fact, you can't even separate one from the other ever since it's 0.0.1 first release. Even Busybox/Linux variants require the GNU system in order to be built...
2. not past, as just suggested, even today GNU project's software is a very significant part of any GNU/Linux variant you have around the neighbourhood
3. not naming, the FSF and Richard Stallman are not taking credit after Linus' work. Linus' taking credit after *their* work. When Stallman asks people to say GNU/Linux, he's not renaming Linux. It is your wrong perception of what Linux is that leads you to think otherwise. All he's asking is for people to recognize the role of the GNU operating system, and to remember the freedom they have, instead of merely focusing on a popular but factually incorrect version of history. Even Linus said that Linux wasn't meant to be something big and professional like GNU . So there you go. At least then Linus didn't try to misappropriate the GNU project.

In fact, that's the WHOLE POINT of open source software--

If Stallman had wanted to keep marketing control over his invention, he shouldn't have open-sourced it.

Anyone can take GPL software and create a product and call it whatever they want, no matter what Stallman says or thinks. No one is forced to prepend "GNU/" to their product name.

Thus, calling it GNU/Linux without written permission is diluting the Linux trademark and implying a business relationship with the Free Software Foundation, isn't it?

1. A Trademark doesn't grant you omnipresent control on the word, and GNU/Linux is is way far from the controlled by trademark law parts
2. It's not diluting anything since he's not calling Linux by nother name, he's asking people to recognize the GNU project and not just merely the kernel of this particular variant.
3. For there to be a business relationship you have to have contracted something.

Does this mean that Richard Stallman trying to shove "GNU/Linux" down people's throats is really a misappropriation of a trademark?

1. In 1983 Richard Stallman initiates the GNU project. That's the project to create a fully Free Software operating system.
2. 8 years later, in 1991, Linus Torvalds first announces a kernel that works with GNU software, and later on releases the very first release of his kernel for an i386 computer to work with the GNU operating system
3. The addition of Linux to the GNU operating system results in the GNU/Linux variant, which is the GNU system running with the kernel called Linux

Big fucking deal. Linux fixed this on April 24th. Maybe OpenBSD could learn a lot from them.

1. Millions of hackers? There isn't a single FOSS project that millions of hackers have contributed too.

2. There are very few people with the experience to write a good much less great 3d driver.

3. Even with the specs I am guessing that the majority of contributions will be security or code clean up and not performance optimizations.

Sure, take a look here. The first Linux kernel was released in October 1991. If you had any reservations about its shortcomings, you can still get fixes for those today.

Let's compare that to a closed source equivalent. In 1991 I bought a copy of Microsoft Windows 3.0 for $40. It had several bugs and I used their support service to complain. Their answer? Those bugs would be fixed in version 3.1. So I asked, when would they send me my version 3.1? Their answer: I could buy version 3.1 as soon as it came out. No, I said, I didn't want to buy version 3.1, I wanted the bugs in version 3.0, for which I had paid $40, fixed. I wasn't interested in paying $45 more to get the additional features in Windows 3.1, all I wanted was the Windows 3.0 for which I had paid $40 working correctly. Can't be done, was their answer.

Now, let's see again, how exactly do you define "service, support, AND legal responsibility"???

And you know what's the worse of it? Although they have, 16 years ago, disclaimed all responsibility for the bugs in Windows 3.0, its copyright won't expire for several decades... Oh, yeah, *LEGAL* responsibility, indeed!

You might want to check out BuildRoot. It's what Gumstix uses for their distributions and it works pretty well.
They have even has the rt patch from Ingo Molnar merged into their standard distribution. Sounds like a Gumstix might
not be a bad way to go now that I think about it. And then you would have some pretty good community support. My $.02 ...

http://www.kernel.org/pub/linux/docs/lkml/reportin g-bugs.html