Slashdot Mirror

I saw some news about Linux doing some work on drive schedule control, and then it kinda died. I've had several people tell me that drive scheduling is unimportant because you always want response from the programs you're running, but while that's true, each one may have a different level of importance.

/sys/block/?d?/queue contains the scheduling configurables for each drive (this can also be specified as "elevator=_____" for a kernel startup option. There are four IO schedulers: noop, deadline, anticipatory, and cfq.

Does that answer your question?

I saw some news about Linux doing some work on drive schedule control, and then it kinda died. I've had several people tell me that drive scheduling is unimportant because you always want response from the programs you're running, but while that's true, each one may have a different level of importance.

/sys/block/?d?/queue contains the scheduling configurables for each drive (this can also be specified as "elevator=_____" for a kernel startup option. There are four IO schedulers: noop, deadline, anticipatory, and cfq.

Does that answer your question?

I don't know how advanced Tridge's tools are, but GNU CSSC (downloadable from the the GNU FTP servers) has a certain amount of read-only compatibility with BitKeeper. See also the BitBucket project (code here).

If you read the thread here, Linus strongly implies that Tridge was not sniffing packets, but analyzing a BK repository on disk:

...I was hoping that that would convince Tridge that trying to muck around with the internal BK file format was not worth it, and avert the BK trainwreck. Larry was ok with the idea to make my export format actually be natively supported by BK (ie the same way you have "bk export -tpatch"), but Tridge wanted to instead get at the native data and be difficult about it.

Your initial assertion was that "anyone" that disagrees with "me" is a zealot. I missed the part where I sadi that. Go read my post again. I was theorizing that the "zealots" who are now cheering and offering their "told you so" blabber will spin this as yet more "proof" that any type of commercial software is "evil".

I was hardly the one to pioneer this concept, BTW. According to Richard Stallman, I am immoral and should find another job. Perhaps things like these are where the term "zealot" applies. There's also an interesting definition over at Wikipedia, if you need further information.

I guess I'm just being a "The Bungi" Zealot

Take a number and get in line.

But in this post, Linus says he was writing scripts to export data in an SCM-independent format. McVoy actually offered to support this directly in BitKeeper, but Tridge insisted on "being difficult about it" and gaining access directly to the binary data. Linus suggests he's bitter at Tridge for it, and now he can't use BitKeeper.

When we were trying to figure out how to avert the BK disaster, and one of
Tridges concerns (and, in my opinion, the only really valid one) was that
you couldn't get the BK data in some SCM-independent way.
So I wrote some very preliminary scripts [...snip...] Larry was ok with the idea to make my export format actually be natively
supported by BK (ie the same way you have "bk export -tpatch"), but Tridge
wanted to instead get at the native data and be difficult about it. As a
result, I can now not only use BK any more, but we also don't have a nice
export format from BK.
Yeah, I'm a bit bitter about it.

Does the name 'git' strike anyone else as an odd name for a (kind-of) SCM system?

Or is this Linus making a not-so-subtle pot-shot at Larry McVoy?

Wasn't one of the provisions of Bitmover's grant of free Bitkeeper licenses to the open source crowd that those who partook of said licenses not work on SCM tools of their own for some time? If so, what's Mr. Linus Torvalds doing working on one just days after announcing that he's giving Bitkeeper up?

JA: What if your job requires you to use non-free software?

Richard Stallman: I would quit that job. Would you participate in something anti-social just because somebody pays you to?

Its not that Linus wasn't happy with the BK licensing, its that there were issues involved with developers (specificially a contractor under OSDL) trying to to reverse-engineer some of the features in BK.
Even though an agreement was reached that this developer would stop doing this, apparently he continued.
Larry McVoy (and BitKeeper) responded by saying that they were removing the Free BK license, and employees of OSDL (which Linus is) were not eligible for a free license under any conditions.

At this point Linus really didn't have a choice whether to switch SCM systems or not.

Kerneltrap story.
Read the 2nd paragraph under "Free Versus Free".

I remember reading this article and leaving comments on it. RMS was quite clear:

JA: What about the programmers...

Richard Stallman: What about them? The programmers writing non-free software? They are doing something antisocial. They should get some other job.

My answer to that was not quite polite.

BTW. don't be surprised to get -2Troll in a short while here.

...especially because RMS said recently that the changes will be details. There is a line in www.gnu.org's license list mentioning that the GPL v3 may be compatible with the Affero General Public License, which seems to implement something similar. However, I definitely don't think it will end up as absurd as this article makes it out to be.

RMS doesn't write code for GCC anymore, or have anything to do with it. He only hacks a little on Emacs these days. Also, there are very practical reasons for not converting GCC over to C++. Keep in mind it has to bootstrap itself when being ported to new platforms; those platforms may not have a good C++ compiler.

Can anyone comment on the possible ramifications of having all these large-scale commercial contributors using a tool that Linus & Co no longer use/have access to?

RTFA. Larry McVoy explains that the reason he's no longer giving Linux devs free copies of bitkeeper is because their needs are so different from what paying customers would want.

BK's advantage was in loosely-coupled projects, without the infrastructure for a central source repository. The only groups that really need that functionality are large open-source projects. Small open-source, or small-to-large commercial projects can afford (and prefer) their own main server.

On the other hand, moving away from bitkeeper expands the kinds of features Linux can accept. Previously, a networked filesystem with versioning metadata would've been a license violation. (Not that such would necessarily be a good idea for Linux, but it is now an option)

I wonder how this will be affected by BitMover's recent decision to end development of their "free as in beer" version. Are enough KDE developers employed by companies willing to buy comercial versions? or will they be moving to back to CVS?

drupal is more than a php based blog, I have been using it since 4 years and I like to think of it as an application framework than can act as a blogserver as myblog is. Drupal was described as community plumbing but now "simpley" is an
" ...open source content management platform.

Out of the box it can do a lot, has clean code, a really cool development and support community and lots of high profile (read: often slashdotted) sites are hosted on drupal such as kerneltrap and spreadfirefox

Drupal is on the forfront of technology. It was one of the first to use real taxanomy, clean URL's and distributed authentication.
The name drupal? That is another story

Harald Welte did a very interesting presentation about GPL Enforcement in Germany at FOSDEM two weeks ago.

He is one of the few, with Theo de Raadt, who really fight against proprietary software. See this Kerneltrap.org feature about OpenBSD fight against closed source drivers for wireless.

Have you looked at the Open Graphics Project?
Here's an interview with the lead developer of the programmable cards.

Mail list and more info: http://lists.duskglow.com/mailman/listinfo/open-gr aphics/

You can tell a great deal about the maturity of a programmer by the quantity, and quality, of comments.

That is, the younger, less mature programmers are commenting tons and actually documenting their work because their teachers stressed how truly important it is. This is really boring. As the programmer matures, the comments decrease but the QUALITY of the comments increase, and end up being quite novel to read.

At the end of September I'm off back to University on a years sabbatical from Red Hat to study for an MBA. I've made the decision that I'm basically going to vanish for the year so I can concentrate on the course, and on the pet side project of learning Welsh.

FYI, Andres Salomon's patchset provides the foundation for Debian's kernels and has been discussed recently on kerneltrap here and here.

FYI, Andres Salomon's patchset provides the foundation for Debian's kernels and has been discussed recently on kerneltrap here and here.

All other replies suggest copying /usr to a RAM disk, but there's a much easier solution if it's just about loading apps quicker after starting up. First, look into the swappiness kernel parameter. Then, create a mechanism to read these files into memory.

Short story: add two lines to /etc/rc.d/rc.local:

echo 100 > /proc/sys/vm/swappiness
for i in `find /opt/OpenOffice`; do cat $i >/dev/null;done &

The first line tells your kernel to swap away as much as possible, leaving lots of space for disk buffering. The second line reads the files of your choice in the cache.

Note that this is all very primitive and there are lots of enhancements, but the basic idea should be clear now.

You couldn't have been searching all that hard
Google: Linux patents kernel

First result: http://kerneltrap.org/node/388

Where are all the geom HOWTOs?

The linked man page is "tasty" n'all, but details on implementing such magical wonders, until recently, have been rather scarce.

This man page is better than the one linked to in the original post. There's also some information from committer (read: major contributor to ggate ) Pawel Jakub Dawidek in Poland.

Not that the info isn't there now, right under man, but for a while it was all very vague.

When searching about all that is BSD, don't forget Google's special google.com/bsd section.

You can also search the freebsd-geom mail list archives to learn more.

geom-gate sure looks nifty! It's akin to block-level NFS (though that's most likely an extremely oversimplified view). All the fun things you can do with geom you can do over your network. Need worldwide distributed, encrypted, multi-level RAID? Go right ahead!

Pretty slick. We'll be hearing more about this.....

You mean like this (2002)?

Or also seen here.


Since the mods didn't see the first comment about this... surely they won't notice I shamelessly ripped this off the post right above this one ;)

You mean like this (2002)?

Or also seen here.

Xfree (and derivatives, X.org incl) is optimized for throughput, not interactivity. IOW welcome to the world of high latency. Even Linus has complained about this on numerous occasions, and it's not going to be fixed without a total rewrite.

Also note that Xfree is single-threaded, so it's bound to suck at serving up multiple window GUIs no matter how it's 'fixed'. see here for more on the official kludgy way of improving X interactivity...

Just from today's news...
Genetic Algorithm in the Kernel

From the description, I'd be concerned that releasing the Accelerator code under a non-free (as in speech) license would be incompatible with the linux kernel's GPL license as it could be argued that it is a derived work.

See also http://kerneltrap.org/node/1735.

In practice, it may be enough of a gray area that it won't be a problem -- although it may scare off any company wishing to invest in it.

Personally, I'm just getting sick and tired with the maintainability and reliability issues that binary modules usually incur..

/* Would you like fries with that? */
hme_write32(hp, tregs + TCVR_FRAME,
(FRAME_WRITE |(hp->ipaddr...

Certainly, the old Sun driver code hits near the 2-5% mark. However, unless you're looking for comedy in the Linux kernel code comments, this doesn't help the maintainability or understanding of this line of code.

There are a lot of good metrics defined in the references for this paper and a lot of good research on maintainability was quoted. From reading this paper, however, I can say that this is not a reference that should join them nor does it say anything useful.

------
1. The metric Netwon defines force, the meters-per-second-squared defines acceleration and the kilogram defines mass. The equation F=MA relates these three quantities, but even in it's absence they have numerious properties they give to any assocated quanities. It is these properties and their implications that make Newtons and kilograms usefull. Quantities without dimension are rare in a real science, such as physics or biology and typically arise from equations that define them as well as any other

What would you have him test? Some imaginary database that is developed on FreeBSD and which runs fastest on FreeBSD? OK. Meanwhile, the rest of us will try to get some useful work done.

We need open hardware

Yes, because we all know that there hasn't been a recent rash of privelege escalation bugs found in linux lately.

A copy from an anonymous kernel trap comment.

Go figure.

It always inevitably boils down to:

JA: What if your job requires you to use non-free software?
Richard Stallman: I would quit that job.

read this and support the opensource movement

http://kerneltrap.org/node/4622
http://kerneltrap.org/node/4622

as it spreads out to new fields of endeavour.

good luck to all the open and honest initiatives and fair projects, social practices and helpful goodwilling people around the world.

cooperation and openness help this world better than pure capitalism and monetary systems.

cheers.

Last I checked it already was.

http://news.zdnet.com/2100-3513_22-5227102.html/
http://linuxgazette.net/107/pramode.html/
http://kerneltrap.org/node/3240?PHPSESSID=262a094f ee677def32a8cc4d1b858f99/
http://searchenterpriselinux.techtarget.com/origin alContent/0,289142,sid39_gci969248,00.html/

Just to name a few

Perhaps you are thinking of UTOTO which RMS mentioned in this interview.

lot of specialty software is very boring, and there just isn't any interest in the OSS community in developing similar software. - but but but according to RMS if software is written as closed source, it should not have been written at all, and the developers who have done it, should have found something else to do instead, like give up their jobs and go do something less harmful to the society!

To make an analogy, it's like buffer overflows in C/C++. Sure, it's possible to write C/C++ code which doesn't have explotable buffer overflows -- the techniques to avoid them are widely known and well documented. However, even extremely talented, security minded programmers STILL wind up making classic mistakes.

If you're worried about buffer overflows in your system, you could take two approches: you could implement it in C/C++ and adopt a very rigorous set of development procedures with extensive code reviews and testing, or you could use a language which is not vulnerable to buffer overlows. Which method do YOU think is going to require less effort and have a greater chance of succeding? Likewise, if you're worried about SQL injections, which is a better approach: one which requires you to not make any mistakes, or one which which won't let you make the mistake in the first place?

If even the hacker gods can't use the tool securely 100% of the time, what chance do we mere mortals have of doing so? The answer is, none at all. Therefore, sound engineering practice dictates that we take our human limitations into account and use tools and techniques which will mitigate the damage caused by those mistakes which we will inevitably make.

Here's another thing to consider: even if we can guarantee that our code is 100% free of exploitable holes, we still have to deal with the possibility that some other vulnerability (say in the operating system or the web server) will allow an attacker to compromise the public server. If all the security is on the web server, we've created single point of failure which will cause the entire system to be subverted.

An attacker who rooted the web server would be able to find out the database connection information, and would then therefore be able to connect to the database with the same permissions as the web application uses. If the web application has permission to read, update, and delete all tables, so does the attacker. If, however, we follow the principle of least privilidge, the public account only has permission to execute a limited set of well-designed stored procedures. This way, having direct access to the database doesn't allow the attacker to do anything that he wouldn't be able to do via the application.

Furthermore, we should practice defense in depth, which means putting in in redundant security features, compartmentalizing the application logic between the web server and the database server, and putting the components on seperate physical machines. This limits the damage which can be done if any single component fails. The fact that this architecture gives us better scalability is a free bonus.

A good engineer always asks himself "If THIS component fails, how will the entire system react?" Using an monolithic security model in a mission-critical system is bad engineering. A software engineer who designs an application which can be compromised by a single line of bad code is just as negligent as a civil engineer who designs a suspension bridge which will collapse if a single cable breaks.

There's a good writeup on this thread at KernelTrap, too. Includes links to the full thread, which is quite fascinating.

There's a good writeup on this thread at KernelTrap, too. Includes links to the full thread, which is quite fascinating.

if I make a basic software package[,] ... I can then release it to the community, where someone else will likely improve it. ...Remember, software itself doesn't really have any value -- raehl

Hmm... http://kerneltrap.org/node/4484?from=150&comments_ per_page=50#comment-18399

It would seem that raehl, IBM and I have reached the same conclusion regarding the value of software and the nature of software as a commodity.

Hope it's contagious...

The article says:

I challenge anyone else who defends the status quo to show me some innovative new titles from the major developers.

Reading this reminded me of some comments I've seen in discussions about free software. Often, the discussion is framed from the perspective of the open source movement and the values that movement promotes, which are not the same as those of the free software movement. As a result, people frame the debate as if we can have either innovative software or free software, or that software freedom isn't worth pursuing because there is insufficient innovation. RMS addressed this in a recent interview: (emphasis mine)

Jeremy Andrews: I have read that the free software model tends to imitate existing software, rather than blaze new trails and developing completely new technologies.

Richard Stallman: To speak of a free software "model" is somewhat misleading. The open source movement speaks of a "development model", but our concern is for the user's freedom, not how the program is developed.

Free software doesn't always imitate, but often it does. There's a good reason for this: freedom is the main goal, and innovation is secondary.

Our goal is to develop free software so that we can use computers exclusively with free software. In 1984, we started with nearly zero (we had TeX, nothing else). We had a lot of catching up to do, so we have done it. Even if GNU/Linux had no technical innovations compared with Unix, it would be completely superior because it respects your freedom as Unix does not.

I was amused to see that in a recent interview with Richard M. Stallman he referred to DRM as Digital RESTRICTIONS Management.

Although I'm not a big fan of spin, the current political climate makes renaming things with misleading names a necessity. When you say "Digital RESTRICTIONS Management", it makes it fairly clear that it's a technology aimed at limiting personal liberties.

P.S. Yes, I know this is a repost, but...

I was amused to see that in a recent interview with Richard M. Stallman he referred to DRM as Digital RESTRICTIONS Management.

Although I'm not a big fan of spin, the current political climate makes renaming things with misleading names a necessity. When you say "Digital RESTRICTIONS Management", it makes it fairly clear that it's a technology aimed at limiting personal liberties.

Is it just me, or does anyone else find the Stallman photo in the interview strikingly similar to one Ron Jeremy?

As I wrote in the comment to another KernelTrap story, using the term "GNU/Linux" (referring to the GCC and glibc essential role in the system) is totally misleading.

Both GCC and glibc are in the current state despite the RMS and FSF efforts. For GCC, remember the situation from the 2.8 times, when an independent team (egcs) had to fork GCC, because the FSF-managed development of GCC was dead. In the same way remember years of work that H.J.Lu invested in Linux libc, because GNU libc was unmaintained and unusable. And of course the work of Ulrich Drepper, who took GNU libc2 and developed it into a form usable in Linux-based system. Ulrich considers none of his work on glibc to be a part of a GNU project (details here, see the bottom part of the text). And it looks like even the present situation in the GCC development is the same (anonymous comment at KernelTrap).

So I can say run GCC/glibc/perl/X.org/TeX/etc/Linux system, but it has nothing special to do with GNU and FSF, and I just prefer the short name "Linux" (named after a single biggest, always-running, and essential component of the system).