This discussion has been archived.
No new comments can be posted.
More From Tanenbaum
Related Links Top of the: day, week, month.
Math is like love -- a simple idea but it can get complicated. -- R. Drabek
Math is like love -- a simple idea but it can get complicated. -- R. Drabek
Article text (Score:5, Informative)
Ken Brown's Motivation, Release 1.2
Background
On 20 May 2004, I posted a statement [cs.vu.nl] refuting the claim of Ken Brown, President of the Alexis de Tocqueville Institution [adti.net], that Linus Torvalds didn't write Linux. My statement was mentioned on Slashdot [slashdot.org], Groklaw [groklaw.net], and many other Internet news sites. This attention resulted in over 150,000 requests to our server in less than a day, which is still standing despite yesterday being a national holiday with no one there to stand next to it saying "You can do it. You can do it." Kudos to Sun Microsystems and the folks who built Apache. My statement was mirrored all over the Internet, so the number of true hits to it is probably a substantial multiple of that. There were also quite a few comments at Slashdot, Groklaw, and other sites, many of them about me. I had never engaged in remote multishrink psychoanalysis on this scale before, so it was a fascinating experience.
The Brown Book
I got an advance copy of Ken Brown's book. I think it is still under embargo, so I won't comment on it. Although I am not an investigative reporter, even I know it is unethical to discuss publications still under embargo. Some of us take ethics more seriously than others. So I won't even reveal the title. Let's call it The Brown Book. There is some precedent for nicknaming books after colors: The International Standard for the CD-ROM (IS 10149) is usually called The Red Book.
Suffice it to say, there is a great deal to criticize in the book. I am sure that will happen when it is published. I may even help out.
Brown's Motivation
What prompted me to write this note today is an email I got yesterday. Actually, I got quite a few :-) , most of them thanking me for the historical material. One of yesterday's emails was from Linus, in response to an email from me apologizing for not letting him see my statement in advance. As a matter of courtesy, I did try but I was using his old transmeta.com address and didn't know his new one until I got a very kind email from Linus' father, a Finnish journalist.
In his email, Linus said that Brown never contacted him. No email, no phone call, no personal interview. Nothing. Considering the fact that Brown was writing an explosive book in which he accused Linus of not being the author of Linux, you would think a serious author would at least confront the subject with the accusation and give him a chance to respond. What kind of a reporter talks to people on the periphery of the subject but fails to talk to the main player?
Why did Brown fly all the way to Europe to interview me and (and according to an email I got from his seat-mate on the plane) one other person in Scandinavia, at considerable expense, and not at least call Linus? Even if he made a really bad choice of phone company, how much could that cost? Maybe a dollar? I call the U.S. all the time from Amsterdam. It is less than 5 cents a minute. How much could it cost to call California from D.C.?
From reading all the comments posted yesterday, I am now beginning to get the picture. Apparently a lot of people (still) think that I 'hate' Linus for stealing all my glory (see below for more on this). I didn't realize this view was so widespread. I now suspect that Brown believed this, too, and thought that I would be happy to dump all over Linus to get 'revenge.' By flying to Amsterdam he thought he could dig up dirt on Linus and get me to speak evil of him. He thought I would back up his crazy claim that Linus stole Linux from me. Brown was wrong on two counts. First, I bear no 'grudge' against Linus at all. He wrote Linux himself and deserves the credit. Second, I am really not a mean person. Even if I were still angry with him aft
Okay. (Score:5, Informative)
You want to read this article. [slashdot.org] It should explain what is happening.
And you would pick this up from the links, but just for the record: Tanenbaum is this european guy who once upon a time in the 80s wrote a textbook on operating systems which came with a simple UNIX-like operating system called "Minix". Ken Brown is some guy who works for something called the "Alex de Torqueville" (sic?) institute and he's writing a book which appears to mostly consist of slander against Linus Tourvalds and/or the Free Software movement.
Re:Little Help? (Score:5, Informative)
Ken Brown works for the Alexis de Tocqueville Institution, who is basicly in the business of writing "impartial" reports for people with money. It's public knowlege that they've taken money from Microsoft in the past for reports. He is writing a book accusing Linus of not writing Linux.
Linux is Obsolete (Score:4, Informative)
For the rest of you who don't know 'the past' Prof. Tanenbaum with Linus, you may refer to the famous mailing list log "Linux is Obsolete" [fluidsignal.com].
Linus seems to be doing excellent work and I wish him much success in the future.
So I guess Prof. Tanenbaum can give higher grade than "F" to Linus now.
Both Prof. Tanenbaum and Linus are my favourite persons. I'm so happy to see this happy ending in real life.
Re:Embargo? (Score:5, Informative)
As for free speech - this isn't a legal thing, it's purely done out of respect for the publishing process and basic good manners. Once the final version is available it's fair game for anyone.
Re:Mirror mirror on the wall ... (Score:3, Informative)
Multiply small integers (Score:4, Informative)
Re:Little Help? (Score:3, Informative)
Really? Can you document that? I worked for Digital Research around that time, and there were many tales about how the MS/IBM thing happened, but I never heard this version. And what do you mean by "a clone of CP/M ported to the 8086"? There were three versions of CP/M at that point, one of which was an 8086 version.
Comment removed (Score:5, Informative)
Re:Little Help? (Score:4, Informative)
Professor Andrew Tanenbaum is a professor who has written some great books. I'm very happy to have read read them. One of his books was Operatin g System Desgin and Implementation [bookpool.com] in which he describes operating systems with a toy, minimal, Unix-like operating system for the 8088 called Minix [minix.org]. It wasn't a really useful OS, but it was small enough to take a look at the code to any particular subsystem and learn how it worked. As an example of its mimilalism, it did have some hardware memory protection between processes, but did so with segment registers. That limited the size of each program to 64k.
Minix wasn't free or open source software. (ideas that were pretty much in their infancy) Tanenbaum sold it through his book publisher. Not for much, probably just enough to make it worth Prentice-Hall's time. Without the Internet as a cost effective distribution medium, someone had to take the orders and mail the disks. People loved tinkering with Minux, though. They ported it to other platforms, (Atari ST, Amiga, Sparc, 80306, etc.) They added to it and started distributing patches. Linus was using Minix-386 before he managed to get Linux to be self-hosting. In some reports, it was Linus' annoyance at having to pay for Minux that inspired him to make Linux free software.
Ken Brown, on the other hand, is someone whose name isn't very recognizable in technical circles. I'm tempted to say that he is a nobody, but maybe I just don't hang around the right circles. (Or on the other hand, maybe if I've never heard of him that means that I hang around the right circles.) I first read about the Alexis de Tocqueville Institution a couple of years ago when they published a paper [adti.net] questioning the security of free and open source software, and sold the paper in a through a system that allowed people to download the paper without purchasing it. Most of the links on their site are either links to articles from news sites about the institutes press releases, or links to papers that they promise will be ready soon.
Re:His comment on Slashdot: (Score:4, Informative)
----
Eh? He did a bit more than just write MINIX!
He's an IEEE & ACM fellow, has written a number of well-known and widely-used books... But don't take my word for it, Google away.
As for sour grapes, I don't sense any, and I exchanged a few emails with him after the last story.
Re:Okay. (Score:3, Informative)
Re:Microkernel in linux? (Score:1, Informative)
2). Writing a simple monolithic system is easier than writing a simple microkernel one (but a complex MK should be easier than a complex mono one, unless it's the original HURD of cource
3). Reading Linuses original posts on the subject (at the time, I have not reread them since) it quickly came apparent that he had not done any OS/systems development on anything but an intel processor, intel x86 (and clones) context switch much slower than for instance a IBM PPC so a microkernel system takes a much higer performance hit on an intel based system than it does on most other processors. OS-9 for instance feels fairly sluggish on an PC compared to some of the other implenmentations. (IBM claims that a PPC performs context switches 10x times faster than an otherwise similarily powerful intel chip, that figure is probably marketspeak but even if only 5x it would mean a lot to a MK performance)
4). Almost all advanced OS design and system programming books/coursewar available at the time dealt with monolithic systems and while most CS peopleetc. had heard the word microkernel they had no real idea what it meant, even today most people have no practical or even theoritical knowledge of a MK system as can be witnessed by those silly WinXP is microkernel" slogans that crop up here from time to time.
5). Linus cannot admit being wrong ? Recent posts to AFC would indicate so, but doing pshycoanalysis based on usenet postings is probably not in the best of taste.
Re:Multiply WHAT? (Score:3, Informative)
Well, given that the number isn't exactly right (the book don't cost $100), here is the calculation:
$100 x 500 x 14 (years) = $700,000. Lets round up that number.
Re:Microkernel in linux? (Score:2, Informative)
The point is, what a micro kernel enforces automatically can be done manually in a monolithic kernel, but faster. The problem is that, not being automatic, more problems are possible, and more work has to be done. It's a tradeoff, and with the MHz and cost wars, tradeoffs generally favor speed.
Re:I'm with Tannebaum about microkernels (Score:4, Informative)
K42 by IBM (Score:2, Informative)
K42 link [ibm.com]
Microkernel reality (Score:5, Informative)
The file systems and networking are user programs. You can add new file systems; there's one that mounts .zip files, there's NFS, and there's Samba.
In Linux terms, visualize a system where there's the /proc file system for inter-program communication, and everything works through that mechanism.
The drivers really are outside the OS. I've written a FireWire camera driver for QNX, and it's a user program. It's privileged in that it does map some real memory shared by the device, and it can talk to the device directly, so it could potentially cause a crash by making the device write someplace it shouldn't. (That's really a weakness in the PC's I/O architecture; there's no MMU between devices and memory, for historical reasons dating back to the original IBM PC.)
Debugging a driver is like debugging a normal program. You can even run a driver under a debugger. You can kill a driver while it's running, and it's no big deal. (If you have real memory mapped, it's not recovered until the next boot, so I had to restart my machine about once a week while doing driver development.) Mainframe people have been doing this since the 1960s, but it's rare on PCs.
The basic penalty for using a microkernel is one extra copy and context switch for every file system operation. If your system is doing anything besides I/O, you'll probably never notice. If you're running a web server that serves mostly plain pages (little Perl, Java, PHP, etc.), you'd probably notice the overhead.
So why are microkernels so rare? They're hard to write well. You can't just hack them together like a UNIX clone. There are some tough design problems to be solved. If those are botched, message passing performance will be terrible. Message passing and CPU scheduling need to work together. This forces certain design decisions in the scheduler. It's also why adding message passing to an existing system tends not to work well. The Hurd crowd has been thrashing on this issue for a decade. I would have loved to see something as good as QNX from the Hurd people. But it didn't happen.
Mach didn't really work out as a microkernel. Mach started from 4.3BSD (considered bloated in its day), and versions of Mach below 3 had 4.3BSD in the kernel. MacOS X is not a microkernel system; the BSD stuff is in the kernel. Basically, retrofitting a microkernel architecture to an existing UNIX kernel didn't work.
What you do get from a microkernel like QNX is predictablity. The kernel changes very little and is very reliable. Good microkernels, like QNX and IBM's VM, settle down into versions that almost never change and have very long MTBFs. This brings down total cost of ownership.
Re:Little Help? (Score:5, Informative)
Re: microkernels the best approach (Score:5, Informative)
Hmmm, No...
It's a widely perpetuated myth that NT is a microkernel. It may have started out that way, but has long since grown through millikernel, centikernel, decikernel to full blown kernel... (and beyond if you count browser, media player and kitchen sink OS embedding)
The linked letter from Prof Tanenbaum touches on this point too... He says:
Re:You lose. [MODERATORS?] (Score:3, Informative)
Well actually if you had read the moderator FAQ you would know that you should really browse at -1 in order to catch abuses.
Re:Changed opinion (Score:5, Informative)
Dynamically loadable modules does not make Linux a microkernel design. It would only be a microkernel if the filesystem code ran in a different address space. But because ext3.o runs in the same address space as the kernel, it is most definitely a monolithic design. It is not a "microkernel design running as a monolithic kernel". That's just a nonsensical statement.
You clearly understand that the significant distinction between microkernel and monolithic is the address space for the subsystems. So I can't understand why you'd suggest that kernel modules makes Linux "essentially a microkernel design". Look at the address space for ext3.o; it's kernel space.
I don't see Linux evolving into a microkernel until there's hardware support for cross address space branching. Don't hold your breath.
Re:Little Help? (Score:2, Informative)
All the TRS-80 machines included Microsoft Basic in ROM, for instance. The IBM-PC had Microsoft Basic in ROM that you could run without even having a floppy disk controller in your PC.
Microsoft was an early entrant in the hobby/personal computer market with one of the first significant commercial products that people found useful.
However, it's more popular to believe the anti-Microsoft drivel and revisionist history written by pundits years later.
Anybody with a sense of ethics and value for historical accuracy would be ashamed to be involved with the early 'history' part of the film 'Revolution OS' or that Robert X. Cringely (not his real name, just a shared psuedonym he stole control of and commercialized) fraud-history work 'Pirates of Silicon Valley.'
Re:Little Help? (Score:4, Informative)
Re:Linux is Obsolete (Score:4, Informative)
Linus would also get a second 'F' for writing i386 specific code, but that problem is long gone. Are there any other OS, apart from NetBSD, that supports as many architectures as Linux?
It seems to me that AST was quite impressed with Linux from the start -- especially the Posix compliance -- but disagreed strongly with the design (or lack thereof). You have to remember that many people actually wanted to turn Minix into something like Linux, and that was out of the question. Linux is not as good for what Minix was supposed to do: teach OS principles. If you consider the context of the discussion, AST does not look as arrogant, and certainly not stupid (although his predictions for the future were a bit off target, but I don't think Linus expected his little hobby to be the subject of a multi-billion dollar suit either).
Re:I resent that! (Score:2, Informative)
And besides Ken, you really work for the Tomás de Torquemada Institute [newadvent.org] who had connections to the Monastery of Santa Cruz! I bet you weren't expecting that!
Re: microkernels the best approach (Score:3, Informative)
No.
Others have already explained why those are not examples of microkernels. If you want a real example of a successful microkernel, look at QNX (which is very successful in its target market).
Re: microkernels the best approach (Score:3, Informative)
Re: microkernels the best approach (Score:3, Informative)
Dunno, what does the one have to do with the other? It's like saying 'If macrokernels are the best approeach, why does Windows suck do much?'
The Hurd is a set of user-space server running on top of a microkernel (currently Mach), together providing the old Unix experience besides other more interesting things.
Really, it's not taking so long because it was a microkernel (it is NOT!), it's just that there's nobody working on it.
Michael
Re:I'm with Tannebaum about microkernels (Score:4, Informative)
Re:Okay. (Score:3, Informative)
Re:Microkernel in linux? (Score:1, Informative)
Microkernels demand interfaces of one sort or another, since the idea is that major components should be pushed out to userland. If the interface is permissive enough (the much-loved AmigaOS, for instance, was a 'microkernel' with no concept of memory protection), your system is modular, but your security/robustness gain is nil. (Stability may improve, to the extent that modularity helps write good code... we all know the arguments for it and agaisnt it.) If you start trying to design 'safe' interfaces, things can get complex, and you may take the performance hit.
A worst case is, perhaps, when you do something insane like Apple Darwin's "xnu," and take a microkernel but implement an entire monolithic kernel beneath it as a 'service.' (I sincerely doubt they'll ever reap benefit from that decision, but you never know.)
Meanwhile, QNX is a *very* fun system, but it's worth noting that running a nuclear plant or a medical monitoring device is a fairly different problem from supporting a massively-multiuser server. It works great, but some of the last of its five 9s or whatever (referring to the concept of 99.999% availability, or whatever's claimed) comes from the availability of a system watchdog to restart failed services... a microkernel concept, sure, but the rest of the stability is simply for good coding under relatively stress-free environments (nobody'd call QNX the most *secure* OS, and we all know how long a Netware server can survive forgotten in a closet), and the monolithic/Linux attitude is that the service should simply not fail in the first place.
(What makes QNX a win for control systems is the modularity -- convenient for developers trying to fit a system on 'embedded' hardware; instead of recompiling a kernel, you just make a new filesystem image -- and the 'realtime' scheduling, which gives developers a hope of determinism that's hard to come by with Linux or Windows... the 'kernel' won't unexpectedly block for two seconds to conduct some disk I/O, while with a 'general purpose' solution like Linux, you'd have to do a lot of profiling and rewriting of the system to assure yourself of the same guarantee. (The 'realtime' Linux vendors have done said work, of course... but who'll actually be better-faster-cheaper for a particular implementation is something of a case-by-case issue.)