UpFront

diff -u: What's New in Kernel Development

Zack Brown

Issue #152, December 2006

Alan Cox, Jeff Garzik and others have unveiled a plan to do away with the IDE subsystem entirely and completely replace it with libata. This will not happen immediately, and it will not happen all at once, but the plan does seem to have universal support. Even the creator of the original IDE subsystem, Mark Lord, thinks this is the way to go. For the immediate future, all that's happening is that more code will merge from Andrew Morton's -mm tree into the official kernel, and users will have the option to use that improved support for various hardware if they so choose. The ultimate removal of the IDE subsystem is undoubtedly years away. Alan's recent announcement is only one step down a long road.

The fork from ext3 to ext4 is a reality. Once upon a time, adding features like journaling to ext2 was considered so invasive that folks had to fork ext3 in order to continue that kind of development. Now, the ext3 developers have had to take the same steps in order to add invasive features, such as extents and large block sizes, to the code. Linus Torvalds has stood firm on the idea that the most relied-upon filesystems should not in themselves undergo significant development, but should be rock solid and totally dependable. Now, the new ext4 code is on the fast track to being included in the official kernel. Whether it will ever be as popular as ext3 remains to be seen. Meanwhile, folks like Hans Reiser feel that ext4's easy entrance into the official tree is just further proof of the favoritism he feels is practiced within kernel development. What he doesn't understand is that intelligence and coding ability are only part of the kernel development culture. After all, one of Linus' great discoveries was that everyone has the ability to contribute, and kernel developers don't all have to be uber-hackers. They don't even have to be particularly nice guys, as Alexander Viro and others proudly proclaim. But, they do have to respond to feedback and present their work in more or less standard ways. The more they can be trusted to “do the right thing”, the easier it is to get their code into the kernel.

Adrian Bunk will maintain the 2.6.16 kernel as a new stable tree. This has roughly the same appearance as if we still had the old even/odd stability model, Linus had forked 2.7 for intensive development, and Adrian were going to maintain 2.6 for stability. The only difference, it seems, are the names of the trees and the fact that Linus will not be stabilizing the stable tree himself for any length of time. Adrian's work on 2.6.16 will hopefully solve some of the issues users have had with the w.x.y.z stable tree maintained by Greg Kroah-Hartman and Chris Wright. That tree, although aiming for run-time stability, did nothing to prevent interface changes between 2.6 versions. Interface stability is not addressed at all by that effort, while the 2.6.16 interfaces will not change under Adrian's maintainership.

Pavel Machek has released a driver for ThinkPad fingerprint sensors. So far, users have reported good success with it, though at the moment, it does seem to have some easy-to-trigger failure modes. The big question for Pavel is whether to leave this as a user-space tool or to migrate it into the kernel proper. This is an interesting case, because typically anything that can reasonably be left outside the kernel, would be. Although at the same time, it is also typical to keep hardware support inside the kernel, with few exceptions. The direction of Pavel's code may influence where other drivers will live in the future as well.

Keith Packard from Intel has announced open-source drivers for Intel 965 Express Chipset family graphics controllers, as part of ongoing work by the Intel Open Source Technology Center. Intel seems to be doing the right thing here, acknowledging that the code needs testing and bug fixes, and inviting kernel folks to participate in development. One interesting detail quick to be noticed on the kernel mailing list is that the code seems to be written to interface with an unavailable binary blob, intel_hal.so, if available. Keith explained, “This module contains stuff that Intel can't publish in source form, like Macrovision register stuff and other trade secrets. It's optional, so if you don't want to use a binary module, you don't get to use code written by Intel agents for these features....The driver remains completely functional in the absence of the binary piece and, in fact, has no reduction in functionality from previous driver releases.”

Another Open Letter to Bill Gates

Dear Bill,

I hope you are enjoying yourself in retirement. I (of course) am busy as ever trying to promote free software.

I read recently in the Wall Street Journal that you too have discovered the freedom of information sharing! I read in an article that you are insisting that researchers who receive your funding share their data, tools and results with each other. Awesome! I know that you may think this is another one of those “innovations” that you have come up with, but I have to tell you that this is the very core of the Free Software movement, and it has been going on for more than 35 years.

I remember back in 1969 when I was a student at Drexel University. I found some computers in the basement of Drexel's main buildings that did not come with software. In order to use these computers, I either had to write the software or buy it.

A single copy of a compiler for some language might cost $100,000 US in those days, and that was when a hundred thousand dollars was a lot of money! I could not afford that on my small stipend for food and beer. But there were people in the Digital Equipment User's Society that wrote software and contributed it to the society's library for distribution to other people. It was the study of this software that allowed me to move into computer science. I have never forgotten that.

Of course, you may not have had the same enlightening experience. You went to Harvard and probably could afford the compilers of those days—or maybe you just used other people's machines and compilers to do your work.

In any case, as I left college and went out into the real world, I knew that working as a team is better than working alone, so I continued to push sharing code segments and even whole programs in order to make the industry move forward.

I just got a couple of great ideas!

  1. All of the software you fund should be Free Software.

  2. Use only Free Software in your own work.

  3. Buy medical equipment only if it is supported by Free Software.

I am sure you will see how these fit into the basis of your new endeavors.

Warmest regards,

Jon “maddog” Hall

LJ Index, December 2006

Doc Searls

Issue #152, December 2006

1. Millions of students in the state of Kerala, India: 1.5

2. Number of government and government-aided schools in Kerala: 2,650

3. Percentage of Kerala schools that will use or switch to free software on Linux: 100

4. Thousands of Kerala high-school teachers being trained on Linux: 56

5. Weeks between Richard M. Stallman's visit to Kerala and the state's decision to switch to free software and GNU/Linux: 2

6. Percentage growth rate of the Indian economy: 8

7. Percentage growth rate of the Chinese economy: >10

8. Billions of dollars Intel plans to invest over the next five years in its “World Ahead” program for emerging markets: 1

9. Percentage of the world's population AMD wants to see connected to the Internet by 2015: 50

10. Billions of people with annual incomes less than $4,000 per year: 3.8

11. Billions of people with annual incomes of $4,000–$20,000 per year: 1.5

12. Percentage yearly growth of Cisco's network load: 100

13. Projected percentage yearly growth of Cisco's network load: 300–500

14. Cisco annual percentage growth rate in emerging markets: 30

15. Percentage of new Cisco employees being hired in emerging markets: 12

16. Percentage of television programs Cisco expects will be broadcast over the Net in the future: 100

17. Current percentage contribution of emerging markets to Cisco's revenues: 10

18. Projected maximum future percentage contribution of emerging markets to Cisco's revenues: 40

19. Millions of dollars Google has donated to the One Laptop Per Child Project: 2

20. Number of cities in Africa that Google intends to connect fully with a wireless network: 7

1–5: rediff news

6, 7: New York Times

8–11: Electronic Engineering Times

12–18: World Resources Institute, reporting on a speech by Cisco CEO John Chambers

19, 20: icicemac.com

Microsoft's New Promise

Doc Searls

Issue #152, December 2006

In the Free Software and Open Source worlds, licensing has always been a big deal. Choice of license has a direct effect on the usefulness of code bases, and on their market growth as well.

Some code, however, makes use of standards that are open, yet to some degree, proprietary. Those degrees are often controlled by patents. Lately, much lawyerly thinking has gone into making those standards useful to development efforts and to disarming the patents involved. One of these—perhaps the first—is the Microsoft Open Specification Promise. The Promise is short on legalese, yet too long to describe here, beyond saying it's about what Microsoft won't sue others for, providing others don't sue Microsoft. Lawrence Rosen, author of Open Source Licensing: Software Freedom and Intellectual Property Law (Prentice Hall 2004), says the Promise “...enables the Open Source community to implement these standard specifications without having to pay any royalties to Microsoft or sign a license agreement. I'm pleased that this OSP is compatible with free and open-source licenses.”

The first standards in question involve SOAP and a variety of protocols from the WS-* portfolio. At the time of the Promise's announcement, in mid-September, approving public statements were made by a variety of folks on the open-source side of the table. These include Mark Webbink, Deputy General Counsel of Red Hat, and R.L. “Bob” Morgan, Senior Technology Architect at the University of Washington.

Before bringing the defenses up, bear in mind that this promise has been hammered out through collaboration between open-friendly folks inside Microsoft and countless cooperative conversations with folks from Red Hat, Mozilla/Firefox, XRI/XDI, OpenID, LID, Sxip, Higgins, VeriSign and others, including customer-side entities such as North Carolina State University. The conversation has a name: OSIS, for Open Source Identity Selector (or something like that...even the initialism is open to change). Everybody involved is interested in developing open-source implementations—or products that interoperate with—Microsoft's CardSpace, an identity selector that will be released with Vista and which Kim Cameron and others at Microsoft have for several years been trying to make as interoperable as possible in a world where many other identity systems will be in use. (See the Identity Metasystem article from the September 2005 issue of Linux Journal.) But, the Promise may end up extending to other standards in other areas as well.

Disclosure: I've been involved in these discussions and have worked for some time to help make them happen and to move forward. It's clear we are at the beginning of something here, not an end.

Of course, the whole matter is open for input, debate and adjustment. To help with that, here are some links:

Let us know what you think.

Progress toward the Hackable Kidtop

Doc Searls

Issue #152, December 2006

One Laptop Per Child (laptop.org) hit the news in January 2005 at Davos, when Nicholas Negroponte of the MIT Media Lab announced plans for a “$100 laptop” in a quantity of 100 million, to “revolutionize the way we educate the world's children”. Formal plans for the project were announced in August, and we covered it for the first time in the November 2005 issue of Linux Journal.

Since then, much progress has been made. Jim Gettys (www.handhelds.org/People/jg.html)—prime mover behind the X Window System, handhelds.org (www.handhelds.org) and earlier fun projects like the Unobtainium (a wild hack on Compaq's original iPAQ)—is now VP of Software Engineering. And, there are prototypes. The current generation is fitted in bright orange and green and features rectangular bunny ears (802.11s mesh network antennae); a lid that twists and flips to form a pad; a dual-mode display that the project wiki says, “can readily be mass produced in standard LCD factories, with no process changes” and that “has higher resolution than 95% of the laptop displays on the market today, approximately 1/7th the power consumption, 1/3rd the price, sunlight readability and room-light readability with the backlight off”.

Although the units are designed for kids (the keyboard is 6/10 the size of an adult one), they're also made to hack. Writing on his blog (www.ethanzuckerman.com/blog/?p=824), Ethan Zuckerman says:

...the 500 prototype boards currently built come with a VGA jack soldered on, but production models will leave the jack leads etched on the board, though unpopulated. Want to turn a laptop into a device that can drive an external monitor? Solder one on. Also on the board, but unpopulated, will be connectors for additional RAM and Flash memory, as well as a mini-PCI slot.

And, although some large companies (AMD, Google, News Corp., and Red Hat) are involved, the whole project is very much a work in progress, and it's open to interest and help with what promises to become the most widespread and good-hearted Linux deployment on earth.

Soweto: Power from the People

Jon “maddog” Hall

Issue #152, December 2006

In the spring of 2005, I attended LinuxWorld in Johannesburg, South Africa. It was not my first time to South Africa, but this time, instead of going to game reserves, I took a different trip after the event. A gentleman I had met at LinuxWorld introduced me to Soweto.

Soweto is a township outside of Johannesburg. Before apartheid ended, it was a township mostly of very poor black people. On June 16, 1976, students were killed in riots in Soweto that led to the beginning of the end for that rule of government called “apartheid”.

My guide and I went on a little tour, first to the Photography Museum/School of Alf Kumalo. Dr Kumalo, who often risked his life to get photographs that illustrated for the world the issues of apartheid, was now using his talents and resources to teach young people in Soweto how to be photographers. I saw they were using Adobe's Photoshop to manipulate the digital images that the students were taking for composition training. I pointed out that they should use GIMP instead, because the students were unlikely to be able to afford Adobe Photoshop at home, and therefore they would have to pirate Photoshop. I promised Paballo Thekiso, the tutor that was my guide to the museum and school, that I would send them a copy of GIMP and a book on how to use GIMP from the USA.

Then we left and went to the Soweto museum. I read a lot of the information about what happened there, and as we looked out across the natural bowl-shaped valley, I mentioned to my guide (who was increasingly becoming my friend) that this would be a wonderful place for a mesh network to deliver Internet services to the entire township. We also talked about the benefits of FOSS and how there were no limitations to what students could learn, other than their own desires, assuming they had access to computer equipment and the Internet. I told him of several “success stories” for this concept, including one about a person who had been programming the kernel since the age of 12 and one about a person who had put out his own distribution at the age of 14. I bet him that there were people in Soweto who could “do Linux”, given the opportunity.

We went to the house where his mother grew up and where his nephew still lived. It was a two-room house, and he talked about how there were sometimes two or three generations living in the same house. Although he was glad his nephew was doing well in school and sports, he was afraid that the nephew might turn to drugs, given the environment that still prevailed in Soweto.

We finally had dinner in a great outdoor restaurant on the edge of the township, where I experienced some of the local food and entrepreneurship that was happening there.

Then we drove back to Johannesburg, and I flew home. What I did not know was that my guide (and now friend) was Nhlanhla Mabaso, Open Source Center Manager for the Meraka Institute, and that he had been listening to me.

During the next year, I tried to send two books and two open CDs to the museum on two different occasions, and neither time did they get through. Eventually, toward the end of the year I got them back, with a note on them saying that they were “undeliverable”. It cost me more than $150 US to buy the books and mail them. When I got them back, I was angry. Undaunted, I built into my 2006 LinuxWorld Johannesburg schedule some time to travel to the Photography Museum and to carry the books and CDs with me.

When I got my schedule from Aldean Prior, Director of Exhibits for Africa, the LinuxWorld producer, I noticed there was built in to it a trip to the Satellite Open Source Research Center that was opening as part of the Meraka Institute. I did not think anything about it, but I did keep asking to “go back to Soweto” so I could deliver the books to the museum. When I got to Johannesburg, I found out that the Satellite Center was in Soweto, and that I was invited to speak at the opening.

Nhlanhla drove me to Soweto that morning. We visited the museum, gave the books and CDs to Paballo and found out that he knew one of the people in the Center. We invited Paballo to the first training class the next day, so he could establish contacts and learn more about Free Software.

The President and CEO of the Council for Scientific and Industrial Research (CSIR), Dr Sibusiso Sibisi, spoke during the opening, as well as Dr Ntsika Msimang and several other dignitaries. Dr Msimang is managing the new center. After the rest of the ceremonies were over, Dr Sibisi came over to me and said, “You have no idea how influential your words were.”

Apparently, my conversation the previous year had inspired Nhlanhla to go back to the Meraka Institute and make a presentation for investigating the potential of opening an open-source research center and training facility in Soweto and also to start to set up a mesh network for the township. His team went to Soweto in August of that year and discovered (during a presentation on FOSS) that several of the young people in the audience already knew about Linux, could work with it and that one young man named Bongani Hlope was doing kernel programming as a hobby and conversing with Linus Torvalds via e-mail on kernel issues. Another person named Kgabo Sepuru was running a FOSS consulting service out of his house in Soweto.

In addition, Nhlanhla's team found that local people already had started to set up a mesh network in their broadband-deprived area.

On my second day at the satellite center, they had a day of training. That day started off a bit slowly because it was their first day, but I believe things will get better as they get settled in. Dr Msimang seems to be a competent, enthusiastic director, and I contributed the first book to their library.

In the meantime, Paballo, the tutor at the museum with whom I had been corresponding via e-mail all this time, got really excited about the rest of FOSS and said that he was going to learn Linux and teach it to the rest of the photography students. I promised to send some more books on other aspects of digital photography and image rendering using free and open-source software.

I have had about five or six people come up to me in my life and say, “I listened to you, followed your advice on FOSS, started my own company and now I am a millionaire” or “I listened to you and it changed my life”, but this was the first time I actually have seen direct action to this extent on something I said almost in passing. I can't take credit (nor do I want to) for the hard work that Nhlanhla and the rest of the staff put into making the center a reality, but it sure felt good to have someone like Dr Sibisi tell me those words.

Each of us affects the people around us with our every thought and deed. I often tell people that if they want to see the most influential person in free software, just look in the mirror when they get up. Lots of people do not believe what I say, and others do. Sometimes the effect of what we do and say goes way beyond what we know. I was fortunate enough to see how influential my words were, and therefore, I encourage others also to speak out and experience the same thing.

They Said It

When brokers turn into toll-takers, it's time to throw the bums out.

—Britt Blaser, from a conversation

Make no decision out of fear.

—Bruce Sterling, from a speech at SXSW 2006

We have decided that we will use only free software for computer education in Kerala schools. We have implemented the Linux platform in high schools; it will be implemented in other schools step by step....Our policy is to migrate computer education to free software platforms. We want to make Kerala the FOSS (Free and Open Source Software) destination in India. That is all.

M A Baby, Education Minister, Kerala, India, www.rediff.com/money/2006/sep/02microsoft.htm

We are getting lots of enquiries and orders for pre-loaded Linux operating systems. The hardware sales have gone up because of this.

P K Harikrishnan, President, Kerala Computer Manufacturers' and Dealers' Association

If there isn't enough food in the fridge, do you say “the store must be down”?

—Greg Elin, at a conference

What we need is an open-source, open hardware, wireless implementation for unlicensed spectrum. It can be done. And if it is, it will blow WiMAX out of the water and change the world.

—Thomas A. Freeburg, at a conference