Month: November 2003

A very interesting thing happened the other day: A very crafty attempt was made to hack the Linux Kernel via the source code. On this occaision the attack was a kludge and developers and server administrators noticed the problem.

In fact, the hacker attacked a copy of the code held on a CVS server and maquerading as a developer made changes to kernel internals that would bump a process to the UID of root uder certain flag/signal conditions.

Had the hack found it’s way into the mainstream release, things could have been pretty bad. Modern Linux kernel development is robust enough to notice such blatant hack attempts though and these kind of problems are likely to remain rare.

Although there are Public CVS trees, a key point here is that they are read rather than write: the compromised code would never have been imported back into the main tree.

On this occaision we have the chaps at BitMover to thank, who alongside the kernel developers realised there was a problem and fixed it within 24 hours.

Worthy also of note: Linux people always tend to use the Best Tool For The Job. Not afraid to use closed source or commercial offerings, the Linux kernel is mainly maintained within BitKeeper, a non-free Source Code Repository system. Perhaps this say a lot about CVS!

For more nitty gritty, read the technical details at kerneltrap.org.

TurboTas

The Cell project which IBM, Sony and Toshiba began researching in 2001 is reaching a critical stage. Depending on which sources you read, chips could begin shipping as early as 2004.

TurboTas understands that the already eagerly awaited PlayStation3 could use the chip at it’s core.

‘Who Cares’ I hear you cry, ‘I’m not botherd about whats in a PS3!’. Think again, perhaps: It’s consistently the games console and games market in general that pushes the consumer technology to it’s limits. GFX cards, CPU’s, Memory, Networks. Modern games tax all these things to the outer limits.

Okay, to the meat: the Cell technology is based around very high speed cores with phenomenal speed broadband interconnects. The massivley parallel devices can be multi core on a die (Probably up to 64 for big server blades) and also use the broadband to talk off die.

Some numbers? Okay, a 64 core chip in something like a graphics server will be able to deliver 2 Teraflops. Ultimatley, this should lead to a single cabinet in the machine room being capable of 16 Teraflops. In context 2200 G5’s can cluster to deliver around 10 Teraflops. This is pretty fast the G5 itself is no slouch: IBM’s Power PC 970 chip is phenomenal.

TurboTas.

IBM Press Release
STI cell processor defined
Two years ago, Sony and Toshiba and IBM (STI) announced that they had teamed up to design an architecture for what is termed a system-on-a-chip (SoC) design. Code-named Cell, chips based on the architecture will be able to use ultra high-speed broadband connectivity to interoperate with one another as one complete system, similar to the way neural cells interoperate over the brain’s network.

Market demand for STI cell processor
IBM expects Cell to define an entirely new way of operating. Cell’s underlying architecture will enable it to manifest itself into many forms for many purposes, helping to open up a whole new set of applications. Incorporating this architecture, chips will be developed for everything from handheld devices to mainframe computers.

IBM strategy with STI cell processor
IBM has an unmatched history and capability of building custom chips and believes the one-size-fits-all model of the PC does not apply in the embedded space; embedded applications will require a flexible architecture, like Cell. Cell also brings together, for the first time, many leading-edge IBM chip technologies and circuit designs developed for its servers.
STI cell processor benefits
Cell will take advantage of IBM’s most advanced semiconductor development and process technologies. These cells will deliver high performance while consuming small quantities of power.

In a keynote speech Wednesday at the Enterprise Linux Forum, Red Hat CTO Michael Tiemann bluntly stated that today the company’s “focus is on the enterprise.”

Tiemann also said the software world “can no longer afford the folly of proprietary architecture.”

While Red Hat will now concentrate on building its enterprise distribution and associated applications, the company will support the now-independent Fedora Project as, essentially, a replacement for the old free-to-download user-level Red Hat.

Tiemann’s speech coincided with Red Hat’s rollout of its new, substantially upgraded and highly scalable Enterprise Linux version 3, which costs $179 per year for a single workstation, including basic support, and nearly $3000 per year per year for the top-end server version, including 24/7 phone support.

These are the same prices Red Hat charged for its previous versions of Red Hat Enterprise Linux, and users who have already purchased RHEL 2.1 can upgrade to the new version at no charge.

Red Hat seems to be taking a cue from Sun’s support of OpenOffice.org as a 100% free project while selling essentially the same code, plus some add-ons, as StarOffice. OpenOffice.org experiments in ways that StarOffice does not. One example is an in-progress Mac OS X port. Successful experiments are available for later StarOffice inclusion.

Fedora, said Tiemann, will provide “the stimulus and the R&D” behind many future Red Hat innovations. And while Fedora explores the leading edge of Linux, Red Hat will concentrate on producing stable, mature enterprise products — and, obviously, on marketing those products.

Fedora has replaced the old, 100% free Red Hat Linux Project, which essentially was “Red Hat” to legions of downloaders. There has been a “free for the download” version of Red Hat since Mark Ewing released his first version of “Red Hat Linux” in October, 1994.

The only mention of the Fedora Project on Red Hat’s main page as of 23 October 2003 was a small button in the lower right-hand corner that did not describe it in any way.

In a Boston Globe article published 16 October 2003, Red Hat CEO Matthew Szulik was quoted as saying, “Our goal is to become the defining technology of the 21st century.”

Obviously, that goal can’t be reached without substantial income.

So far, even though some old-line members of the free and open source software communities do not like Red Hat’s recent concentration on the enterprise market, the strategy seems to be working in a financial sense. Red Hat declared its first operating profit last quarter — $240,000 — and the company’s revenues have consistently increased year over year (this year’s target gross is $115 million) at a rate similar to that experienced by Microsoft when the first Wintel boom was in full swing.

In the short term, Red Hat’s decision to divorce its “free” and “pay for” offerings from each other has obviously been a roaring success. Whether this move will help Red Hat (or Linux or open source) become “the defining technology of the 21st century” remains to be seen.

But don’t forget: Unlike Microsoft, which is and always has been the only publisher of the Windows operating system, Red Hat faces competition from at least 10 ‘major’ Linux distributions. And hundreds of ‘minor’ ones. That is a factor that will keep Red Hat from ever achieving 100% domination of the Linux market — or from ever realizing profit margins as large as Microsoft earns from its core products.