Leading items

The demise of PlayStation Linux

Sony uses embedded Linux in a wide variety of its consumer electronics products: televisions, video recorders, navigation devices, and more. Up until fairly recently, it gave something back to the community by supporting Linux installation on its PlayStation game consoles. While Sony removed the "Install Other OS" option on new PlayStation 3 (PS3) systems back in August 2009, there were millions of older PS3s that could be used for Cell processor hacking on Linux—no more.

Sony announced [Japanese press release] (LWN coverage) that the v3.21 firmware release will disable the "Install Other OS" feature, but it also threatens users with a long list of features that will no longer work if the "upgrade" isn't installed. One might guess that either April fools day is not celebrated in Japan or that Sony Computer Entertainment (SCE) is irony-impaired because the release date for the firmware is April 1. The timing is also a bit suspicious in that it seems like a measure aimed at punishing the "mod" community for a recent successful PS3 "jailbreak".

The ostensible reason for removing the "Install Other OS" feature is "security concerns", but as PS3 hacker George Hotz points out, jailbreaking the PS3 requires opening up the enclosure. The procedure is not for the faint-of-heart—it involves pulsing a particular solder point on the PS3 board "low for ~40ns". This is not something a casual user will have any way to do, if they were even willing to try it. It certainly isn't a vector for malware attacks either.

Unsuspecting PS3 Linux users who upgrade will lose the contents of the Linux portion of the hard drive. Because of that, and other restrictions enforced in the new firmware, Hotz has vowed to find a way around those restrictions. He previously had no plans to create custom firmware for the device, but because of SCE's latest move, he does now:

Hotz has also been part of the iPhone/iPod/iPad jailbreaking community, having released multiple software and hardware jailbreaking hacks for those platforms, so his track record is good. That means it is pretty likely he will come up with a way to run v3.21 and still run Linux—just what SCE evidently fears. But he is also clear that doing so is not about "piracy", it is about taking back the control of your device: "Hacking isn't about getting what you didn't pay for, it's about making sure you do get what you did."

That is the crux of the matter for many. Without the firmware update, PS3 owners will not be able to do a number of things they thought they got when buying the device: playing games online, watching new Blu-ray videos, playing new PS3 games, and so on. The EFF is concerned that new Blu-ray disks could even completely disable the Blu-ray drive by revoking the AACS decryption key in the older firmware. Because of the digital rights management (DRM) features included in content meant to be used by PS3s, SCE has the technical means to stop existing, working devices from performing those tasks on new content. The EFF puts it this way:

SCE would undoubtedly argue that it needs to maintain the integrity of its online games, so requiring certain firmware upgrades to participate in its network is reasonable. There is something to that argument, but there is zero evidence that allowing Linux (or any other OS) to be installed had played any kind of role in game "cheats"—in fact its hard to see how it could. If anything is flawed, it is the hardware which allowed Hotz to essentially circumvent the hypervisor that SCE put in place to wall off Linux from the 3D video hardware. In addition, that argument falls flat when considering playing new DVDs or local games.

It is believed that PS3s which need to be serviced for a hardware problem of some kind will be silently upgraded to the latest firmware, which would wipe out any Linux partition on the disk. So, who owns this device that you have, supposedly, bought and paid for? Once a given set of features is released, and works, isn't the manufacturer honor-bound (if not legally bound) to not actively work to disable those features? Some PS3 customers relied on being able to install Linux, while still keeping the other features of the console. In fact, SCE made assurances that the "Install Other OS" feature would be maintained as recently as February.

It is interesting to note that there are folks in the HPC community who were buying PS3s by the thousands to create Linux clusters. Other than the occasional fragging expedition at lunch, one would guess that the vast majority of those machines never actually run games at all. There is clearly a market for low-cost Cell-based machines, but SCE evidently doesn't see that—or can't make money at it. It may be running PS3s on the razor blade model; selling the consoles at a loss, while making up the difference by selling peripherals and games.

Its hard to see how SCE comes out of this looking like anything other than a bully. It sold hardware with a feature set that folks found attractive, so they bought them. Now, when it is somehow inconvenient for SCE to continue supporting some of those features, it turns them off, with little warning and almost no recourse. The vast majority of PS3 owners may be completely unaffected, but those who relied upon SCE's word may think twice before buying from it again. In the meantime, they are likely to follow Hotz's progress with great interest.

Comments (27 posted)

Open-source biotechnology

The free software community, along with the commercial ecosystem which surrounds it, is widely seen as having pointed the way toward successful, collaborative development of common resources. We have seen a number of attempts to port the free software model to other areas of endeavor. Open content, headlined by sites like Wikipedia, has adopted this model with considerable success. Other areas, such as open hardware, are still trying to find their way. Your editor recently read an interesting book (Rob Carlson's Biology is Technology), which raises an interesting question: is there a place for an ecosystem based around free "software" running on biological processors?

The core point of the book is that biological hacking is quickly headed toward becoming yet another engineering discipline. The "device physics" of standard parts are being worked out, the development tools are becoming more sophisticated, and the level of skill required to do interesting things is dropping. The annual International Genetically Engineered Machine competition which is intended, among other things, to increase the number of "biological parts" available, is getting high-scoring entries from high school students. The amount of hacking on biological substrates is increasing quickly, and will continue to do so.

The amount of creativity we will be seeing in this area inspires both hope and outright terror. Biological hacking has the potential to transform health care, address energy problems, mitigate climate change, and more. Or it could wreak environmental devastation and facilitate horrifying attacks by either individuals or governments. Carlson strongly advocates openness as the best policy for dealing with this technology. Only through openness, he says, will we develop the kind of economy we need to make the best use of this new technology while simultaneously understanding what others are up to and defending ourselves against mistakes and abuses. Trying to keep technology under wraps never works. Your editor might compare attempts to restrict biotechnology with governmental efforts to restrict encryption technology a generation ago.

Openness does not just mean freedom from regulatory interference, though; Carlson takes a long look at the possibility of creating a successful commercial ecosystem based on the open source model. At an abstract level, the idea looks compelling: it is not hard to see programming with nucleotides as being fundamentally the same task as programming with bits. A nucleotide is able to encode two bits rather than one, and the underlying processor is smaller, wetter, and smellier, but it's a program nonetheless. Given that tools for working with DNA are following a path similar to that of computers - they are rapidly becoming smaller, cheaper, and more powerful - there is a lot to be said for the creation of freely-licensed libraries based on genetic programs developed in garages and basements.

There are some efforts afoot to do exactly that. The BioBricks Foundation is working toward the creation of a set of freely-available biological components. Another initiative is Biological Open Source, appropriately known as BiOS. These efforts are promising, but there is a large problem looming - one which will be familiar to LWN readers.

That problem, of course, is patents. Genetic sequences are currently patentable in the US and elsewhere, so companies operating in this area are accumulating as many of them as possible. Things are quickly getting to the point where it is difficult to work commercially in biotechnology without running into patents held by others - patents which, often, cover fundamental natural phenomena. Carlson brings up some interesting history; it seems that the automotive and aviation industries both ran into this problem; in both cases, it got to where companies could not do anything because they were forever caught up in patent litigation. In both cases, in the US, the government intervened, forcing the creation of patent pools so that companies would stop suing each other and get back to doing interesting things with the technology.

Patent pools (like patents in general) favor large, established corporations over smaller companies. But it's the small companies which are the source of most innovation in any field. Carlson worries that the US is headed toward a situation where those companies cannot afford to exist and innovation will be strangled. An open-source-like approach to biotechnology might just be a way out of that situation.

But doing open source in this field, despite its similarities with software, is going to be hard. Software gets copyright protection worldwide; that makes it easy to use copyright licensing to create a legal regime where people (and companies) feel that it is in their interest to contribute. Genetic sequences have no such protection, so patents are the only way to go for anybody feeling the need to gain a degree of control over how a discovery is used. Copyleft-style patent licensing is possible, but it is more awkward, and, in any case, the high cost of obtaining patents creates a barrier to entry that does not exist for licensing based on copyrights. Lone biohackers working in their garages are not going to be contributing components to a community based on patents.

As a result of the different legal environment, open-source-like efforts in biotechnology must form their understandings under different terms than the software community uses. BioBricks must be placed in the public domain; the draft BioBrick Public Agreement - a contributor agreement, not a license - requires contributors to give "An irrevocable promise not to assert any property rights held by the Contributor over Users of the contributed Materials." BiOS, instead, is organized more like a patent pool with a fee to enter. Neither of these approaches is seen (by Carlson) as being ideal, but he also admits to being short of better ideas.

What may be required, in the end, is a new and different legal regime for biological discoveries. As Carlson notes, neither patents nor copyrights are mentioned by name in the US Constitution; they are legislative creations. Someday, maybe, a legislature rather more enlightened than those governing us now will find a way to foster open biotechnology development that works at all levels. It will be interesting to see whether the recent US District Court ruling throwing out genetic patents inspires any useful thinking in that direction.

One need not be a speculative fiction author to imagine a future world where the freedom to use, modify, and distribute biological code is (at least) as important as those freedoms applied to software running on silicon. We do not seem to be building a world which includes those freedoms, though; we do not even really have a good sense for what that world would look like. The biotechnology industry, it seems, is in need of its own personalities to fill the roles Richard Stallman, Linus Torvalds, and the many others who have helped to make free software work.

Comments (25 posted)

SCO loses again

It has been just over seven years since the "SCOSource" initiative showed its true colors and filed suit against IBM asking for $1 billion in damages. Much has happened in that time; the nature of the charges has evolved considerably, the price tag has increased, other companies have been pulled in, and SCO has gone into bankruptcy. In the process, SCO attack has, like a vaccination, served to strengthen the community's legal defenses considerably. It has been a wild show to watch.

One of the most surprising developments came when Novell abruptly announced that, in fact, it (and not SCO) was the owner of the Unix copyrights upon which much of the litigation was based. SCO responded with a "slander of title" lawsuit which has been a convoluted multi-year affair in its own right. But, on March 30, a jury in US District Court ruled that, in fact, no copyrights had been transferred to SCO and that Novell remained the owner of Unix.

This outcome was far from guaranteed. SCO's claims against IBM and the Linux community have been repeatedly shown to be without merit, but the Novell litigation was different. That was all based on a vague "asset purchase agreement," twice amended, which was seemingly written by lawyers who were not doing their job. What SCO purchased was truly unclear, and there was no telling what a jury might decide. Now, perhaps, we have some clarity.

It is tempting to see this ruling as the end of the story; rightfully it should be. But anybody who has watched this case for any period of time knows better. The SCO affair is kind of like a bad zombie movie; the plot is implausible, the acting is horrible (ask any of us who sat though all those "Chris and Darl show" conference calls back at the beginning), and, even though you know the good guys must win in the end, that obnoxious zombie just keeps coming back and ruining the party. SCO, which has just obtained a $2 million loan from Ralph Yarro (one of the original architects of this whole mess), may well appeal. Or perhaps it will find a willing buyer for the lawsuit out there. It does not seem like SCO to slip quietly into Chapter 7 bankruptcy at this point.

One should also remember that the Novell case was a sideshow; IBM is the main event. Losing the ownership of the copyrights clearly cannot help SCO's case, but it has long since been shown that there's nothing covered by those copyrights in Linux anyway. Much of the dispute was over code clearly owned by IBM, but which SCO claimed could only be distributed under the proprietary Unix terms. SCO might just choose to pursue its breach of contract (and related) claims. The fact that Novell claims the right to issue "get out of jail free" cards with regard to Unix licensing will complicate matters, and could prove fodder for extensive legal maneuvering in its own right.

Meanwhile, as long as some shred of SCO remains, one can only imagine that IBM will not be in a mood to forgive, forget, or drop its counterclaims. IBM has clearly put many millions of dollars into this fight; its chances of getting any of that back would appear to be approximately zero, but IBM is unlikely to want to leave SCO in a position to plan another attack.

SCO has long since ceased to be a threat to Linux, of course; at this point, most of us can afford to sit back and watch the remainder of the show. But we should not forget that SCO set out to kill our community. The attack was baseless and clownish, but it still created considerable uncertainty and expense. The sad fact of the matter is that, when companies lose in the market, they try to win in the courtroom, especially in the US. There will be other attackers, and at least some of those attackers will not make so many silly mistakes.

In the past, LWN has asked Novell to clarify what it intends to do with the Unix copyrights that, once again, it seems to clearly own. Those copyrights cannot have much - if any - commercial value at this point. It would make great sense for Novell to release all of that code as free software, optimally under a permissive license. Until that happens, those copyrights will be a temptation to those who think they can somehow use them in litigation. That zombie, at least, can be definitively killed.

Comments (9 posted)

Page editor: Jonathan Corbet
Next page: Security>>