March 1, 2001

The next Linux: A mobile version

Author: JT Smith

- by Jack Bryar -
Open Source Business -
OK, now we've been through the 2.4 hype. Get ready for the next Linux
wave. And it's likely to come from a source we should have

We've been through 2.4 and examined it to death. But now get ready
for next wave of Linux-hype. When it hits we should learn the

  • Is developing source code in a somewhat more private environment
    easier and more efficient than the long, painfully drawn-out tweak-fest
    that characterized 2.4's public evolution?
  • Does 2.4 make sense as the core of a mobile operating system?
  • Does Linus Torvalds do even better work when he gets paid?

We may all find out all this and more in the next few weeks.
According to Ziff Davis, Torvalds' employer is getting set to release
source code for its shaved-down, memory- and energy-efficient version of
Linux. According to the report, Transmeta should be posting the
source code for its Mobile Linux operating system to its Web site in the next two weeks or so. Although the exact release date isn't set, it should be before
they set up the display booth at Hall 13 Germany's massive CeBIT show in mid-March.

This could be interesting. Admittedly the Transmeta version might
not be exactly Alan Cox embodied
in silicon
as one satirical Web site suggested, or LinusTorvalds, the
portable edition, either. But the Transmeta design may be as close as
we've gotten to see what Torvalds thinks a stripped down version of the
operating system ought to look like.

Over the last month we've been tracking every iteration and
permutation of the tech and business world's response to 2.4. From my
perspective, I've talked about how the kernel's design now seems
solidly oriented to meet the needs of the big server and back end systems guys.
That's not to say there weren't plenty of fixes and patches to meet the
needs of hacker community that gave Linux its first popularity. But it
seemed like the priorities of the new kernel were focused elsewhere.

If the base of Linux enthusiasts felt somewhat left behind, they got
something, at least.

Developers of mini-Linuxes, those systems based on the operating system but shaved down to meet the memory and processor requirements of their equipment,
got even less. In the last couple of weeks between my day job and my
other gigs, I developed a fairly extensive list of appliance and
embedded systems developers working various compacted and embedded
versions of Linux. I was able to spare time to call about half of them.
Most said they'd get back to me. Few did. The one's I did talk to went
off the record, but their response was essentially: 2.4 is nice and
Linus is to be congratulated and we'll certainly look at it, and bring
its features to our platform, but there's not a lot there for

The challenges associated with paring down the kernel to meet the
requirements of mobile devices and computing appliances was taking
these developers in a rather different direction. The code they were
generating was certainly Open Source and recognizably Linux, but 2.4
wasn't a source of much excitement.

Linux for Transmeta may mark a change in that thinking. For one
thing, the Transmeta chip has been designed for portable, low power
applications, so the chip and the mobile Linux operating system together should be
very compelling to equipment developers who really don't want to
develop and maintain their own platform. And Transmeta, together with Mobile
Linux, is already being used in Internet appliances being developed by
Gateway and Hitachi among others. Indeed, Transmeta distributed the operating system
to its OEM customers some time back.

But the interesting thing about the Mobile Linux platform was that
it wasn't exactly open until now. Officials from the company told
Ziff-Davis that, of course, they were always going to release source
code for their platform, but they wanted to wait until they "were
satisfied" that the operating system was sufficiently complete that developers and
third parties could effectively test it and begin to suggest improvements.

"We want to put it out there in a way that's useful to people," said
a Transmeta official.

But that represents a rather different than the approach than the
Open Source community has been used to. Typical development work has
almost been painfully open, in an effort to keep the doors open to all
third-party developers. Closing the circle in the manner chosen by
Transmeta may be understandable; commercial applications developers
have taken this approach, and so have some embedded systems developers. But
the result has been to effectively lock other parties out in favor of a
selection of favored OEMs.

Whatever the political implications of that decision, the code
itself is also bound to be interesting, because Mobile Linux is a shaved down
version of the operating system that's been generated by the company where Torvalds works. Thus the memory management and power consumption strategies
embedded into Transmeta's version of Linux may be of interest to other
developers of mobile devices, whether they use Transmeta's platform or
not. And Mobile Linux does run on Intel as well as Transmeta hardware,
and it uses "the standard" Linux kernel.

And memory and power management issues are likely to extend well
past the world of portable devices. As the power consumption of big server
banks begins to be an issue in electricity starved locations like
California, look for an energy-conscious iteration of Linux to play an
important role in servers routers and other back-end equipment. Is this
the center of the next generation of Linux-based equipment?

NewsForge editors read and respond to comments posted on our discussion page.


  • Linux
Click Here!