November 08, 2010, 7:36 AM — In the open source world, leading off an elevator pitch with a line like "six industry giants ARM, IBM, Texas Instruments, Samsung, ST-Ericsson, and Freescale founded a not-for-profit open source engineering organization..." is just bound to start an argument.
After all, there are elements within the open source community who eschew large corporate players with the same enthusiasm devoted to avoiding the plague. At best, even the least cynical among the community might flinch a bit at such a weighty pronouncement. Not just one industry giant, but six?
And yet, that's what Linaro is about, and they're proud of it.
At first blush, Linaro comes across as yet-another-embedded consortium trying to get their version of hardware/software standards more firmly placed in the marketplace. Superficially, that seems a fair assessment, but take another look and you've really got an organization that's waded into the trenches of embedded Linux and is trying to make sense of it all.
Linaro, which kicked off in June, is nearing the end of its first six-month engineering cycle--a self-imposed release period that should hopefully keep up with the rate of innovations in the mobile/embedded industry--and is about to show off some of its first wave of tools this week at the ARM Technology Conference in Santa Clara, CA.
Linaro's primary focus is around ARM-based technology, which might not seem too difficult, since it's just one platform. But according to Linaro CTO David Rusling, "ARM is not a monolithic technology." Rusling added the analogy that instead of thinking of ARM as an aircraft carrier, it should actually be thought of as more of a flotilla, sailing along the same general course.
The diversity within the ARM platform is indeed one of the reasons why Linaro was put together. Not only do embedded manufacturers have to work with multiple distros (MeeGo, Android, WebOS) and Linux kernel versions, but they also have to deal with the System on a Chip (SoC) fragmentation that comes from different SoC vendors bringing different approaches to kernel, power management, graphics, and multimedia technology.
Faced with that kind of hullabaloo around the ARM sector, Linaro was formed to go in and try to insert some sanity into the mix. This would, the theory goes, bring all players in the market better performance and a faster time to market for devices.
Linaro's approach to this problem is not to sit around and form committees to discuss the problems at hand.
"We're a not-for-profit engineering organization that actually gets things done," Rusling boasted. To do that, Linaro has formed active working groups tasked with providing solutions to the problems each group is focused on. Right now, there are three groups: toolchain, kernel consolidation, and the newest group, power management. The graphics and multimedia groups are about to launch soon.
The toolchain group is the primary area of interest for participants now, since these tools will form the framework for most of the work to follow, but the other groups are already seeing some action, too.
I asked Rusling about the power management group's involvement in the current discussion between the Android and Linux kernel teams, since power management is at the heart of the disagreement, and he replied that indeed Linaro is trying to set itself up as a mediator between the two sides.
Rusling believes that while the Android team may be on to something with their approach to the embedded power management problem, large vendors like Google (and even the actual members of the Linaro organization) need to learn that existing open source/free software projects don't like it when vendors come along with a huge bundle of shiny new code and say "here, accept this."
Instead, it's important that vendors learn to release code in increments to upstream projects, so changes are gradual and easier to parse and--ideally--get accepted. This is something that embedded vendors in particular need to embrace, since historically they will hold on to every bit of code until the last possible moment, fearing that an early release of software code will give competitors too much of a clue on intentions.
That seems to be changing, Rusling indicated.
"I have seen the ARM partners realigning around Linaro," he said, with some partners releasing architecture development code before the release of the actual hardware.
It's a cultural challenge, but Linaro seems ready to roll up its metaphorical sleeves and produce the results it needs to demonstrate that open source can be a positive force in embedded space.