Linux in Embedded OSs

Carnage4Life writes "ZDNet has an article on the viability of Linux as the future belle of embedded OSes. It quotes Linus as mentioning the fact that since license fees are free and developer support is relatively abundant, Linux is a prime candidate for startups creating Web appliances and the like. It lists Sony's, tiVo, Lineo, Transmeta, Intel and national Semiconductor as major industry players who are embracing embedded Linux. "

What does "Embedded" mean these days?

Let me preface this by saying that I've been an embedded developer for about 9 years.

Traditionally, the term "embedded" denoted a system with minimal (or no) UI. Code for such a system could not be developed natively so development was done on a host and 'installed' on the target via an emulator, debug port, PROMs, etc. Typical examples are heating/cooling systems, flight control systems and microwave ovens. In my case, they are traffic controllers and other associated equipment.

Things like Web Pads and Palm Pilots seem to blur the distinction between embedded and non-embedded systems. While code is still developed on a host (for the most part), many of these devices have quite sophisticated UI's (even GUI's). What do you think? Should they still be considered embedded or should there be some new term to describe them? Maybe I'm just old-fashioned, but I have a hard time thinking of anything with a GUI as embedded.

Windows CE

There is already a project to port Linux to devices which currently run Windows CE. They have already made excellent progress and Linux now boots on several MIPS and SH3 handhelds, such as the Casio E-100 and IBM WorkPad Z50.

See the Linux CE Project [linuxce.org].

ZDNot

Hmm, another filler article for mass consumption by slashdotters, and pleased to be clickink on the banner ads. Quick, somebody pass me the salt, this article tastes bad!

It may be flamebait, but I think ZDNet has become see-through in the lip-service it's paying to linux. I point you no further than Jesse Berst, who in his "berst alerts" went from "linux sucks - it'll never compete with windows!" to "I always said linux could be a contender" to "linux rulez" in a span of 5 months. Something ZDNet should try someday: balanced reporting. It's a novel concept I urge any reporter (slashdot included) to employ - *represent both sides equally and without bias*.

An important point was missed

Writing the code for an embedded product under Linux provides one huge benefit. The embedded OS does not depend on the plans of an outside company for porting to the next hardware platform you choose. And the application should port easily as long as you avoid, or isolate, hardware dependent code. For companies for whom the OS they deliver their product on is a commodity item, open source OSs offer them a measure of control over the future of their product. Any hardware they choose can run the OS if the port is worth the effort. If having Linux run on your next hardware platform is worth enough to pay a few good programmers to do it, Linux will run on that hardware. No one can say no to you.

Re:Windows CE

Same for NetBSD:

See NetBSD/hpcmips [netbsd.org] for MIPS based PDAs.

There's of course a SH3 port as well: NetBSD/SH3 [netbsd.org], but they currently don't support any SH3-handhelds (AFAIK).

Re:ZDNot

OK, at the risk of being flame-bait... may I point this out: has it ever occurred to you that Jesse Berst might have gone through a "conversion"? Is it so inconceivable for a person to believe in popular FUD against Linux and to speak out his belief? Is it so inconceivable that this same person may have found out eventually that there is more to Linux than the FUD would have people believe? You have to realize that reporters usually do NOT have the means nor the time to do a 100% accurate research about their subject. They go by what they judge to be an accurate picture based on the majority of information they collect -- if this majority happens to be tainted with FUD, it should not be surprising that their views show this too.

But after 5 months, if the reporter is worth his salt at all, he'd have dug deeper and perhaps discovered that Linux really isn't what the FUD depicts it as, and that there is a glimmer of truth to the claims made by Linux supporters.

I hate to say this, but Slashdot seems to be home to a lot of paranoid people who believes that popular media is a Big Satan that is totally clueless and always inaccurate. While it *might* be true that popular media is usually inaccurate, that doesn't justify the conclusion that *anything* from the media is not reliable. So please, people, before flaming this article to death, let's do some research and let's show some hard evidence of why this article is so lousy, as it's claimed to be. Pointing fingers at a reporter's reputation is not sufficient grounds to dismiss an article.

BSD vs GPL for embedded systems

The article didn't touch on licensing. Distributing linux in an embedded system all takes place under the GPL, correct? As the purchaser of a hypothetical LinuxPilot, I'd be entitle also to receive the source. But, would it be the source to the OS... or everything? I mean, if the code is all in binary form in a ROM, and the OS never exposes itself directly, where does the app stop and the OS start? Do the developers simply have to observe a "chinese-wall" model for developing, keeping the OS in one set of directories and the "app" in the other, and linking/stitching them together only as the last step before ROMming?

It seems that economic model for open source that makes a lot of sense in the embedded market is for the chipmakers, who are not software companies particularly, to port an open OS and then give it to their OEMs to make a more attractive packaged offering for their chips. But in that case, wouldn't some embedded systems makers who are in hotly competitive areas prefer to have an OS without the "gotta hand over the source to our modifications" feature? I.e, isn't there room for the BSD license here? Linux and the *BSDs are essentially the same from the chipmakers standpoint. Hire a small staff, do/maintain a port, give it away with the chips... what do their customers want?

I prefer the GPL myself, but it does have to compete against the BSD license and I want to explore/understand the implications, since the embedded systems area is going to be so big.

Too much OS?

I can't help but wonder of an embedded-linux will have problems similar to WinCE. A large OS crammed into a too-small package.

Is it wrong to wonder if, perhaps, we might be better off having different OSes that serve different purposes? Do we have to have only one, jack-of-all-trades OS?

Anyway, there is a reason that PalmOS beat WinCE, and it wasn't necessarily the normal Microsoft bugginess. It was the fact that PalmOS just does what is needed, and doesn't have a bunch of extra fluff brought down from bigger machines. Would an embedded-linux be any different?