Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

[P]
Real Time Linux Patent

By Maniac in Technology
Thu Aug 09, 2001 at 02:59:18 PM EST
Tags: Software (all tags)
Software

A recent press release from Lineo announces an agreement with Lineo and FSM Labs on patent licensing related to Linux real time. There is a big discussion on RTAI related mailing lists, but I expect there are broader issues than this specific case to be discussed.


For those who don't know what all the fuss is about, here's a brief summary of it.
  • U.S. Patent No. 5,995,745, titled ADDING REAL-TIME SUPPORT TO GENERAL PURPOSE OPERATING SYSTEMS was issued November 30, 1999. This patent describes use of patches to drivers and a special scheduler that runs real time tasks w/ Linux as the "idle task".
  • The Real Time Application Interface (RTAI) uses basically the same method as described in this patent. RTAI is released under the GPL.
  • There is an Open RTLinux Patent License that basically allows GPL software to use the patent w/o royalty (or use FSM Labs RTLinux & pay appropriate fees).
  • The recent announcement by Lineo that they have a license agreement with FSM Labs that covers Lineo's customers (using Embedix Real Time).
As a possible commercial user of a real time Linux, it appears I have the following options:
  • Go to FSM Labs and use RTLinux.
  • Go to Lineo and use Embedix Real Time.
  • Use RTAI and assume I'm covered.
  • Ignore these particular real time solutions and use something else.
Our current system implementation is leaning heavily on the last option. Our application is quite large and we can't live with the restrictions imposed by either RTLinux or Embedix Real Time. The real problem occurs if we get into trouble and need to use a higher performance option for a small part of our system. We have a current support contract in place w/ Lineo, we should be covered for our current customer, but need to address the future as well.

A few questions come to mind at this point:

  • Are there other areas where software patents can cause legal problems for developers?
  • Should we (as a large company) fund an effort to get this patent invalidated?
  • Has FSM Labs covered itself properly with respect to the GPL?
My answers to those three questions are... likely, not at this time, and "not sure". Other feedback on this is appreciated.

Sponsors

Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure

Login

Related Links
o recent press release from Lineo
o Lineo
o FSM Labs
o Real Time Application Interface
o Open RTLinux Patent License
o Also by Maniac


Display: Sort:
Real Time Linux Patent | 24 comments (13 topical, 11 editorial, 0 hidden)
Don't make me think for you... (3.40 / 5) (#1)
by taruntius on Wed Aug 08, 2001 at 02:41:59 PM EST

Your company needs to figure out its strategy on its own.




--Believing I had supernatural powers I slammed into a brick wall.
Why not use it? (4.50 / 2) (#2)
by wiredog on Wed Aug 08, 2001 at 02:47:02 PM EST

The only requirement is that you release your source under the GPL. If you are doing an app that only one or two customers will use, GPL is OK, most people will never look at the code anyway. If you are doing it for, say, GM or Ford, use QNX. Higher license fees, but you can do closed source. Under the GPL you can still charge money for your work. Heck, I'd like to see more companies do it this way.

I strongly suspect this is a defensive patent. They had something relatively new and they patented it so that no one else could patent it and sue them. See Rambus for details of nasty patents. Or GIF.

There's lots of stuff at O'Reilly about bad patents. Check it out.

If there's a choice between performance and ease of use, Linux will go for performance every time. -- Jerry Pournelle

Unfortunately, no... (4.00 / 2) (#3)
by trhurler on Wed Aug 08, 2001 at 03:00:12 PM EST

As far as I can tell, the owner of that patent intends to make sure nobody makes money off his "great idea" except him. Nevermind that it is absurdly obvious to anyone who knows anything about how operating systems really work, and nevermind that there's prior art from(at least) Sun and a few research labs that had RT capable versions of unix back in the early 90s.

Sure, he offered a free license to anyone using the GPL, but in the real time market, the GPL is as good as a death sentence for any kind of profits; anyone who really needs RT does not need a support hotline or a box with an instruction manual.

The truth is that this patent is worse than the one click, because at least one click is a trivial convenience feature; good old Victor is preventing legitimate OS research and development efforts solely on the basis of the fact that someone other than him might profit by them, and he's doing it on the basis of nothing more than having had an idea that probably occurs to anyone taking a Unix-centric OS theory class at least once during a semester.

That said, I could care less from a personal standpoint, because realtime versions of unix are a stupid idea anyway. If you really need realtime, you won't be using unix, and if you don't, then quit wasting time with "soft realtime" systems that half-ass work and just get yourself an extra machine to run those critical jobs on. I occasionally disagree quite strongly with the man, but this is one place where Linus Torvalds has it right. (Now, if only he'd admit that GNU software is a bunch of shit and get some of his countless minions to work on an alternative toolchain and a port of the BSD userland...)

--
'God dammit, your posts make me hard.' --LilDebbie

[ Parent ]
Gack... (5.00 / 1) (#4)
by beergut on Wed Aug 08, 2001 at 03:45:47 PM EST

(Now, if only he'd admit that GNU software is a bunch of shit and get some of his countless minions to work on an alternative toolchain and a port of the BSD userland...)

It would be great to have hands and eyes involved in that effort, but something here reminds me of that old chestnut about infinite monkeys...

i don't see any nanorobots or jet engines or laser holography or orbiting death satellites.
i just see some orangutan throwing code-feces at a computer screen.

-- indubitable
[ Parent ]

Quickly (none / 0) (#7)
by trhurler on Wed Aug 08, 2001 at 04:20:11 PM EST

Go and read my story near the top of the queue on the test machine(bubba.kuro5hin.org). You'll have to log in, same password and username as here. Tell your friends:)

Oh, and yeah, you're right. Geh. Infinite monkeys suck.

--
'God dammit, your posts make me hard.' --LilDebbie

[ Parent ]
Real-time unix (5.00 / 1) (#20)
by Pseudonym on Thu Aug 09, 2001 at 11:57:06 AM EST

That said, I could care less from a personal standpoint, because realtime versions of unix are a stupid idea anyway.
That won't always be true. You may not need to run an ATM router at the moment. You may not need to serve streaming media at the moment. Eventually you will, and then you'll want real-time support from your server OS. You may not need to mix 12-channel audio or multiple video sources in real time at the moment, but eventually you will and then you'll want real-time support from your desktop OS.

Right now, probably the only kinds of real-time support you really need is something like burning CDs. You can hack around it by putting the job at a high priority, making sure you time the job so your machine won't start running any cron jobs in the mean time, then not touching anything until it's done. Eventually, people will learn that things need not be this way.

If you really need realtime, you won't be using unix, and if you don't, then quit wasting time with "soft realtime" systems that half-ass work and just get yourself an extra machine to run those critical jobs on.
You tell 'em! And 640kb should be enough for anybody, dammit!



sub f{($f)=@_;print"$f(q{$f});";}f(q{sub f{($f)=@_;print"$f(q{$f});";}f});
[ Parent ]
I disagree (none / 0) (#21)
by tzanger on Thu Aug 09, 2001 at 05:06:32 PM EST

Sure, he offered a free license to anyone using the GPL, but in the real time market, the GPL is as good as a death sentence for any kind of profits; anyone who really needs RT does not need a support hotline or a box with an instruction manual.

I must disagree. For the developers using the RT extensions, yes, they have no (little) need of a support line or instruction manual. However, if I, as a developer, make GPL software with the RT extensions, my customers will require support. And the GPL doesn't restrict me from selling software at all -- all I need to do is make the source available upon request. I can guarantee you right now that any of my embedded software that I design (I do this for a living) has no benefit to any of my customer base, who just want a unit that works. My competition may like a copy of the source but that is different again.



[ Parent ]
Realtime Unix already here. (none / 0) (#24)
by zak on Fri Aug 10, 2001 at 09:21:17 PM EST

> realtime versions of unix are a stupid idea anyway.
This is correct, since at least _some_ commercial Unixes already have sufficient RT features to compete with the likes of VxWorks and QNX (I am speaking of Solaris and IRIX). This means finely-grained kernel and process threading, high resolution timers (10's of nanosec), guaranteed sub-microsecond response (scheduling), RT support in STREAMS etc. Linux and BSD have to resort to backward solutions like RT microkernels and running as an application, since they _cannot_ properly support the realtime paradigm. I have seen IRIX and Solaris used in hard-realtime environments, with great success. I think that if your project is not constrained with your hardware choice and money is no object (as is usually the case for many hard-RT projects, at least moneywise), there should be very little reason not to choose a real operating system instead of a mongrel halfbreed (especially one which chooses to be burdened with ridiculous patents).

[ Parent ]
I don't understand what your problem is (none / 0) (#6)
by delmoi on Wed Aug 08, 2001 at 04:16:11 PM EST

It isn't very clear. You claimed that RT linux could be used in GPLd software without paying a license fee. So what exactly is the problem?

And you didn't give us very much information to to go on teither. Like, what is it your actualy trying to do, etc.
--
"'argumentation' is not a word, idiot." -- thelizman
More from Lineo (3.00 / 1) (#9)
by Maniac on Wed Aug 08, 2001 at 05:18:32 PM EST

There's a pretty long message posted at Linux Today from David Beal that provides some reasoning why Lineo decided to put the license agreement in place. A specific quote is key
... we must provide assurance to our customers that there is no legal risk in the use of this technology.
Where "this technology" is generally the RTAI solution and specifically the Embedix Real Time product from Lineo.

As a commercial company, not able to release all our software under the GPL, this gives us an option for delivery of our product to our customer (at least in this case). I am still concerned about other areas where patents may exist that restrict our rights as well. None of the replies to this point appear to identify any.

Could be a Good Thing (3.00 / 1) (#10)
by turtleshadow on Wed Aug 08, 2001 at 10:18:11 PM EST

This could be a very good way of Linux getting more mainstream.
A vendor who licensed you and now can be sued if RT kernel burps and the safety widget goes kaput is something the corporate suits look for in a project.
Same situation with A GPL author that is unfortunatly probably earning squat and community praise is something the suits don't like.
Secondly FSM and Lineo seem to have produced something needed not just patent squatted. The community is about 3 years from this at most. That's the day I fear.

As companies exist to make money; the model is that they will provide what the market needs at a cost the market can bear.
Im surprised their licensing is constricting you. I'd hope to think they'd want many of your widgets out there, ie license revenue, and would work to meet your requirements so you can actually sell your product. Perhaps its a matter of volume & size. As M$ is not in this market yet I'd rather see someone else get incumbent quickly, FSM and Lineo don't seem the evil empire types.

My two cents

Turtleshadow

Prior Art (4.50 / 2) (#22)
by cjensen on Thu Aug 09, 2001 at 05:41:08 PM EST

It seems to me prior art exists.... Here's an excerpt from an email I wrote about the Embedded Systems Conference in September 1996:
RT/Win was showing their system for allowing Microsoft Windows and vxWorks to both share a single processor. Windows runs as a task under vxWorks. New this year is Win95 support (WinNT is not supported, nor did they mention any future support).
The filing date for the patent in question is Nov 10 1997... so it seems to me RT/Win is prior art. I'm genuinely interested in anyone providing a reason why I'm wrong.

Here's a link to an EDN Magazine article about the same product from 1995: EDN. You'll have to scroll down the page to get to RT-Win.

I would note that the RT/Win product appears to not handle the "interrupts in the nonrealtime OS" problem well, and that FSM's patent does include a trivial enhancement to guarantee that nonrealtime OS interrupts do not interfere with the realtime threads.

Also, I haven't been able to determine if RT/Win and FSMLabs are the same organization. Even if they are, I didn't think you could file a patent after publicly disclosing a device.

Sounds like Intel's original plan for NSP (none / 0) (#23)
by John Miles on Thu Aug 09, 2001 at 10:18:05 PM EST

The idea behind the Native Signal Processing architecture would have been for Win95 to run as just another "task" scheduled by Intel's own ring-0 multitasking kernel. NSP would have turned a PC into a real-time DSP platform capable of acting as a MIDI synthesizer, modem data pump, or 3D audio spatializer.

Users, of course, would have had to upgrade to the latest and greatest Intel processor if they wanted their Windows apps to run reasonably well under NSP... and PC gaming would have been catapulted back to the Zork days. (Not that that would necessarily have been a Bad Thing, but just try to imagine a Winmodem running on a 486.) NSP was one of the industry's all-time worst ideas, and Microsoft, for all their faults, was quite justified in knifing that particular baby.

Intel first floated the NSP idea in 1995, before Win95 was even released. It's hard to imagine any of the RTL patent claims surviving a comparison with prior work by Intel. Perhaps some good can come from the evil that was NSP...

For so long as men do as they are told, there will be war.
[ Parent ]

Real Time Linux Patent | 24 comments (13 topical, 11 editorial, 0 hidden)
Display: Sort:

kuro5hin.org

[XML]
All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
See our legalese page for copyright policies. Please also read our Privacy Policy.
Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
Need some help? Email help@kuro5hin.org.
My heart's the long stairs.

Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!