I agree with Adam : the best is to try to port ucLinux (the NO_MMU and NO_F=
PU version of Linux) to a PIC18 target (the only one to let you manage the =
hardware stack).=20
Anyway don't forget the RAM space dedicated to the task descriptor is somet=
hing like 2Kbytes with ucLinux, so you need a PIC with a lot of RAM space a=
vailable to do it ! But if you have free time enough (one year probably) an=
d a solution to have a PIC with many Kbytes free in the RAM space you can t=
ry to start the job....
Let us know when you get a first result.
Xavier

> Message du 01/04/04 07:20
> De : M. Adam Davis=20
> A : PICLIST@MITVMA.MIT.EDU
> Copie =E0 :=20
> Objet : Re: [PIC:] Want to port Linux to PIC16f877...Help needed from all=
 you big guys
> Well, I could easily assume that from where you sent this email it was
> April 1st when you sent it, but then I'd have to assume that the entire
> world enjoys the American April 1st holiday, which would then expose me
> as the sort of ameri-centric snob I've worked so long to hide.
>=20
> So I'll assume this is a serious request for help and feed the trolls,
> er, I mean expose some of the issues you'll run into when you design
> Linux into a PIC.
>=20
> First of all, don't let anyone tell you it can't be done. Anything is
> possible, though there are some tradeoffs:
>=20
> First, the PIC has no MMU. Heck, it doesn't really even have a fully
> manipulable stack. Linux typically requires at minimum an MMU, but some
> smart folks somewhere on the internet have cobbled together an embedded
> MMU-less version of linux. Find it and learn from them. Also find out
> what kind of life they lead, and expect to live that life for the
> remainder of your development period.
>=20
> Second, the PIC doesn't really have a fully capable ANSI C compiler.
> Don't worry, linux won't compile on just any old ANSI compiler anyway.
> Go ahead and pull out your copy of C18 and run it on a few linux source
> files. When it runs into a problem, comment that line out. You can
> always sort it out later. Don't let anyone fool you about linking -
> just copy all the object files into one big file (cp *.o linux.obj, for
> instance) and let the PIC worry about it. Then find a PIC with a large
> enough code space to hold the entire thing. Be aware, though, that
> since you commented out most of the code in an earlier step, you may
> need a larger PIC when you get all that sorted out.
>=20
> Thirdly, After realizing the futility of compiling linux directly for
> the PIC, take up the bright idea of creating an emulator that runs on
> the PIC. You can pick an existing core which already has emulators in C
> (such as ARM, MIPS, x86, HAL9000, etc) or make up your own. Tha
> advantage of picking one that exists is you don't have to create the C
> compiler as well. Code up the emulator for the PIC, and create external
> address/data busses for flash and ram. Look up the PCI spec and try to
> emulate that bus standard - it'll save you loads of time adding
> peripherals later. The performance of the system may be low (a few
> hundred instructions per second of emulated code) but be content in the
> fact that you were able to shoehorn Linux into an 8-bit microcontroller.
>=20
> All it'll take, really, is a bunch of elbow grease and headlamp fluid.
> Good luck! Keep us informed of your progress, ok?
>=20
> -Adam
>=20
> P.S. While the post is mainly in jest, the third option really would
> work. The PIC, after all, is Turing Complete, and in theory can perform
> any logical sequence of steps. I just can't see the educational value
> in doing it...
>=20
> Vishal Changrani wrote:
>=20
> >hi,
> >hi to all you big guys there..man ..after reading your questions and ans=
wers i feel i m still in nursery :(
> >anyway heres my question:
> >I want to develop (or rather 'Port') the linux kernel on to the Pic16f87=
7. Now the reason I wish to do this is:
> >1. I have to present a research paper due on april 22nd on the topic Rea=
l Time operating Systems: Case Study Microcontrollers.
> >2. I think since both Linux and PIc are so very commercially successful =
in there respective domains that it would be good idea to combine them in s=
ome way.
> >3. I love linux and the PIC and couldnt think of a better topic.
> >About the Real time part I think i will save that till later that is aft=
er I have finsihing porting Linux to the PIC (if I am able to do it i.e.)
> >
> >Now i went through a book - Building Embedded Linux Systems ..and it say=
s that it isnt easy (not impossible ) though to port linux on to a 'less th=
an 32 bit' microcontroller or microprocessor. So I am stuck now! since Micr=
ochip doesnt offers a 32 bit microcontroller.
> >Please suggest me something! anything!
> >
> >Thankyou
> >Bye
> >Vishal
> >
> >
> >---------------------------------
> >Do you Yahoo!?
> >Yahoo! Small Business $15K Web Design Giveaway - Enter today
> >
> >--
> >http://www.piclist.com hint: To leave the PICList
> >mailto:piclist-unsubscribe-request@mitvma.mit.edu
> >
> >
> >
> >
> >
>=20
> --
> http://www.piclist.com hint: PICList Posts must start with ONE topic:
> [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads
>

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads