[ECOS] eCos on 80186?

[ECOS] eCos on 80186?

[osrookie]

Hi all,

I've seen a couple of messages in the archives about
people wanting to run eCos on a 16-bit Intel
80186-based processor, but they weren't very recent. 
Do any of you know of any efforts to get this to work?
 Our company is very interested in doing this, but we
don't have the resources and know-how to pull it off
in a reasonable timeline.

I've seen that eCos relies heavily on gcc, which is
strictly a 32-bit compiler.  The options were to
either port gcc to 16-bit or to port eCos to a
different 16-bit compiler.  Could any experts out
there give a ballpark estimate of what it would take
to get either of these done?  Are there other easier
options?

I'm guessing that porting eCos to a different compiler
would be the easier route - any open-source 16-bit
compilers out there that you recommend?

Thanks, Dave.



		
__________________________________
Do you Yahoo!?
New and Improved Yahoo! Mail - Send 10MB messages!
http://promotions.yahoo.com/new_mail 

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] Re: eCos on 80186?

[Grant Edwards]

On Wed, Jun 23, 2004 at 08:24:14AM -0700, osrookie wrote:

> I've seen that eCos relies heavily on gcc,

True.  

> which is strictly a 32-bit compiler.

Not true.  

There are gcc back-ends for a number of 8 and 16 bit
architectures (68HC12, AVR, H8, etc.) 

That said, I don't recall there being an 8086 backend.  

What GCC doesn't support is Intel's messed-up far/near/huge
pointer scheme where pointers to one data type might be a
different size than pointers to another data type. That means
that only 2 of the 4 8086 memory models could be supported (all
16-bit pointers or all 32-bit pointers).  It would not be a
simple thing to support a memory model with 16-bit data
pointers and 32-bit function pointers.

> The options were to either port gcc to 16-bit

You mean port it to the 8086.

> or to port eCos to a different 16-bit compiler.  Could any
> experts out there give a ballpark estimate of what it would
> take to get either of these done?  Are there other easier
> options?
> 
> I'm guessing that porting eCos to a different compiler would
> be the easier route - any open-source 16-bit compilers out
> there that you recommend?

At one point in time, there was somebody who claimed to have
ported eCos to another 32-bit compiler, but I don't know if
the work was ever made public...

-- 
Grant Edwards
grante@visi.com

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Re: eCos on 80186?

[John Newlin]

>At one point in time, there was somebody who claimed to have
>ported eCos to another 32-bit compiler, but I don't know if
>the work was ever made public...
>
>  
>
We have Redboot compiling and running with Tensilica's compiler for the 
Xtensa architecture.  Their compiler uses the gcc front end, and then a 
different back end.  I tried compiling a simple ECOS application, and it 
linked ok, but I haven't tried running it yet, so I'm not sure if all of 
the constructor initialization works.

At some point in time we would like to contribute back our changes.

-john





-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

RE: [ECOS] eCos on 80186?

[Paul D. DeRocco]

> From: osrookie
>
> I've seen a couple of messages in the archives about
> people wanting to run eCos on a 16-bit Intel
> 80186-based processor, but they weren't very recent.
> Do any of you know of any efforts to get this to work?
>  Our company is very interested in doing this, but we
> don't have the resources and know-how to pull it off
> in a reasonable timeline.
>
> I've seen that eCos relies heavily on gcc, which is
> strictly a 32-bit compiler.  The options were to
> either port gcc to 16-bit or to port eCos to a
> different 16-bit compiler.  Could any experts out
> there give a ballpark estimate of what it would take
> to get either of these done?  Are there other easier
> options?
>
> I'm guessing that porting eCos to a different compiler
> would be the easier route - any open-source 16-bit
> compilers out there that you recommend?

I don't know any open-source ones. However, Borland (or whatever their name
is now) had a very configurable compiler that worked better on 16-bit, in my
opinion, than Microsoft's. If you can find an old copy of Borland C++
Builder or earlier, you could see what happens when you try to feed some
eCos code through it. My guess is that you'll have the biggest headaches
with the makefiles, rather than the C.

--

Ciao,               Paul D. DeRocco
Paul                mailto:pderocco@ix.netcom.com


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] Re: eCos on 80186?

[Grant Edwards]

On Wed, Jun 23, 2004 at 10:34:52PM -0400, Paul D. DeRocco wrote:

> If you can find an old copy of Borland C++ Builder or earlier,
> you could see what happens when you try to feed some eCos code
> through it. My guess is that you'll have the biggest headaches
> with the makefiles, rather than the C.

Firstly, why would changing C compilers cause make to break?
The compiler flags are completely user-configurable with the
config tool, so I see no reason why there should be any problem
with makefiles.

Secondly, there are going to be problems with the C -- and with
the C++.  eCos uses a number of GCC and binutils extensions.

Ones that come readily to mind are:

 * Prioritized constructors for C++ objects.

 * "section" attributes and section sorting used to place
    objects located in different source files into contiguous
    tables in memory in specified order.

 * Link-time garbage collection used to discard unused
   functions and variables.

You can probably find work-arounds for such things, but it's
going to be a bit of work.

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Re: eCos on 80186?

[Chris Gray]

On Thursday 24 June 2004 05:44, Grant Edwards wrote:
> On Wed, Jun 23, 2004 at 10:34:52PM -0400, Paul D. DeRocco wrote:
> > If you can find an old copy of Borland C++ Builder or earlier,
> > you could see what happens when you try to feed some eCos code
> > through it. My guess is that you'll have the biggest headaches
> > with the makefiles, rather than the C.
>
> Firstly, why would changing C compilers cause make to break?

Borland had their own version of make, and from what I remember (from my own 
80186 days) it was pretty restricted even compared to standard un*x version 
of make back then. It wouldn't have a snowflake in hell's chance of running a 
makefile written for GNU make. So OP isn't going to be able to use Borland C+
+ Builder as such, only the compiler component (invoked from GNU make).

-- 
Chris Gray                      /k/ Embedded Java Solutions
Embedded & Mobile Java, OSGi        http://www.kiffer.be/k/
chris.gray@kiffer.be                         +32 3 216 0369


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

RE: [ECOS] Re: eCos on 80186?

[Gregg C Levine]

Hello from Gregg C Levine
What about using the tools for 16 bit arches from DJ Delorie? Oh he
claims they are basically unsupported, I don't know if they won't
work.
-------------------
Gregg C Levine hansolofalcon@worldnet.att.net
------------------------------------------------------------
"The Force will be with you...Always." Obi-Wan Kenobi
"Use the Force, Luke."  Obi-Wan Kenobi

> -----Original Message-----
> From: ecos-discuss-owner@ecos.sourceware.org [mailto:ecos-discuss-
> owner@ecos.sourceware.org] On Behalf Of Chris Gray
> Sent: Thursday, June 24, 2004 2:17 AM
> To: Grant Edwards; Paul D. DeRocco
> Cc: eCos Discuss
> Subject: Re: [ECOS] Re: eCos on 80186?
> 
> On Thursday 24 June 2004 05:44, Grant Edwards wrote:
> > On Wed, Jun 23, 2004 at 10:34:52PM -0400, Paul D. DeRocco wrote:
> > > If you can find an old copy of Borland C++ Builder or earlier,
> > > you could see what happens when you try to feed some eCos code
> > > through it. My guess is that you'll have the biggest headaches
> > > with the makefiles, rather than the C.
> >
> > Firstly, why would changing C compilers cause make to break?
> 
> Borland had their own version of make, and from what I remember
(from my own
> 80186 days) it was pretty restricted even compared to standard un*x
version
> of make back then. It wouldn't have a snowflake in hell's chance of
running a
> makefile written for GNU make. So OP isn't going to be able to use
Borland C+
> + Builder as such, only the compiler component (invoked from GNU
make).
> 
> --
> Chris Gray                      /k/ Embedded Java Solutions
> Embedded & Mobile Java, OSGi        http://www.kiffer.be/k/
> chris.gray@kiffer.be                         +32 3 216 0369
> 
> 
> --
> Before posting, please read the FAQ:
http://ecos.sourceware.org/fom/ecos
> and search the list archive:
http://ecos.sourceware.org/ml/ecos-discuss


--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] Re: eCos on 80186?

[Grant Edwards]

On Thu, Jun 24, 2004 at 08:17:18AM +0200, Chris Gray wrote:

> > Firstly, why would changing C compilers cause make to break?
> 
> Borland had their own version of make, and from what I
> remember (from my own 80186 days) it was pretty restricted
> even compared to standard un*x version of make back then.

Right, but we're talking about using a different compiler, not
a different make.

> It wouldn't have a snowflake in hell's chance of running a 
> makefile written for GNU make. So OP isn't going to be able to
> use Borland C+ + Builder as such, only the compiler component
> (invoked from GNU make).

Why would one even consider not using Gnu make?  ecosconfig
already creates gnu makefiles automagically, and there's no
reason at all to try to port the project to an incompatible
version of make.

-- 
Grant Edwards
grante@visi.com

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Re: eCos on 80186?

[Mark Grosberg]

On Thu, 24 Jun 2004, Grant Edwards wrote:

> ELKS uses the bcc compiler and ld86 linker from the dev86
> package.  Googling for "dev86 bcc" should provide more info.

What about the OpenWatcom compiler; the commercial Watcom compiler has
gone open source. It produces 16-bit as well as 32-bit code. Also, like
GCC it is quite flexible for use in embedded development (albeit,
incompatibly). See http://www.openwatcom.org/ for more info.

The compiler targets several architectures as well as generating excellent
code (sometimes better than GCC, sometimes worse). It might be interesting
to try and port eCos to to it.

OpenWatcom can also generate code for Alpha (not maintained?) and PPC, I
think some other architectures may be in there as well (even IBM 370,
IIRC).

Cheers,
Mark G.


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Re: eCos on 80186?

[Grant Edwards]

On Thu, Jun 24, 2004 at 02:12:17PM +0200, Neundorf, Alexander wrote:

>> What about using the tools for 16 bit arches from DJ Delorie?
>> Oh he claims they are basically unsupported, I don't know if
>> they won't work.
> 
> Maybe you could check out which compiler is used by the ELKS
> project: http://elks.sourceforge.net/ This is intended to run
> a linux-like kernel on 80x86 with x<3

ELKS uses the bcc compiler and ld86 linker from the dev86
package.  Googling for "dev86 bcc" should provide more info.

-- 
Grant Edwards
grante@visi.com

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss