[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RTEMS or1k platform port




Ralf Corsepius wrote:
> 
> Am Don, 2001-08-02 um 14.23 schrieb Ivan Guzvinec:
> > Hello,
> >
> > I'm currently maintaining RTEMS port for OpenRisc1000 platform and I
> > would need some help in order to get things started. Problem is that
> > developer that ported RTEMS to or1k platform, AFAIK, decided not to
> > pursue this project anymore.
> >
> > I would like to to see if there is anyone interested in helping with the
> > project.
> > As this e-mail is being written I'm trying to compile current snapshot
> > of RTEMS or1k port with very little success. Snapshot comes with no
> > configure, Makefile.in, aclocal.m4 scripts/files
> Really? They ought to be in.
> 
> The standard procedure to re-generate them is to run ./bootstrap from
> the toplevel of RTEMS' source-tree.

But this is not a "standard" snapshot or release. :(  

I did not know they had gotten far enough to have anything worth
distributing.  My best recommendation is to merge their or1k 
port (score/cpu/or1k), any libcpu code (lib/libcpu/or1k), and
BSP code (libbsp/or1k) into the current snapshot.  Then 
submit the result so at least it is in future snapshots.

> > and I can't figure the
> > right way of building them using autoconf, aclocal, automake,... as I
> > have almost no experience using these tools.
> Note that the snapshots are using autoconf-2.52 and automake-1.5 (Joel
> promised to put rpms on ftp://ftp.oarcorp.com)

But we don't know what version this fork was based upon.  It could
be buildable using older autotools. Who knows?

This is precisely why I push hard on submitters to get things into
the main tree.  It is easier for everyone. :(

> > If there is anyone interested, please take a look at the port source and
> > post some hints or suggestions on how to deal with this problem.
> If they were into RTEMS's source tree, this would not be a major deal.

Yep.

> However, IIRC, last time somebody was working on an or* port there were
> major problems with the or-tool-chain being based on some outdated
> version of the gnu-toolchain, and therefore the toolchain did not make
> it into RTEMS's standard toolchain packages.

They did not have support in a real gcc release so it took major patches.

> As a consequence of this probably nobody tried to build this port since
> then.

I am sure of it.

My recommendation stands -- merge the source into a current RTEMS snapshot
and submit it.   I would be happy to merge it.

Are there pre-built RTEMS or1k tools available?
 
> > or1k port of RTEMS can be found at
> > http://www.opencores.org/cores/or1k-new/rtems/
> > It should be a working kernel that runs on or1ksim simulator (link to
> > source and info can be fond on abovementioned URL).
> > Any questions regarding the or1k platform and simulator can, with great
> > probability, be answered by me so feel free to ask.
> 
> I am willing to assist you wrt. the autotools and integration into, but
> can't actually perform any actual testing, due to personal time
> constraints (I am a volunteer myself, being unpaid for RTEMS work and
> having other obligations.).
> 
> Ralf

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel at OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
Support Available                (256) 722-9985