[ECOS] eCos 2.0 for V850

[Wolfram Kattanek <wolfram dot kattanek at imms dot de>]

Hi,

I currently try to upgrade some V850 based systems to eCos 2.0. During my
experiments I noticed a couple of strange things and issues:

1. I've built the "official" toolset for eCos 2.0 (exactly following the
   instructions from http://sources.redhat.com/ecos/build-toolchain.html)
   (BTW: is anywhere a precompiled V850 toolset for eCos 2.0?).
   Using this toolset I tried to build and run the ecos tests for
   a default configuration. All tests explicitly using cyg_scheduler_start()
   (e.g. kclock1.c) are hanging (I suspect they're idling in idle_thread).
   If I turn off gcc optimization (via -O0) all runs well. If I enable
   CYGPKG_INFRA_DEBUG I can compile with -O2 and run the tests without any
   problems. I searched through gcc mailing lists but didn't find any
   particular problems regarding the V850 optimizer. Does anyone has an
   explanation for this weird behaviour?

   BTW: a quick experiment using an older toolchain (gcc 2.95.3) revealed
   that it is possible to build eCos 2.0 with them. First test results were
   encouraging too. Is a gcc 3.x in fact a prerequisite for eCos 2.0 (for
   all platforms)?

2. Did anyone test the eCos 2.0 distribution with a V850 CEB platform?
   Looking at the latest documentation under
   http://ecos.sourceware.org/ecos/docs-latest/user-guide/setup-v850-cebsa1.html
   I noticed that it seems to be a bit out of date (e.g. references to a
   "loaders" directory). Additionally I'm unable to build a gdb_module.bin
   (using the above mentioned toolchain) that is binary equivalent to the
   one from anoncvs (the latter one was last modofied 12 months ago!). Now
   I'm wondering if the V850 platform is actively supported and tested
   anymore?

3. The kernel tests kexcept1.c and except1.cxx have changed a little bit in
   eCos 2.0. As a result they're executed on V850 platforms although to my
   knowledge the appropriate HAL doesn't support e.g. data exceptions. This
   results in test failures (or hanging boards)...

Any hints and experiences regarding these problems are very welcome!

Wolfram
--
Wolfram 'L.A.' Kattanek     Institut fuer Mikroelektronik- und
Email:       LA@imms.de     Mechatronik-Systeme (IMMS) gGmbH     
Tel: +49 3677 / 6783-55     Langewiesener Str. 22
Fax: +49 3677 / 6783-38     98693 Ilmenau / Germany

-------------------------------------------------
This mail sent through IMP: imp.imms.de

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