Re: [ECOS] does ecos support generating core files or kernel crashdumps?

["Reinhard JESSICH" <Reinhard.JESSICH@frequentis.com>]

Hello

We had similar needs for our development.
As you know we have implemented a process model and memory protection
for the PowerPC. We also implemented the first step of a core dump feature.

If an exception occures (dived by zero, illegal read/write, assert, ... ), we write
the exception text, the thread, the registers, a stackdump of the running thread
and if this is not possible a stackdump of the interrupt stack. The stackdump is
not a simple memory dump, but the context of each called function. This allowes
us to see the call graph and the exact position where the exception occures very
easily. We write the whole information to an unused memory, which is not erased
during reset. We can store as much core dumps as memory is available.
After reset we can read the core's via telnet or serail line.

We plan to store the complete context of the thread and its memory in
compressed form. This allowes us to see the content of variables in gdb. This would
be a real core feature. If it is too big for the unused memory, we can store it on a
flash.

We have submitted a patch for our process extension, but EcosCentric does not
accept it. If you are interrested, search in the archives for memory protection.
The patch was submitted by my colleague Thomas Binder.
You will find the core code there. In the meantime we have continued the
development, but this parts should be still the same.

Best regards,
   Reinhard


-- 
 Ing. Reinhard Jessich                Phone: +43/1/81150/2395
 Software Design                      Fax: +43/1/81150/3169
 Frequentis Nachrichtentechnik GmbH   A-1120 Wien, Spittelbreitengasse 34
 http://www.frequentis.com            eMail: reinhard.jessich@frequentis.com


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