[U-Boot-Users] boot control concept

[U-Boot-Users] boot control concept

[Andrew Wozniak]

Hello everyone,

Our ppc440gx-based project needs to develop a "boot control" mechanism
without any operator intervention. We have a customer PCI interface
where "special jumper settings" will indicate if an OS should be booted
from Flash or via TFTP.

We also desire that once U-Boot has been installed in Flash, the user
will ONLY be allowed to update OS and application components. This
insures that the user can always get to the U-Boot prompt in case they
somehow corrupted the Flash. We don't want to embed our code within
U-Boot because our code needs to be field upgradeable.

I was hoping to construct a U-Boot application (similar to hello_world
example) and do all of our "boot control" processing their. This
application would configure U-Boot to boot in one of the desired modes
based on special jumper settings.

It seems there are several ways to perform this. This app could use
native U-Boot calls to setup the environment before returning to U-Boot
where the OS load/boot would be completed. Would this work or is their a
better approach?

I've looked thru the docs and email list and have not seen anything done
like this. Any observations and suggestions about this are very welcome.

Thanks and best regards

[U-Boot-Users] boot control concept

[Wolfgang Denk]

In message <4214AD81.5010007@mc.com> you wrote:
> 
> Our ppc440gx-based project needs to develop a "boot control" mechanism
> without any operator intervention. We have a customer PCI interface
> where "special jumper settings" will indicate if an OS should be booted
> from Flash or via TFTP.

There are several examples in the U-Boot code how to handle this; see
for  example   the   LWMON,   R360   or   TRAB   boards.   See   also
board/lwmon/README.keybd (in German).

> We also desire that once U-Boot has been installed in Flash, the user
> will ONLY be allowed to update OS and application components. This

Just tell him in the documentation.  You  cannot  really  prevent  it
anyway.

> I was hoping to construct a U-Boot application (similar to hello_world
> example) and do all of our "boot control" processing their. This

Should be straightforward.

> It seems there are several ways to perform this. This app could use
> native U-Boot calls to setup the environment before returning to U-Boot
> where the OS load/boot would be completed. Would this work or is their a
> better approach?

Just call  you  standalone  application  as  one  step  in  the  boot
sequence.  This  is preferrable, as no changes to the U-Boot code are
required.

You  may  have  to   export   some   additional   functions   -   see
doc/README.standalone for instructions how to extend the jump table.

> I've looked thru the docs and email list and have not seen anything done
> like this. Any observations and suggestions about this are very welcome.

There are several examples - see above. Actually this is just another
are where U-Boot's flexibility  ourperforms  nearly  any  other  boot
loader I know of.

Best regards,

Wolfgang Denk

-- 
See us @ Embedded World, Nuremberg, Feb 22 - 24,  Hall 10.0 Booth 310
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
If something is different, it's either better or worse,  and  usually
both.                                                    - Larry Wall