RTEMS on MVME230X
Eric Valette
valette at crf.canon.fr
Tue Jul 18 07:29:59 UTC 2000
Till Straumann wrote:
>
> Is there anybody out there running RTEMS on a MVME230X board (or any
> other powerpc-VME board)?
>
> I had severe problems with the bootloader: (Eric, are you still
> maintaining this?)
Yep.
> - the kernel `start.S' code corrupts the residual passed by the
> bootloader (see my last message)
Not here and not for the last guy that tested mvme2307...
> - the bootloader re-configures the pci devices without considering the
> Tundra-Universe
> (PCI-VME bridge) 's PCI slave image settings. This means that
> devices can be reconfigured to
> occupy an address range that is already in use by the Universe,
> i.e. a range where the Universe maps
> VME addresses.
Note that this is documeneted in the README.MVME2307 :-)
> - The USE_PPCBUG compiler/assembler switch should take care of
> disabling the MMU.
I gave up trying to use PPCBUG as it does not work with cache on...
> IMO, the bootloader should either not re-configure PCI or it should be
> able to properly set up or disable the Universe. In any case, it should
> at least _print_a_message_ when doing any re-configuration.
We can debate this for years. The debate has already taken place on the
linux-ppc mailling list. The reason for reconfiguring the PCI devices is
to compres the PCI address space so that a single BAT is usable and let
one available for VME address space :-)
Note that the guy how origaly wrote the boot loader even decided to
reconfigure the system using a CHRP map rather than a Prep MAP to have
contiguous memory MAP...
> Also, the user should be advised to manually disable the Universe's PCI
> slave if the reconfiguration is still done.
Again this is documented... So do not complain too much if you do not
read the doc...
--
__
/ ` Eric Valette - Canon CRF
/-- __ o _. Canon Development Europe Team Leader
(___, / (_(_(__ Rue de la touche lambert
35517 Cesson-Sevigne Cedex
FRANCE
Tel: +33 (0)2 99 87 68 91 Fax: +33 (0)2 99 84 11 30
E-mail: valette at crf.canon.fr http://www.crf.canon.fr
More information about the users
mailing list