Eric Valette valette at
Mon May 14 12:36:33 UTC 2001

"D. Peter Siddons" wrote:
> Hi Eric,
>     Thanks very much for pointing out the info in the various README's.
> I know, always rtfm! Oh well. Anyway, with that and
> a pointer from Joel Sherrill about modifying the mvme2307.cfg file,
> I got a boot image and was able to load it. Here's what happened for
> the 'paranoia' sample:

Stuff deleted.

> and then nothing. THe processor activity light goes out and it just
> hangs. I have tried using the PCI and VME chip configurations given
> in the README's, and also the suggested configuration from the Motorola
> manual for a PREP setup. It gives the same result. Am I doing something
> wrong?

If I remember something correctly, there is a conflict for the MVME
mapping and usual Prep PCI mapping  one's.
At least this was the case for MVME2307. Did you perform what is
mentionned below? 

--------------------------- README.MVME23000------------------------
This BSP was adapted from Eric Valette MCP750 Generic motorola
port to MVME2300 by Jay Kulpinski <jskulpin at>.
In other to work correctly, the Tundra Universe chip must
be turned off using PPCBug as explained below.

The Tundra Universe chip is a bridge between the PCI and VME buses.
It has four programmable mapping windows in each direction, much like
the Raven.  PPCBUG lets you specify the mappings if you don't want
to do it in your application.  The mappings on our board, which may
or not be the default Motorola mappings, had one window appearing
at 0x01000000 in PCI space.  This is the same place the bootloader
code remapped the Raven registers.  The windows' mappings are 
very likely to be application specific, so I wouldn't worry too
much about setting them in the BSP, but it would be nice to have
a standard interface to do so.  Whoever needs that first can 
incorporate the ppcn_60x BSP code for the Universe chip. :-)

These options in PPCBUG's ENV command did the job:

VME3PCI Master Master Enable [Y/N] = Y?
PCI Slave Image 0 Control                = 00000000?   <-----
PCI Slave Image 0 Base Address Register  = 00000000?
PCI Slave Image 0 Bound Address Register = 00000000?
PCI Slave Image 0 Translation Offset     = 00000000?
PCI Slave Image 1 Control                = 00000000?   <-----
PCI Slave Image 1 Base Address Register  = 01000000?
PCI Slave Image 1 Bound Address Register = 20000000?
PCI Slave Image 1 Translation Offset     = 00000000?
PCI Slave Image 2 Control                = 00000000?   <-----
PCI Slave Image 2 Base Address Register  = 20000000?
PCI Slave Image 2 Bound Address Register = 22000000?
PCI Slave Image 2 Translation Offset     = D0000000?
PCI Slave Image 3 Control                = 00000000?   <-----
PCI Slave Image 3 Base Address Register  = 2FFF0000?
PCI Slave Image 3 Bound Address Register = 30000000?
PCI Slave Image 3 Translation Offset     = D0000000?        

  /  `                   	Eric Valette - Canon CRF
 /--   __  o _.          	Canon Development Europe Team Leader
(___, / (_(_(__         	Rue de la touche lambert
				35517 Cesson-Sevigne  Cedex
Tel: +33 (0)2 99 87 68 91	Fax: +33 (0)2 99 84 11 30
E-mail: valette at

More information about the users mailing list