Port-powerpc archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

NetBSD on Motorola PowerStack II and IBM RS/6000 B50



Hi.

I ported NetBSD to the Motorola PowerStack II during the last months.
The PowerStack II is a 300 MHz PPC604e PReP machine with OpenFirmWare. 
(The OFW doesn't supply Residual Data, so port-prep can't run on it.)

Butattachments / bus_space(9) is working and the kernel reaches the
point where it wants to mount the root-FS or tries to scan the SCSI
bus. This does not work due to non working PCI DMA. I am using the 
same generic PPC PCI DMA code that port-prep uses und it woks on 
PReP machines. (My development machine is PReP, a IBM RS/6000 43P-100.) 
I am at a dead end here, as I don't know how to debug this. Beyond the 
DMA issus, I am sure that the Interrupt handling code needs work. But 
it should work enough to get to single user or at least beyond the 
SCSI bus scan. 

If someone else wants to play around with this stuff:
http://www.unixag-kl.fh-kl.de/~jkunz/tmp/ofwboot.elf.gz
http://www.unixag-kl.fh-kl.de/~jkunz/tmp/netbsd.gz
http://www.unixag-kl.fh-kl.de/~jkunz/tmp/ofppc.tar.gz
The last is my complete src/sys/arch/ofppc from a NetBSD-1.6 source tree.

Also: This code has initial support for the IBM RS/6000 B50, a CHRP/RPA
machine based on a 375 MHz PPC604e, Motorola MPC106 and an OpenPic. 

The OFW of this machine seams to be broken in some ways. If I "load net"
the bootloader it terminates with an error. tcpdump indicates that the
machine is trying to load more data via TFTP then there is available. (?)
But if I start the bootloader with "go" it runs just fine. Real strange 
is the bootloader. The same bootloader works well on the PowerStack II 
and it loads the kernel just fine from a NetBSD NFS server. But on the 
B50 it chokes after a few NFS packets, but works well with a Tru64 UNIX
NFS server. 

There is an AMD PCnet PCI Ethernet in the B50. The firmware and the NetBSD 
kernel get a VendorID of 0x1023 = Trident instead 0x1022 = AMD from this 
chip. The chip has a big, bright AMD logo on it, so there is somthing 
strange going on. Other PCI devices are recognized correctly. 

If there is someone else with a B50 it would be nice to know if my machine
is the only one with this woes. 
-- 



tschüß,
         Jochen

Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/




Home | Main Index | Thread Index | Old Index