Updated MPC5200B support (pcm030, kp-mcb2, ...)
Carsten Schlote
c.schlote at konzeptpark.de
Thu Apr 22 06:10:56 EDT 2010
Hi,
yesterday I spent some time to get BareBox up and running on MPC5200B
again. Finally I got it working on our hardware. Not yet tested for
phytec pcm030, but I'm sure that without some fixes it won't work out of
the box.
I found and hacked/fixed the following problems so far:
- cfi_flash driver has problems with big-endian. Some byte swapping is
done in readb()/writeb() at HW access layer and also at several other
places in the CFI driver. Swapping things twice isn't exactly what the
driver and the flashdevice expects and needs a lot of more cleanups.
- Detection of flash/bank layout was hacked to workaround 'faulty'
values read from flash device (might be also a chip or HW design
problem, I need to check that in more detail)
- CFI flash buffer writes are broken with big-endian as well
- Some problems and old bugs from the u-boot-v1 era found in cpu_init.c.
Fortunately the HW did the right things although some register setups
were broken.
- Missing headers for arch/ppc to compile the liblzo code.
(unsigned.h ?)
Now with network, flash and basic debugging outputs available I would
like to port some more drivers and at least RO filesystems (fat, jffs2,
ext2/3/4) for MPC5xxx platform and replace the stone-age crappy u-boot
asap.
Just in case - Is there any on-going or even experimental work for MPC I
could start with?
--
Carsten Schlote <c.schlote at konzeptpark.de>
konzeptpark.de
More information about the barebox
mailing list