LibBSD PowerPC motorola_shared BSP PCI Support

[Chris Johns]

On 21/10/20 2:52 am, Sebastian Huber wrote:
> On 20/10/2020 03:52, chrisj at rtems.org wrote:
> 
>> Tested on a MVME2700 (mvme2307) BSP:
>>
>> nexus0: <RTEMS Nexus device>
>> pcib0 pcibus 0 on motherboard
>> pci0: <PCI bus> on pcib0
>> pci0: <bridge, HOST-PCI> at device 0.0 (no driver attached)
>> pci0: <bridge, PCI-ISA> at device 11.0 (no driver attached)
>> pci0: <mass storage, ATA> at device 11.1 (no driver attached)
>> pci0: <mass storage, SCSI> at device 12.0 (no driver attached)
>> pci0: <bridge> at device 13.0 (no driver attached)
>> pci0: <network, ethernet> at device 14.0 (no driver attached)
> Does this mean that the legacy x86 PCI bus driver works on this PowerPC board?
> Are there no big-endian vs. little-endian issues?

It seems there are no issues with the limited testing I have performed. I have
not done extensive testing but the if_dc.c (tulip driver) does probe the bus and
the correct device/vendor id is returned. The fact there currently is no support
for the 21140 on the MVMEW2700 in the driver is different issue and not related
to this patch.

I had a detailed look at the few calls that are being made and everything seems
to work as expected. There is a lot of hardware in the PCI bridges on these
boards to handle endian mapping. The same hardware set up is being used as 4.10
and legacy stack combination and that works.

I suspect there may be changes needed in relation to resource allocations but I
thought those can be in follow up patches.

Chris