Re: [rfc][patch] driverfs multi-node(board) patch [2/2]

[Matthew Dobson <colpatch@us.ibm.com>]

Patrick Mochel wrote:
> Matt,
> 
> I have some comments about the structure of the code, but those will come 
> in another email..
Cool...  Any/all comments are definitely welcome...

>>[root@elm3b79 devices]# tree -d root/sys/
>>root/sys/
>>|-- node0
>>|   `-- sys
>>|       |-- cpu0
>>|       |-- cpu1
>>|       |-- cpu2
>>|       |-- cpu3
>>|       `-- memblk0
>>|-- node1
>>|   `-- sys
>>|       |-- cpu4
>>|       |-- cpu5
>>|       |-- cpu6
>>|       |-- cpu7
>>|       `-- memblk1
>>|-- pic0
>>`-- rtc0
> 
> 
> Shouldn't nodes (or, erm, boards) be added as children of the root? 
Um, yes!  I don't quite know how, though...  I call sys_register_root() 
for each of the nodes...  That call seems to parent them under the 
root/sys directory...  How can I change that?

> Aren't all types of devices present on the various boards (PCI, etc)?
Yes...  I have some patches that I'm working that will put PCI busses 
and devices into the topology infrastructure (both in-kernel & via 
driverfs).  Again, this is just a first pass of what I'd like to see... ;)

Cheers!

-Matt


> 
> 	-pat
> 
>