Re: removing get_immrbase()??

[Kumar Gala <galak@kernel.crashing.org>]

On Apr 22, 2009, at 4:57 PM, Timur Tabi wrote:

> Kumar Gala wrote:
>
>> New nodes.  For example I've proposed a "local access window" node.
>> Once I add it I plan on changing code to use it.  This will break an
>> old device tree booting with the new kernel and I'm completely ok  
>> with
>> that.
>
> Are we having two different conversations?  I was talking about this
> block from your email:
>
>>> arch/powerpc/include/asm/cpm2.h:#define CPM_MAP_ADDR  
>>> (get_immrbase() +
>>> 0x80000)
>>> arch/powerpc/sysdev/cpm2.c:     cpm2_immr = ioremap(get_immrbase(),
>>> CPM_MAP_SIZE);
>>>       these two are related and seem like we could look for  
>>> "fsl,cpm2"
>>
>> That's okay, as long as you don't break compatibility with older
>> device trees that don't have that property, unless you can  
>> demonstrate
>> that these trees would never work with the current kernel anyway.
>
> Specifically, I was referring to this comment:
> 	
> 	these two are related and seem like we could look for "fsl,cpm2"
>
> And my point was that not all device trees have "fsl,cpm2" in their  
> CPM
> nodes.

Yes -- we are having two different conversations.  I've moved on from  
the specific issue of "fsl,cpm2" not existing in old device trees.   
I've moved to a more general statement about how I can solve some of  
the CPM2 related uses of cpm2_immr.  For example we assign cpmp based  
on cpm2_immr.  I could stop using cpm2_immr and solve this problem by  
adding a new device node for the comm-proc registers in the device  
trees at which point I'd break older .dts working with the kernel.

- k