Re: [PATCH 5121 pci 1/3] powerpc: 83xx: pci: Remove need for get_immrbase from mpc83xx_add_bridge.

[John Rigby <jrigby@freescale.com>]

Grant Likely wrote:
> On Wed, Aug 13, 2008 at 10:06 AM, John Rigby <jrigby@freescale.com> wrote:
>   
>>> On Thu, Aug 07, 2008 at 11:36:25AM -0600, John Rigby wrote:
>>> Can you use something like 'fsl,primary-pci-bridge' instead?  'primary'
>>> is a little too generic for my taste.  Also, the purpose of identifying
>>> one of the PCI bridges as primary should be documented (This is me
>>> pushing against encoding Linux internal implementation details into the
>>> device tree, I suspect that 'primary' doesn't belong in the device tree
>>> at all).
>>>
>>>       
>> Ok, I got the primary idea from sam440ep.dts, I'm willing to do something
>> different.
>>
>> I have thought about adding an is_primary argument to mpc83xx_add_bridge
>> like fsl_add_bridge has and make the callers figure out which is primary.
>>
>> The simple case is the platform that have only one bus:
>>   for_each_compatible_node(np, "pci", "fsl,mpc8540-pci")
>>       fsl_add_bridge(np, 1);
>>
>> Callers with multiple bridges do something like this:
>>   for_each_compatible_node(np, "pci", "fsl,mpc8641-pcie") {
>>       struct resource rsrc;
>>       of_address_to_resource(np, 0, &rsrc);
>>       if ((rsrc.start & 0xfffff) == 0x8000)
>>           fsl_add_bridge(np, 1);
>>       else
>>           fsl_add_bridge(np, 0);
>>   }
>>
>> So now we are using hardcoded offsets again.
>>     
>
> Go with the hardcoded offset.  Linux is broken, so the workaround
> should be in Linux code until Linux PCI code is fixed.
>
> g.
>
>   
Ok, I'll leave the offset checking in mpc83xx_add_bridge rather than 
having it duplicated in the eight different callers.