Re: [PATCH v6 2/7] xen/arm: add pci-domain for disabled devices

[Oleksandr Andrushchenko <Oleksandr_Andrushchenko@xxxxxxxx>]

Hi, Julien!

On 23.11.21 18:05, Julien Grall wrote:
> Hi Oleksandr,
>
> On 23/11/2021 06:31, Oleksandr Andrushchenko wrote:
>>
>>
>> On 22.11.21 19:17, Julien Grall wrote:
>>> Hi,
>>>
>>> On 22/11/2021 16:23, Oleksandr Andrushchenko wrote:
>>>> On 22.11.21 17:29, Julien Grall wrote:
>>>>> I would prefer to go with my way. This can be refined in the future if we 
>>>>> find Device-Tree that matches what you wrote.
>>>> Ok, so just to make it clear:
>>>>    >a PCI device would always be described as a child of the hostbridges. 
>>>> So I would rework the 'if' to also check if the parent type is not "pci"
>>>
>>> That's correct. Thank you!
>> Ok, so how about
>>       if ( is_pci_passthrough_enabled() && dt_device_type_is_equal(node, 
>> "pci") )
>>       {
>>           bool skip = false;
>>
>>           /*
>>            * If the parent is also a "pci" device, then "linux,pci-domain"
>>            * should already be there, so nothing to do then.
>>            */
>
> This comment is a bit confusing.
Do you have something on your mind?
> I think what matter if the parent is a "pci" device, then the current node 
> must not be a hostbridge. So we can skip it.
By skipping you only mean we do not need to add/assign "linux,pci-domain", 
right?
> However...
>
>>           if ( node->parent && dt_device_type_is_equal(node->parent, "pci") )
>>               skip = true;
>>
>>           if ( !skip && !dt_find_property(node, "linux,pci-domain", NULL) )
>>           {
>> I played with a single if and it looks scary...
>
> ... how about introducing an helper that will return true if this node is 
> likely an hostbridge?
This is yet a single use of such a check: why would we need a helper and what 
would that
helper do?
>
> Cheers,
>
Thank you,
Oleksandr