Re: [PATCH v2 06/29] x86/PCI: convert to pci_create_root_bus() andpci_scan_root_bus()

[Yinghai Lu]

On 10/14/2011 01:54 PM, Bjorn Helgaas wrote:

> On Fri, Oct 14, 2011 at 2:39 PM, Yinghai Lu <yinghai.lu@xxxxxxxxxx> wrote:
>> On 10/14/2011 01:32 PM, Bjorn Helgaas wrote:
>>
>>
>>>>>       int node;
>>>>> @@ -353,11 +349,18 @@ struct pci_bus * __devinit pci_acpi_scan_root(struct acpi_pci_root *root)
>>>>>               memcpy(bus->sysdata, sd, sizeof(*sd));
>>>>>               kfree(sd);
>>>>>       } else {
>>>>> -             bus = pci_create_bus(NULL, busnum, &pci_root_ops, sd);
>>>>> -             if (bus) {
>>>>> -                     get_current_resources(device, busnum, domain, bus);
>>>>> -                     bus->subordinate = pci_scan_child_bus(bus);
>>>>> +             INIT_LIST_HEAD(&resources);
>>>>> +             get_current_resources(device, busnum, domain, &resources);
>>>>> +             if (!pci_use_crs) {
>>>>> +                     pci_free_resource_list(&resources);
>>>>> +                     x86_pci_root_bus_resources(busnum, &resources);
>>>>>               }
>>>>
>>>>
>>>> You may need to update get_current_resources() to return status about handling _CRS...
>>>> and check that status insteaf of !pci_use_crs.
>>>
>>> Is the current patch broken here?  I don't think it will be simpler to
>>> have get_current_resources() return a status and check that.  But if
>>> something's actually broken, I want to fix it, of course.
>>>
>>
>>
>> yes. if for some reason, _CRS is not handled right, that root bus resources will not get set to default res.
>>
>> +             status = get_current_resources(device, busnum, domain, &resources);
>> +             if (status) {
>> +                     pci_free_resource_list(&resources);
>> +                     x86_pci_root_bus_resources(busnum, &resources);
>>              }
>>
>> get_current_resources() will not return 0 if _CRS is not really used.
> 
> OK.  I think it was that way before, too.  We created the bus with
> default resources, but the first thing get_current_resources() did was
> clear them all out.  If there was a problem parsing _CRS, we never
> went back and put the defaults back in.


that means you broke it when you clean up it last time. 

> 
> But I think this will do what you suggest and is nicer than what I had before:
> 
>                 INIT_LIST_HEAD(&resources);
>                 get_current_resources(device, busnum, domain, &resources);
>                 if (list_empty(&resources))
>                         x86_pci_root_bus_resources(busnum, &resources);


yes, that should work.

Yinghai
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/