[PATCH v8 30/45] powerpc/pci: Delay populating pdn

[Alexey Kardashevskiy]

On 04/20/2016 12:13 PM, Gavin Shan wrote:
> On Tue, Apr 19, 2016 at 06:19:20PM +1000, Alexey Kardashevskiy wrote:
>> On 02/17/2016 02:44 PM, Gavin Shan wrote:
>>> The pdn (struct pci_dn) instances are allocated from memblock or
>>> bootmem when creating PCI controller (hoses) in setup_arch(). PCI
>>> hotplug, which will be supported by proceeding patches, releases
>>> PCI device nodes and their corresponding pdn on unplugging event.
>>> The memory chunks for pdn instances allocated from memblock or
>>> bootmem are hard to reused after being released.
>>>
>>> This delays creating pdn by pci_devs_phb_init() from setup_arch()
>>> to core_initcall() so that they are allocated from slab. The memory
>>> consumed by pdn can be released to system without problem during
>>> PCI unplugging time. It indicates that pci_dn is unavailable in
>>> setup_arch() and the the fixup on pdn (like AGP's) can't be carried
>>> out that time. We have to do that in ppc_md.pcibios_root_bridge_prepare()
>>> on maple/pasemi/powermac platforms where/when the pdn is available.
>>>
>>> At the mean while, the EEH device is created when pdn is populated,
>>> meaning pdn and EEH device have same life cycle. In turn, we needn't
>>> call eeh_dev_init() to create EEH device explicitly.
>>>
>>> Signed-off-by: Gavin Shan <gwshan at linux.vnet.ibm.com>
>>
>>
>> Uff. It would not hurt to mention that  pcibios_root_bridge_prepare is called
>>from subsys_initcall() which is executed after core_initcall() so the code
>> flow does not change.
>>
>
> Yes, will do in next revision.
>
>> Have you checked if there is anything in between
>> core_initcall(pci_devs_phb_init) and subsys_initcall(pcibios_init) which
>> might need device tree nodes? For example, subsys_initcall(pcibios_init)
>> calls (eventually) pnv_pci_ioda_fixup(), if we are unlucky and pcibios_init()
>> (and therefore pnv_pci_ioda_fixup() or what pseries/others do) is called
>> before pcibios_init() - won't we crash or something?
>>
>
> I don't catch what you were asking. device-tree nodes (struct device_node)
> are always there. This patch doesn't affect them. Perhaps you were talking
> about pdn (PCI_DN). If it's the case, this patch delays creating pdn from
> setup_arch() to core_initcall(pci_devs_phb_init).


While thinking of explaining what I wanted to ask, I found my answer :)

pcibios_init() calls ppc_md.pcibios_root_bridge_prepare() first, then 
ppc_md.pcibios_fixup() so we are fine here with ordering.


> I don't see anything need pdn between setup_arch() and core_initcall().
> The changes introduced to powermac/pasemi platforms are: move fixing the child
> pdns of the specifiec PHB's pdn from setup_arch() to subsys_initcall(pcibios_init).
> I don't see anything between them needs the fixed pdns.
>
> I don't understand how pcibios_init() is called before pcibios_init() in your

pcibios_init() is used twice in the sentence above :)

Anyway,


Reviewed-by: Alexey Kardashevskiy <aik at ozlabs.ru>




> context. Sorry for my bad English. Perhaps you're asking the the called sequence
> on core_initcall() and subsys_init()? If so, they're defined like below:
>
> #define core_initcall(fn)		__define_initcall(fn, 1)
> #define subsys_initcall(fn)		__define_initcall(fn, 4)
 >
>
>>
>> --
>> Alexey
>>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-pci" in
> the body of a message to majordomo at vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>


-- 
Alexey