Re: [PATCH 2/3] PCI: of: create DT nodes for PCI devices if they do not exists

[Lizhi Hou]

On 5/4/22 06:43, Clément Léger wrote:
> Le Tue, 3 May 2022 17:53:53 -0500,
> Bjorn Helgaas <helgaas@xxxxxxxxxx> a écrit :
>
> > In subject:
> >
> >    PCI: of: Create DT nodes ... if they do not exist
> >
> > The subject could be read as saying that you're going to create DT
> > nodes before the PCI devices exist, but I think you mean that when we
> > enumerate a PCI devices, we're *always* going to create a DT node for
> > it, even if the DT didn't mention it.
> Hi Bjorn,
>
> Indeed ! I'll modify that.

Linking the dynamic generated dt node to PCIe device through 
pci_dev->dev.of_node may cause issues. Kernel and driver code may check 
of_node pointer and run complete different code path if of_node is not 
NULL.

For example:  in of_irq_parse_pci(): 
https://elixir.bootlin.com/linux/v5.18-rc2/source/drivers/pci/of.c#L492

I encountered different issues when I tried to create a prototype. And I 
have sent all may questions/thoughts through 
https://lore.kernel.org/lkml/79e4c876-e5a4-861f-cfbc-c75ed1a07ddd@xxxxxxxxxx/#t

I am wondering what would be the right way to resolve it?


Thanks,

Lizhi

> > Maybe something like:
> >
> >    PCI: of: Create DT node for every PCI device
> >
> > although I see Rob thinks this should be done on demand instead of
> > doing it for every device, which sounds sensible to me.
> Agreed, I'll rework this series.
>
> Thanks,
>
> > On Wed, Apr 27, 2022 at 11:45:01AM +0200, Clément Léger wrote:
> > > In order to apply overlays to PCI device nodes, the nodes must first
> > > exist. This commit add support to populate a skeleton tree for PCI bus
> > > and devices. These nodes can then be used by drivers to apply overlays.
> > s/This commit add support/Add support/
> >
> > Bjorn