On Fri, 28 Feb 2025 14:58:55 -0600
Bjorn Helgaas <helgaas@xxxxxxxxxx> wrote:
> On Mon, Feb 24, 2025 at 03:13:52PM +0100, Herve Codina wrote:
> > The commit 407d1a51921e ("PCI: Create device tree node for bridge")
> > creates of_node for PCI devices. The newly created of_node is attached
> > to an existing device. This is done setting directly pdev->dev.of_node
> > in the code.
> >
> > Even if pdev->dev.of_node cannot be previously set, this doesn't handle
> > the fwnode field of the struct device. Indeed, this field needs to be
> > set if it hasn't already been set.
> >
> > device_{add,remove}_of_node() have been introduced to handle this case.
>
> I guess another way to say this is:
>
> - If dev->of_node has already been set, it is an error and we want
> to do nothing. The error is impossible in this case because
> of_pci_make_dev_node() returns early if dev->of_node has been set.
>
> - Otherwise, we want to set dev->of_node (just as we previously
> did), and
>
> - if dev->fwnode has not been set, we want to set that too.
>
> So the whole point of this is to set dev->fwnode, which we didn't do
> before. But has np->fwnode been set to anything? Maybe it's buried
> somewhere inside of_changeset_create_node(), but I didn't see it.
np->fwnode can be set by ACPI. We are at the frontier between ACPI and
device-tree.
The ofnode is created and filled from an already existing device. This
device can be created from information provided by the ACPI world.
In that case, np->fwnode is set to and ACPI fwnode.
Best regards,
Hervé
