Re: [PATCH 06/15] PCI/ACPI: Link host bridge to its ACPI fw node

[Dan Williams]

Bjorn Helgaas wrote:
> On Fri, Sep 09, 2022 at 12:20:45PM +0200, Robert Richter wrote:
> > On 08.09.22 12:45:16, Dan Williams wrote:
> > > Rafael J. Wysocki wrote:
> > > > On Thu, Sep 8, 2022 at 8:05 AM Dan Williams <dan.j.williams@xxxxxxxxx> wrote:
> > > > >
> > > > > Robert Richter wrote:
> > > > > > A lookup of a host bridge's corresponding acpi device (struct
> > > > > > acpi_device) is not possible, for example:
> > > > > >
> > > > > >       adev = ACPI_COMPANION(&host_bridge->dev);
> > > > > >
> > > > > > This could be useful to find a host bridge's fwnode handle and to
> > > > > > determine and call additional host bridge ACPI parameters and methods
> > > > > > such as HID/CID or _UID.
> 
> > ...
> > No, it is x86. And true, it is set. So this series is actually working
> > without this patch. It can be dropped.
> > 
> > Now, I just checked my logs. The reason I was adding this is that
> > during code development I modified the code to have bridge->dev.parent
> > set. Then, the fwnode is not linked. I later dropped that change but
> > kept this patch.
> 
> If this patch does the same thing as the ACPI_COMPANION_SET() in
> several pcibios_root_bridge_prepare() implementations, I would love to
> keep this patch, which does it in a generic place, and drop the
> corresponding code from those arch-specific functions.
> 
> But I don't understand the fwnode stuff well enough to know if this is
> feasible.

I took a brief look, but I could not convince myself that these lookups:

arch/ia64/pci/pci.c ((struct pci_controller *) bridge->bus->sysdata)->companion
arch/loongarch/pci/acpi.c to_acpi_device(((struct pci_config_window *) bridge->bus->sysdata)->parent)
arch/arm64/kernel/pci.c to_acpi_device(((struct pci_config_window *) bridge->bus->sysdata)->parent)
arch/x86/pci/acpi.c ((struct pci_sysdata *) bridge->bus->sysdata)->companion

...are identical to what acpi_pci_root_add() used to create the
acpi_pci_root.