[PATCH v2 2/2] [usb] add support for ACPI identification to xhci-platform

Arnd Bergmann arnd at arndb.de
Wed Nov 5 12:41:23 PST 2014


On Wednesday 05 November 2014 11:55:07 Greg KH wrote:
> On Wed, Nov 05, 2014 at 01:44:43PM -0600, Mark Langsdorf wrote:
> > On 11/05/2014 01:11 PM, Greg KH wrote:
> > >On Wed, Nov 05, 2014 at 07:59:32AM -0600, Mark Langsdorf wrote:
> > >>>>  static struct platform_driver usb_xhci_driver = {
> > >>>>          .probe  = xhci_plat_probe,
> > >>>>          .remove = xhci_plat_remove,
> > >>>>@@ -294,6 +304,7 @@ static struct platform_driver usb_xhci_driver = {
> > >>>>                  .name = "xhci-hcd",
> > >>>>                  .pm = DEV_PM_OPS,
> > >>>>                  .of_match_table = of_match_ptr(usb_xhci_of_match),
> > >>>>+         .acpi_match_table = ACPI_PTR(usb_xhci_acpi_match),
> > >>>
> > >>>Shouldn't the reworked driver core code handle this differently with the
> > >>>ability to handle either OF or ACPI in the same driver?
> > >>
> > >>I'm not sure I understand the question. With these changes, the driver
> > >>handles both ACPI and DTB/OF. It's the same style of code as used
> > >>in drivers/ata/plat-xgene.c, which also handles both ACPI and DTB/OF.
> > >>Why do you think this code isn't correct?
> > >
> > >There is a new framework in the kernel that keeps a driver from having
> > >to query both of and acpi to get the needed resources, it just does one
> > >query and depending on the platform, everything "just works".  Shouldn't
> > >that be used here as well?
> > 
> > Would you send me a pointer to a driver that's using this new
> > framework? I can't find any references to it and all the other
> > drivers that support ACPI and OF are doing it the way I'm doing
> > it.
> 
> See the email on lkml:
>  Subject: [PATCH v6 00/12] Add ACPI _DSD and unified device properties support
> 
> for the latest patch series.
> 

The _DSD approach is for devices that do not follow the ACPI specification
but do have a DT binding. Those will work without the .acpi_match_table
entry when the firmware uses the compatible value in the new properties.

In this case, the device does have an official ACPI ID "PNP0D10", so we should
use that for compatibility with other operating systems and with BIOS
versions that provide the standard IDs.

	Arnd



More information about the linux-arm-kernel mailing list