[PATCH] ACPI: glue: Use acpi_device_adr() in acpi_find_child_device()

[PATCH] ACPI: glue: Use acpi_device_adr() in acpi_find_child_device()

[Rafael J. Wysocki]

From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Instead of evaluating _ADR in acpi_find_child_device(), use the
observation that it has already been evaluated and the value returned
by it has been stored in the pnp.type.bus_address field of the ACPI
device object at hand.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---

Note that the patch at

https://patchwork.kernel.org/project/linux-acpi/patch/2223516.ElGaqSPkdT@kreacher/

will have an effect of this, so I'm going to prioritize the current patch
over the one above.

---
 drivers/acpi/glue.c |    7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

Index: linux-pm/drivers/acpi/glue.c
===================================================================
--- linux-pm.orig/drivers/acpi/glue.c
+++ linux-pm/drivers/acpi/glue.c
@@ -113,13 +113,10 @@ struct acpi_device *acpi_find_child_devi
 		return NULL;
 
 	list_for_each_entry(adev, &parent->children, node) {
-		unsigned long long addr;
-		acpi_status status;
+		acpi_bus_address addr = acpi_device_adr(adev);
 		int score;
 
-		status = acpi_evaluate_integer(adev->handle, METHOD_NAME__ADR,
-					       NULL, &addr);
-		if (ACPI_FAILURE(status) || addr != address)
+		if (!adev->pnp.type.bus_address || addr != address)
 			continue;
 
 		if (!ret) {

Re: [PATCH] ACPI: glue: Use acpi_device_adr() in acpi_find_child_device()

[Andy Shevchenko]

On Wed, Oct 27, 2021 at 06:59:06PM +0200, Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> 
> Instead of evaluating _ADR in acpi_find_child_device(), use the
> observation that it has already been evaluated and the value returned
> by it has been stored in the pnp.type.bus_address field of the ACPI
> device object at hand.

...

> +		acpi_bus_address addr = acpi_device_adr(adev);
>  		int score;
>  
> -		status = acpi_evaluate_integer(adev->handle, METHOD_NAME__ADR,
> -					       NULL, &addr);
> -		if (ACPI_FAILURE(status) || addr != address)
> +		if (!adev->pnp.type.bus_address || addr != address)
>  			continue;

I'm not sure I understand the new check, i.e. !adev->pnp.type.bus_address.
IIUC _ADR == 0 is a valid value and children may have it like this.

I believe this change will break the working things (first comes to my mind
is drivers/mfd/intel_quark_i2c_gpio.c).

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] ACPI: glue: Use acpi_device_adr() in acpi_find_child_device()

[Rafael J. Wysocki]

On Wed, Oct 27, 2021 at 7:17 PM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
>
> On Wed, Oct 27, 2021 at 06:59:06PM +0200, Rafael J. Wysocki wrote:
> > From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> >
> > Instead of evaluating _ADR in acpi_find_child_device(), use the
> > observation that it has already been evaluated and the value returned
> > by it has been stored in the pnp.type.bus_address field of the ACPI
> > device object at hand.
>
> ...
>
> > +             acpi_bus_address addr = acpi_device_adr(adev);
> >               int score;
> >
> > -             status = acpi_evaluate_integer(adev->handle, METHOD_NAME__ADR,
> > -                                            NULL, &addr);
> > -             if (ACPI_FAILURE(status) || addr != address)
> > +             if (!adev->pnp.type.bus_address || addr != address)
> >                       continue;
>
> I'm not sure I understand the new check, i.e. !adev->pnp.type.bus_address.

See acpi_set_pnp_ids().

adev->pnp.bus_addres is the address value and
adev->pnp.type.bus_address is a flag that is set to 1 when _ADR is
valid.

> IIUC _ADR == 0 is a valid value and children may have it like this.

That's true.

> I believe this change will break the working things (first comes to my mind
> is drivers/mfd/intel_quark_i2c_gpio.c).

I don't think so.

Re: [PATCH] ACPI: glue: Use acpi_device_adr() in acpi_find_child_device()

[Andy Shevchenko]

On Wed, Oct 27, 2021 at 07:23:43PM +0200, Rafael J. Wysocki wrote:
> On Wed, Oct 27, 2021 at 7:17 PM Andy Shevchenko
> <andriy.shevchenko@linux.intel.com> wrote:
> > On Wed, Oct 27, 2021 at 06:59:06PM +0200, Rafael J. Wysocki wrote:

...

> See acpi_set_pnp_ids().
> 
> adev->pnp.bus_addres is the address value and
> adev->pnp.type.bus_address is a flag that is set to 1 when _ADR is
> valid.
> 
> > IIUC _ADR == 0 is a valid value and children may have it like this.
> 
> That's true.
> 
> > I believe this change will break the working things (first comes to my mind
> > is drivers/mfd/intel_quark_i2c_gpio.c).
> 
> I don't think so.

With your explanation I see where my logic is wrong. Thanks!

-- 
With Best Regards,
Andy Shevchenko