[PATCH v2 0/2] ACPI: scan: Turn off unused power resources during initialization

[PATCH v2 0/2] ACPI: scan: Turn off unused power resources during initialization

[Rafael J. Wysocki]

Hi,

Power resources that aren't associated with any devices are expected to be
turned off by the OS, but Linux only does that during resume from
suspend-to-RAM.  That turns out to be problematic, so turn them off during
system initialization too (patch [1/2]).

In addition to that, turn them off unconditionally, without checking the
current status, which should be safe and may help to work around firmware
issues (patch [2/2]).

Please refer to the patch changelogs for details.

Thanks!

[PATCH v2 1/2] ACPI: scan: Turn off unused power resources during initialization

[Rafael J. Wysocki]

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

It is reported that on certain platforms there are power resources
that are not associated with any devices physically present in the
platform.  Those power resources are expected to be turned off by
the OS in accordance with the ACPI specification (section 7.3 of
ACPI 6.4) which currently is not done by Linux and that may lead
to obscure issues.

For instance, leaving those power resources in the "on" state may
prevent the platform from reaching the lowest power state in
suspend-to-idle which leads to excessive power draw.

For this reason, turn all of the unused ACPI power resources off
at the end of the initial namespace scan for devices in analogy with
resume from suspend-to-RAM.

Link: https://uefi.org/specs/ACPI/6.4/07_Power_and_Performance_Mgmt/device-power-management-objects.html
Reported-by: David Box <david.e.box@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---

-> v2: Expand the changelog somewhat and make it build with CONFIG_PM_SLEEP
       unset.

---
 drivers/acpi/internal.h |    1 +
 drivers/acpi/power.c    |    2 +-
 drivers/acpi/scan.c     |    2 ++
 drivers/acpi/sleep.h    |    1 -
 4 files changed, 4 insertions(+), 2 deletions(-)

Index: linux-pm/drivers/acpi/internal.h
===================================================================
--- linux-pm.orig/drivers/acpi/internal.h
+++ linux-pm/drivers/acpi/internal.h
@@ -139,6 +139,7 @@ int acpi_device_sleep_wake(struct acpi_d
 int acpi_power_get_inferred_state(struct acpi_device *device, int *state);
 int acpi_power_on_resources(struct acpi_device *device, int state);
 int acpi_power_transition(struct acpi_device *device, int state);
+void acpi_turn_off_unused_power_resources(void);
 
 /* --------------------------------------------------------------------------
                               Device Power Management
Index: linux-pm/drivers/acpi/scan.c
===================================================================
--- linux-pm.orig/drivers/acpi/scan.c
+++ linux-pm/drivers/acpi/scan.c
@@ -2360,6 +2360,8 @@ int __init acpi_scan_init(void)
 		}
 	}
 
+	acpi_turn_off_unused_power_resources();
+
 	acpi_scan_initialized = true;
 
  out:
Index: linux-pm/drivers/acpi/sleep.h
===================================================================
--- linux-pm.orig/drivers/acpi/sleep.h
+++ linux-pm/drivers/acpi/sleep.h
@@ -8,7 +8,6 @@ extern struct list_head acpi_wakeup_devi
 extern struct mutex acpi_device_lock;
 
 extern void acpi_resume_power_resources(void);
-extern void acpi_turn_off_unused_power_resources(void);
 
 static inline acpi_status acpi_set_waking_vector(u32 wakeup_address)
 {
Index: linux-pm/drivers/acpi/power.c
===================================================================
--- linux-pm.orig/drivers/acpi/power.c
+++ linux-pm/drivers/acpi/power.c
@@ -996,6 +996,7 @@ void acpi_resume_power_resources(void)
 
 	mutex_unlock(&power_resource_list_lock);
 }
+#endif
 
 void acpi_turn_off_unused_power_resources(void)
 {
@@ -1025,4 +1026,3 @@ void acpi_turn_off_unused_power_resource
 
 	mutex_unlock(&power_resource_list_lock);
 }
-#endif

[PATCH v2 2/2] ACPI: power: Turn off unused power resources unconditionally

[Rafael J. Wysocki]

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

According to the ACPI specification (section 7.2.2 in ACPI 6.4), the
OS may evaluate the _OFF method of a power resource that is "off"
already [1], and in particular that can be done in the case of unused
power resources.

Accordingly, modify acpi_turn_off_unused_power_resources() to
evaluate the _OFF method for each of the unused power resources
unconditionally which may help to work around BIOS issues where the
return value of _STA for a power resource does not reflect the
actual state of the power resource [2].

Link: https://uefi.org/specs/ACPI/6.4/07_Power_and_Performance_Mgmt/declaring-a-power-resource-object.html#off # [1]
Link: https://lore.kernel.org/lkml/20210314000439.3138941-1-luzmaximilian@gmail.com/ # [2]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---

New patch in v2.

---
 drivers/acpi/power.c |   11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)

Index: linux-pm/drivers/acpi/power.c
===================================================================
--- linux-pm.orig/drivers/acpi/power.c
+++ linux-pm/drivers/acpi/power.c
@@ -1005,18 +1005,9 @@ void acpi_turn_off_unused_power_resource
 	mutex_lock(&power_resource_list_lock);
 
 	list_for_each_entry_reverse(resource, &acpi_power_resource_list, list_node) {
-		int result, state;
-
 		mutex_lock(&resource->resource_lock);
 
-		result = acpi_power_get_state(resource->device.handle, &state);
-		if (result) {
-			mutex_unlock(&resource->resource_lock);
-			continue;
-		}
-
-		if (state == ACPI_POWER_RESOURCE_STATE_ON
-		    && !resource->ref_count) {
+		if (!resource->ref_count) {
 			dev_info(&resource->device.dev, "Turning OFF\n");
 			__acpi_power_off(resource);
 		}