ACPI / PM: Do not infer power-state if there are no D0 power-resources
PATCH 1/1 ACPI / PM: Do not infer power-state if there are no D0 power-resources
Some devices do not have a power-resource-list for D0, but do have a power-resource-lists for e.g. D3 (_PR3).
On these devices the "if (device->power.flags.power_resources)" check in acpi_device_get_power() succeeds because of the presence of the _PR3 resources, so the code used to try and infer the power-state.
In this case since there is no power-resource-list for D0, we can never infer that the device is in D0 even though it very well might be in D0. This results in the code inferring that the device is in D3HOT and on the first suspend acpi_device_set_power() skips calling _PS3 for the device because it thinks the device is already in D3.
An example of a family of devices which are affected by this are Bay Trail based devices. The ACPI device for the XHCI controller on these devices does not have a _PR0 method, but it does have a _PR3 method. The problem described above causes the XHCI controller's _PS3 method not getting called on the first suspend of the device, which causes these devices to not reach the S0i3 power-state during suspend.
Since we cannot infer if the device is in D0 or not when there is no power-resource-list for D0, the best thing to do is to change acpi_power_get_inferred_state() to return ACPI_STATE_UNKNOWN in this case.
--- drivers/acpi/power.c | 11 +++++++++++ 1 file changed, 11 insertions(+)
drivers/acpi/power.c +11 -0
@@ -807,6 +807,17 @@ int acpi_power_get_inferred_state(struct acpi_device *device, int *state)if (!device || !state)return -EINVAL ; +/* +* Some devices do not have a power-resource-list for D0, but do +* have a power-resource-lists for e.g. D3 so we do end up here. +* In this case we can never infer that the device is in D0 even +* though it might very well be in D0, so return ACPI_STATE_UNKNOWN. +*/ +if (list_empty (&device->power.states[ACPI_STATE_D0 ].resources)) { + *state =ACPI_STATE_UNKNOWN ; +return 0 ; + } +/* * We know a device's inferred power state when all the resources * required for a given D-state are 'on'.
Signature · 1 line
2.26.2
Rafael J. Wysocki
maintainerQuoted 3 lines · Hans de Goede
Some devices do not have a power-resource-list for D0, but do have a power-resource-lists for e.g. D3 (
_PR3).
This looks like a bug in the firmware.
It is hard to imagine a design in which some power resources only need to be "on" in the D3hot power state of a device and not in D0 (which is implied by the lack of _PR0 and the presence of _PR3).
So when the device goes from D0 to D3hot, we are expected to turn some power resources "on"? What sense does this make?
I'm guessing that this only works at all, because we only use D0 and D3cold with the affected devices and _PS0 simply turns the power resource(s) in question on.
Quoted 36 lines
On these devices the "if (
device->power.flags.power_resources)" check inacpi_device_get_power()succeeds because of the presence of the_PR3resources, so the code used to try and infer the power-state.In this case since there is no power-resource-list for D0, we can never infer that the device is in D0 even though it very well might be in D0. This results in the code inferring that the device is in D3HOT and on the first suspend
acpi_device_set_power()skips calling_PS3for the device because it thinks the device is already in D3.An example of a family of devices which are affected by this are Bay Trail based devices. The ACPI device for the XHCI controller on these devices does not have a
_PR0method, but it does have a_PR3method. The problem described above causes the XHCI controller's_PS3method not getting called on the first suspend of the device, which causes these devices to not reach the S0i3 power-state during suspend.Since we cannot infer if the device is in D0 or not when there is no power-resource-list for D0, the best thing to do is to change
acpi_power_get_inferred_state()to returnACPI_STATE_UNKNOWNin this case.Signed-off-by Hans de Goede <[email protected]>--- drivers/acpi/power.c | 11 +++++++++++ 1 file changed, 11 insertions(+)drivers/acpi/power.c +0
-0@@ -807,6 +807,17 @@ int acpi_power_get_inferred_state(struct acpi_device *device, int *state)if (!device || !state)return -EINVAL ;
drivers/acpi/power.c int acpi_power_get_inferred_state(struct acpi_device *device, int *state) · line 819
Quoted 10 lines
drivers/acpi/power.c +10
-0+/* +* Some devices do not have a power-resource-list for D0, but do +* have a power-resource-lists for e.g. D3 so we do end up here. +* In this case we can never infer that the device is in D0 even +* though it might very well be in D0, so return ACPI_STATE_UNKNOWN. +*/ +if (list_empty (&device->power.states[ACPI_STATE_D0 ].resources)) { + *state =ACPI_STATE_UNKNOWN ; +return 0 ; + }
Well, this makes things work on the particular affected platform, but that seems to be just by accident, because _PS0 does something special on it.
IMO this needs to be addressed elsewhere and in a different way.
Namely, it looks like if _PR0 is not present (or its return package is empty), but _PR3 is present, we should use the _PR3 list of power resources for D0 as well as for D3hot.
Let me cut a patch for that.
Thanks!
Hi,
Quoted 6 lines · Rafael J. Wysocki
Quoted 3 lines · Hans de Goede
Some devices do not have a power-resource-list for D0, but do have a power-resource-lists for e.g. D3 (
_PR3).This looks like a bug in the firmware.
The DSDT for these devices definitely has somw warts, see below.
TBH I'm not sure of the whole inferring state from power-resource-lists is the best idea. I think just going for UNKNOWN as initial state would be better. The issue is that getting the initial state wrong will lead to either skipping _PS3 (as is happening here) or _PS0 on the first transition to the matching state which can have undesirable side-effects, where as just running _PS0/PS3 for a second time _usually_ should be fine.
Anyways I know we have the inferring code for a long time, and I guess it mostly does its job fine...
Quoted 3 lines
It is hard to imagine a design in which some power resources only need to be "on" in the D3hot power state of a device and not in D0 (which is implied by the lack of
_PR0and the presence of_PR3).
Right, except that the listed resource is a dummy power resource, here is the AML for the USB power-resource which is the only resource in the _PR3 list:
PowerResource (USBC, 0x00, 0x0000)
{
Method (_STA, 0, NotSerialized) // _STA: Status
{
Return (0x0F)
}
Method (_ON, 0, NotSerialized) // _ON_: Power On
{
}
Method (_OFF, 0, NotSerialized) // _OFF: Power Off
{
}
}So not having this in _PR0 does not matter because the whole thing is a dummy implementation.
If you want the whole DSDT, the DSDT of one of the affected devices is attached here: https://bugzilla.kernel.org/show_bug.cgi?id=69011
Note it seems that pretty much any Bay Trail tablet version based device suffers from this.
Quoted 58 lines
So when the device goes from D0 to D3hot, we are expected to turn some power resources "on"? What sense does this make?
I'm guessing that this only works at all, because we only use D0 and D3cold with the affected devices and
_PS0simply turns the power resource(s) in question on.Quoted 36 lines
On these devices the "if (
device->power.flags.power_resources)" check inacpi_device_get_power()succeeds because of the presence of the_PR3resources, so the code used to try and infer the power-state.In this case since there is no power-resource-list for D0, we can never infer that the device is in D0 even though it very well might be in D0. This results in the code inferring that the device is in D3HOT and on the first suspend
acpi_device_set_power()skips calling_PS3for the device because it thinks the device is already in D3.An example of a family of devices which are affected by this are Bay Trail based devices. The ACPI device for the XHCI controller on these devices does not have a
_PR0method, but it does have a_PR3method. The problem described above causes the XHCI controller's_PS3method not getting called on the first suspend of the device, which causes these devices to not reach the S0i3 power-state during suspend.Since we cannot infer if the device is in D0 or not when there is no power-resource-list for D0, the best thing to do is to change
acpi_power_get_inferred_state()to returnACPI_STATE_UNKNOWNin this case.Signed-off-by Hans de Goede <[email protected]>--- drivers/acpi/power.c | 11 +++++++++++ 1 file changed, 11 insertions(+)drivers/acpi/power.c +0
-0@@ -807,6 +807,17 @@ int acpi_power_get_inferred_state(struct acpi_device *device, int *state)if (!device || !state)return -EINVAL ;drivers/acpi/power.c int acpi_power_get_inferred_state(struct acpi_device *device, int *state) · line 819
Quoted 10 lines
drivers/acpi/power.c +10
-0+/* +* Some devices do not have a power-resource-list for D0, but do +* have a power-resource-lists for e.g. D3 so we do end up here. +* In this case we can never infer that the device is in D0 even +* though it might very well be in D0, so return ACPI_STATE_UNKNOWN. +*/ +if (list_empty (&device->power.states[ACPI_STATE_D0 ].resources)) { + *state =ACPI_STATE_UNKNOWN ; +return 0 ; + }Well, this makes things work on the particular affected platform, but that seems to be just by accident, because
_PS0does something special on it.
Quoted 5 lines
IMO this needs to be addressed elsewhere and in a different way.
Namely, it looks like if
_PR0is not present (or its return package is empty), but_PR3is present, we should use the_PR3list of power resources for D0 as well as for D3hot.
That should work too.
Quoted 1 line
Let me cut a patch for that.
Ok, let me know when you have a patch ready then I will test it.
Regards,
Hans
Quoted 33 lines · CEST Hans de Goede
Hi,
Quoted 6 lines · Rafael J. Wysocki
Quoted 3 lines · Hans de Goede
Some devices do not have a power-resource-list for D0, but do have a power-resource-lists for e.g. D3 (
_PR3).This looks like a bug in the firmware.
The DSDT for these devices definitely has somw warts, see below.
TBH I'm not sure of the whole inferring state from power-resource-lists is the best idea. I think just going for UNKNOWN as initial state would be better. The issue is that getting the initial state wrong will lead to either skipping
_PS3(as is happening here) or_PS0on the first transition to the matching state which can have undesirable side-effects, where as just running _PS0/PS3 for a second time _usually_ should be fine.Anyways I know we have the inferring code for a long time, and I guess it mostly does its job fine...
Quoted 3 lines
It is hard to imagine a design in which some power resources only need to be "on" in the D3hot power state of a device and not in D0 (which is implied by the lack of
_PR0and the presence of_PR3).Right, except that the listed resource is a dummy power resource, here is the AML for the USB power-resource which is the only resource in the
_PR3list:
Quoted 6 lines
PowerResource (USBC, 0x00, 0x0000)
{
Method (_STA, 0, NotSerialized) // _STA: Status
{
Return (0x0F)
}This is wrong, because the _STA return value is required to return the current state of the power resource, which evidently is not the case here.
Quoted 10 lines
Method (_ON, 0, NotSerialized) // _ON_: Power On
{
}
Method (_OFF, 0, NotSerialized) // _OFF: Power Off
{
}
}Quoted 2 lines
So not having this in
_PR0does not matter because the whole thing is a dummy implementation.
Now that is an important piece of information, because it basically means that on the affected platform(s) the _PR3 is present only to indicate D3cold support.
It also means that the device doesn't really support D3hot and _PS3 puts it straight into D3cold.
In the face of this, I don't think it is safe to use power resources if the D0 list is empty (either _PR0 is not present or it returns an empty package).
Quoted 5 lines
If you want the whole DSDT, the DSDT of one of the affected devices is attached here: https://bugzilla.kernel.org/show_bug.cgi?id=69011
Note it seems that pretty much any Bay Trail tablet version based device suffers from this.
Well, that's sad. :-/
Quoted 58 lines
Quoted 58 lines
So when the device goes from D0 to D3hot, we are expected to turn some power resources "on"? What sense does this make?
I'm guessing that this only works at all, because we only use D0 and D3cold with the affected devices and
_PS0simply turns the power resource(s) in question on.Quoted 36 lines
On these devices the "if (
device->power.flags.power_resources)" check inacpi_device_get_power()succeeds because of the presence of the_PR3resources, so the code used to try and infer the power-state.In this case since there is no power-resource-list for D0, we can never infer that the device is in D0 even though it very well might be in D0. This results in the code inferring that the device is in D3HOT and on the first suspend
acpi_device_set_power()skips calling_PS3for the device because it thinks the device is already in D3.An example of a family of devices which are affected by this are Bay Trail based devices. The ACPI device for the XHCI controller on these devices does not have a
_PR0method, but it does have a_PR3method. The problem described above causes the XHCI controller's_PS3method not getting called on the first suspend of the device, which causes these devices to not reach the S0i3 power-state during suspend.Since we cannot infer if the device is in D0 or not when there is no power-resource-list for D0, the best thing to do is to change
acpi_power_get_inferred_state()to returnACPI_STATE_UNKNOWNin this case.Signed-off-by Hans de Goede <[email protected]>--- drivers/acpi/power.c | 11 +++++++++++ 1 file changed, 11 insertions(+)drivers/acpi/power.c +0
-0@@ -807,6 +807,17 @@ int acpi_power_get_inferred_state(struct acpi_device *device, int *state)if (!device || !state)return -EINVAL ;drivers/acpi/power.c int acpi_power_get_inferred_state(struct acpi_device *device, int *state) · line 819
Quoted 10 lines
drivers/acpi/power.c +10
-0+/* +* Some devices do not have a power-resource-list for D0, but do +* have a power-resource-lists for e.g. D3 so we do end up here. +* In this case we can never infer that the device is in D0 even +* though it might very well be in D0, so return ACPI_STATE_UNKNOWN. +*/ +if (list_empty (&device->power.states[ACPI_STATE_D0 ].resources)) { + *state =ACPI_STATE_UNKNOWN ; +return 0 ; + }Well, this makes things work on the particular affected platform, but that seems to be just by accident, because
_PS0does something special on it.
Quoted 7 lines
Quoted 5 lines
IMO this needs to be addressed elsewhere and in a different way.
Namely, it looks like if
_PR0is not present (or its return package is empty), but_PR3is present, we should use the_PR3list of power resources for D0 as well as for D3hot.That should work too.
No, it wouldn't, because the _PR3 power resources above are not functional.
Quoted 3 lines
Quoted 1 line
Let me cut a patch for that.
Ok, let me know when you have a patch ready then I will test it.
Sure, please find appended.
That just avoids using power resources at all if the D0 list is empty.
--- drivers/acpi/scan.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-)
drivers/acpi/scan.c +11 -6
@@ -919,12 +919,9 @@ static void acpi_bus_init_power_state(stif (buffer.length && package && package->type ==ACPI_TYPE_PACKAGE - && package->package.count) { -int err =acpi_extract_power_resources (package,0 , - &ps->resources); -if (!err) - device->power.flags.power_resources =1 ; - } + && package->package.count) +acpi_extract_power_resources (package,0 , &ps->resources); +ACPI_FREE (buffer.pointer); }
@@ -970,6 +967,14 @@ static void acpi_bus_get_power_flags(strfor (i =ACPI_STATE_D0 ; i <=ACPI_STATE_D3_HOT ; i++)acpi_bus_init_power_state (device, i); +/* +* Use power resources only if the D0 list of them is populated, because +* some platforms only provide _PR3 to indicate D3cold support and in +* those cases the power resources list returned by _PR3 may be bogus. +*/ +if (!list_empty (&device->power.states[ACPI_STATE_D0 ].resources)) + device->power.flags.power_resources =1 ; +INIT_LIST_HEAD (&device->power.states[ACPI_STATE_D3_COLD ].resources);if (!list_empty (&device->power.states[ACPI_STATE_D3_HOT ].resources)) device->power.states[ACPI_STATE_D3_COLD ].flags.valid =1 ;
PATCH 1/1 ACPI: PM: Avoid using power resources if there are none for D0
In reply to Hans de GoedeFrom: Rafael J. Wysocki <[email protected]>
As recently reported, some platforms provide a list of power resources for device power state D3hot, through the _PR3 object, but they do not provide a list of power resources for device power state D0.
Among other things, this causes acpi_device_get_power() to return D3hot as the current state of the device in question if all of the D3hot power resources are "on", because it sees the power_resources flag set and calls acpi_power_get_inferred_state() which finds that D3hot is the shallowest power state with all of the associated power resources turned "on", so that's what it returns. Moreover, that value takes precedence over the acpi_dev_pm_explicit_get() return value, because it means a deeper power state. The device may very well be in D0 physically at that point, however.
Moreover, the presence of _PR3 without _PR0 for a given device means that only one D3-level power state can be supported by it. Namely, because there are no power resources to turn "off" when transitioning the device from D0 into D3cold (which should be supported since _PR3 is present), the evaluation of _PS3 should be sufficient to put it straight into D3cold, but this means that the effect of turning "on" the _PR3 power resources is unclear, so it is better to avoid doing that altogether. Consequently, there is no practical way do distinguish D3cold from D3hot for the device in question and the power states of it can be labeled so that D3hot is the deepest supported one (and Linux assumes that putting a device into D3hot via ACPI may cause power to be removed from it anyway, for legacy reasons).
To work around the problem described above modify the ACPI enumeration of devices so that power resources are only used for device power management if the list of D0 power resources is not empty and make it mart D3cold as supported only if that is the case and the D3hot list of power resources is not empty too.
--- drivers/acpi/device_pm.c | 2 +- drivers/acpi/scan.c | 28 +++++++++++++++++++--------- 2 files changed, 20 insertions(+), 10 deletions(-)
drivers/acpi/scan.c +19 -9
@@ -919,12 +919,9 @@ static void acpi_bus_init_power_state(stif (buffer.length && package && package->type ==ACPI_TYPE_PACKAGE - && package->package.count) { -int err =acpi_extract_power_resources (package,0 , - &ps->resources); -if (!err) - device->power.flags.power_resources =1 ; - } + && package->package.count) +acpi_extract_power_resources (package,0 , &ps->resources); +ACPI_FREE (buffer.pointer); }
@@ -971,14 +968,27 @@ static void acpi_bus_get_power_flags(stracpi_bus_init_power_state (device, i);INIT_LIST_HEAD (&device->power.states[ACPI_STATE_D3_COLD ].resources); -if (!list_empty (&device->power.states[ACPI_STATE_D3_HOT ].resources)) - device->power.states[ACPI_STATE_D3_COLD ].flags.valid =1 ; -/* Set defaults for D0 and D3hot states (always valid) */ +/* Set the defaults for D0 and D3hot (always supported). */ device->power.states[ACPI_STATE_D0 ].flags.valid =1 ; device->power.states[ACPI_STATE_D0 ].power =100 ; device->power.states[ACPI_STATE_D3_HOT ].flags.valid =1 ; +/* +* Use power resources only if the D0 list of them is populated, because +* some platforms may provide _PR3 only to indicate D3cold support and +* in those cases the power resources list returned by it may be bogus. +*/ +if (!list_empty (&device->power.states[ACPI_STATE_D0 ].resources)) { + device->power.flags.power_resources =1 ; +/* +* D3cold is supported if the D3hot list of power resources is +* not empty. +*/ +if (!list_empty (&device->power.states[ACPI_STATE_D3_HOT ].resources)) + device->power.states[ACPI_STATE_D3_COLD ].flags.valid =1 ; + } +if (acpi_bus_init_power (device)) device->flags.power_manageable =0 ; }
drivers/acpi/device_pm.c +1 -1
@@ -186,7 +186,7 @@ int acpi_device_set_power(struct acpi_de* possibly drop references to the power resources in use. */ state =ACPI_STATE_D3_HOT ; -/* If _PR3 is not available, use D3hot as the target state. */ +/* If D3cold is not supported, use D3hot as the target state. */ if (!device->power.states[ACPI_STATE_D3_COLD ].flags.valid) target_state = state; }else if (!device->power.states[state].flags.valid) {
Re: [PATCH] ACPI: PM: Avoid using power resources if there are none for D0
Hi,
Quoted 38 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
As recently reported, some platforms provide a list of power resources for device power state D3hot, through the
_PR3object, but they do not provide a list of power resources for device power state D0.Among other things, this causes
acpi_device_get_power()to return D3hot as the current state of the device in question if all of the D3hot power resources are "on", because it sees thepower_resourcesflag set and callsacpi_power_get_inferred_state()which finds that D3hot is the shallowest power state with all of the associated power resources turned "on", so that's what it returns. Moreover, that value takes precedence over theacpi_dev_pm_explicit_get()return value, because it means a deeper power state. The device may very well be in D0 physically at that point, however.Moreover, the presence of
_PR3without_PR0for a given device means that only one D3-level power state can be supported by it. Namely, because there are no power resources to turn "off" when transitioning the device from D0 into D3cold (which should be supported since_PR3is present), the evaluation of_PS3should be sufficient to put it straight into D3cold, but this means that the effect of turning "on" the_PR3power resources is unclear, so it is better to avoid doing that altogether. Consequently, there is no practical way do distinguish D3cold from D3hot for the device in question and the power states of it can be labeled so that D3hot is the deepest supported one (and Linux assumes that putting a device into D3hot via ACPI may cause power to be removed from it anyway, for legacy reasons).To work around the problem described above modify the ACPI enumeration of devices so that power resources are only used for device power management if the list of D0 power resources is not empty and make it mart D3cold as supported only if that is the case and the D3hot list of power resources is not empty too.
Quoted 6 lines
Fixes ef85bdbec444 ("ACPI / scan: Consolidate extraction of power resources lists")Reported-by Hans de Goede <[email protected]>Cc 3.10+ <[email protected]> # 3.10+Signed-off-by Rafael J. Wysocki <[email protected]>
Thank you.
I've tested this and I can confirm that it fixes the issue:
I've also looked at the code and it looks good to me:
Regards,
Hans
Quoted 72 lines
--- drivers/acpi/device_pm.c | 2 +- drivers/acpi/scan.c | 28 +++++++++++++++++++--------- 2 files changed, 20 insertions(+), 10 deletions(-)Index:
linux-pm/drivers/acpi/scan.c
===================================================================drivers/acpi/scan.c +19
-9@@ -919,12 +919,9 @@ static void acpi_bus_init_power_state(stif (buffer.length && package && package->type ==ACPI_TYPE_PACKAGE - && package->package.count) { -int err =acpi_extract_power_resources (package,0 , - &ps->resources); -if (!err) - device->power.flags.power_resources =1 ; - } + && package->package.count) +acpi_extract_power_resources (package,0 , &ps->resources); +ACPI_FREE (buffer.pointer); }@@ -971,14 +968,27 @@ static void acpi_bus_get_power_flags(stracpi_bus_init_power_state (device, i);INIT_LIST_HEAD (&device->power.states[ACPI_STATE_D3_COLD ].resources); -if (!list_empty (&device->power.states[ACPI_STATE_D3_HOT ].resources)) - device->power.states[ACPI_STATE_D3_COLD ].flags.valid =1 ; -/* Set defaults for D0 and D3hot states (always valid) */ +/* Set the defaults for D0 and D3hot (always supported). */ device->power.states[ACPI_STATE_D0 ].flags.valid =1 ; device->power.states[ACPI_STATE_D0 ].power =100 ; device->power.states[ACPI_STATE_D3_HOT ].flags.valid =1 ; +/* +* Use power resources only if the D0 list of them is populated, because +* some platforms may provide _PR3 only to indicate D3cold support and +* in those cases the power resources list returned by it may be bogus. +*/ +if (!list_empty (&device->power.states[ACPI_STATE_D0 ].resources)) { + device->power.flags.power_resources =1 ; +/* +* D3cold is supported if the D3hot list of power resources is +* not empty. +*/ +if (!list_empty (&device->power.states[ACPI_STATE_D3_HOT ].resources)) + device->power.states[ACPI_STATE_D3_COLD ].flags.valid =1 ; + } +if (acpi_bus_init_power (device)) device->flags.power_manageable =0 ; }Index:
linux-pm/drivers/acpi/device_pm.c
===================================================================drivers/acpi/device_pm.c +1
-1@@ -186,7 +186,7 @@ int acpi_device_set_power(struct acpi_de* possibly drop references to the power resources in use. */ state =ACPI_STATE_D3_HOT ; -/* If _PR3 is not available, use D3hot as the target state. */ +/* If D3cold is not supported, use D3hot as the target state. */ if (!device->power.states[ACPI_STATE_D3_COLD ].flags.valid) target_state = state; }else if (!device->power.states[state].flags.valid) {