Skip to contentSkip to conversation

pwm: lpss: Check PWM powerstate after resume on Cherry Trail devices

PATCH In review Hans de Goede 4 participants 11 messages Oct 11, 2018 – Oct 12, 2018

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

v4 cover pwm: lpss: Check PWM powerstate after resume on Cherry Trail devices

Hi All,

As requested here is a new version of this series adding a patch exporting acpi_device_get_power() and using that.

This means that the PWM changes now rely on an ACPI patch.

I've 3 other LPSS PWM changes in flight which all have already been acked or reviewed by Andy. One option would be to merge the entire series + the 3 patches through Rafael's PM tree. If we want to go that route I can send out a new series with all 6 patches together to Rafael.

Another option would be to merge the single line ACPI change through the PWM tree.

Rafael, Thierry how do you want to proceed with this? And if you prefer for the changes to go through the other tree, may we have your Acked-by for the patch(es) which would normally go in through your own tree ?

Regards,

Hans

#1 lore raw

PATCH v4 1/3 ACPI / PM: Export acpi_device_get_power() for use by modular build drivers

Export acpi_device_get_power() for use by modular build drivers.

---
Changes in v4:
-New patch in v4 of this patch-set

---
 drivers/acpi/device_pm.c | 1 +
 1 file changed, 1 insertion(+)
drivers/acpi/device_pm.c +1 -0
index a7c2673ffd36..824ae985ad93 100644
@@ -126,6 +126,7 @@ int acpi_device_get_power(struct acpi_device *device, int *state)
 
 	return 0;
 }
+EXPORT_SYMBOL(acpi_device_get_power);
 
 static int acpi_dev_pm_explicit_set(struct acpi_device *adev, int state)
 {
Signature · 1 line
2.19.0
reviewer Reviewed-by on v4 1/3
#2 lore raw

Re: [PATCH v4 1/3] ACPI / PM: Export acpi_device_get_power() for use by modular build drivers

Quoted 2 lines · Hans de Goede

Export acpi_device_get_power() for use by modular build drivers.

non-GPL export is used to be consistent with setter counterpart, right?

Quoted 23 lines

---
Changes in v4:

(quoted diff) +0 -2
-New patch in v4 of this patch-set
---
 drivers/acpi/device_pm.c | 1 +
 1 file changed, 1 insertion(+)
drivers/acpi/device_pm.c +1 -0
index a7c2673ffd36..824ae985ad93 100644
@@ -126,6 +126,7 @@ int acpi_device_get_power(struct acpi_device *device, int *state)
 
 	return 0;
 }
+EXPORT_SYMBOL(acpi_device_get_power);
 
 static int acpi_dev_pm_explicit_set(struct acpi_device *adev, int state)
 {
Signature · 1 line
2.19.0
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#3 lore raw

Re: [PATCH v4 1/3] ACPI / PM: Export acpi_device_get_power() for use by modular build drivers

Hi,

Quoted 5 lines · Andy Shevchenko
Quoted 2 lines · Hans de Goede

Export acpi_device_get_power() for use by modular build drivers.

non-GPL export is used to be consistent with setter counterpart, right?

Right.

Quoted 1 line

Thanks.

Regards,

Hans

Quoted 25 lines
Quoted 24 lines

---
Changes in v4:

(quoted diff) +0 -2
-New patch in v4 of this patch-set
---
  drivers/acpi/device_pm.c | 1 +
  1 file changed, 1 insertion(+)
drivers/acpi/device_pm.c +1 -0
index a7c2673ffd36..824ae985ad93 100644
@@ -126,6 +126,7 @@ int acpi_device_get_power(struct acpi_device *device, int *state)
 
  	return 0;
  }
+EXPORT_SYMBOL(acpi_device_get_power);
 
  static int acpi_dev_pm_explicit_set(struct acpi_device *adev, int state)
  {
Signature · 1 line
2.19.0
Acked-by on v4 1/3
#4 lore raw

Re: [PATCH v4 1/3] ACPI / PM: Export acpi_device_get_power() for use by modular build drivers

In reply to Hans de Goede
Quoted 16 lines · CEST Hans de Goede

Export acpi_device_get_power() for use by modular build drivers.

---
Changes in v4:

(quoted diff) +0 -2
-New patch in v4 of this patch-set
---
 drivers/acpi/device_pm.c | 1 +
 1 file changed, 1 insertion(+)
drivers/acpi/device_pm.c +0 -0
index a7c2673ffd36..824ae985ad93 100644

drivers/acpi/device_pm.c int acpi_device_get_power(struct acpi_device *device, int *state) · line 132

Quoted 7 lines
drivers/acpi/device_pm.c +1 -0
@@ -126,6 +126,7 @@ int acpi_device_get_power(struct acpi_device *device, int *state)
 
 	return 0;
 }
+EXPORT_SYMBOL(acpi_device_get_power);
 
 static int acpi_dev_pm_explicit_set(struct acpi_device *adev, int state)
 {
#5 lore raw

Re: [PATCH v4 1/3] ACPI / PM: Export acpi_device_get_power() for use by modular build drivers

Hi,

Quoted 21 lines · Rafael J. Wysocki
Quoted 20 lines · CEST Hans de Goede

Export acpi_device_get_power() for use by modular build drivers.

---
Changes in v4:

(quoted diff) +0 -2
-New patch in v4 of this patch-set
---
  drivers/acpi/device_pm.c | 1 +
  1 file changed, 1 insertion(+)
drivers/acpi/device_pm.c +1 -0
index a7c2673ffd36..824ae985ad93 100644
@@ -126,6 +126,7 @@ int acpi_device_get_power(struct acpi_device *device, int *state)
 
  	return 0;
  }
+EXPORT_SYMBOL(acpi_device_get_power);
 
Quoted 5 lines

drivers/acpi/device_pm.c int acpi_device_get_power(struct acpi_device *device, int *state) · line 132

Quoted 3 lines
drivers/acpi/device_pm.c +0 -0
  static int acpi_dev_pm_explicit_set(struct acpi_device *adev, int state)
  {

Does this mean that you are ok with taking this patch upstream through Thierry's PWM tree ?

Thierry, do you want me to resend the latest version of all my outstanding PWM LPSS patches as a single series?

Regards,

Hans

#6 lore raw

Re: [PATCH v4 1/3] ACPI / PM: Export acpi_device_get_power() for use by modular build drivers

Quoted 25 lines · Hans de Goede

Hi,

Quoted 21 lines · Rafael J. Wysocki
Quoted 20 lines · CEST Hans de Goede

Export acpi_device_get_power() for use by modular build drivers.

---
Changes in v4:

(quoted diff) +0 -2
-New patch in v4 of this patch-set
---
  drivers/acpi/device_pm.c | 1 +
  1 file changed, 1 insertion(+)
drivers/acpi/device_pm.c +1 -0
index a7c2673ffd36..824ae985ad93 100644
@@ -126,6 +126,7 @@ int acpi_device_get_power(struct acpi_device *device, int *state)
 
     return 0;
  }
+EXPORT_SYMBOL(acpi_device_get_power);
 
Quoted 8 lines
Quoted 5 lines

drivers/acpi/device_pm.c int acpi_device_get_power(struct acpi_device *device, int *state) · line 132

Quoted 3 lines
drivers/acpi/device_pm.c +0 -0
  static int acpi_dev_pm_explicit_set(struct acpi_device *adev, int state)
  {

Does this mean that you are ok with taking this patch upstream through Thierry's PWM tree ?

Yes, it does.

Thanks,
Rafael

#7 lore raw

PATCH v4 2/3 pwm: lpss: Move struct pwm_lpss_chip definition to the header file

In reply to Hans de Goede

Move struct pwm_lpss_chip definition from pwm-lpss.c to pwm-lpss.h, so that the pci/platform drivers can access the info member (struct pwm_lpss_boardinfo *).

This is a preparation patch for adding platform specific quirks, which the drivers need access to, to pwm_lpss_boardinfo.

---
Changes in v4:
-No changes in v4 of this patch-set

Changes in v3:
-There was no v3, but I accidentally put v3 in the Subject of the v2
patches, so lets skip v3

Changes in v2:
-No changes in v2 of this patch-set

---
 drivers/pwm/pwm-lpss.c | 9 ---------
 drivers/pwm/pwm-lpss.h | 9 ++++++++-
 2 files changed, 8 insertions(+), 10 deletions(-)
drivers/pwm/pwm-lpss.c +0 -9
index 4721a264bac2..e602835fd6de 100644
@@ -32,15 +32,6 @@
 /* Size of each PWM register space if multiple */
 #define PWM_SIZE			0x400
 
-#define MAX_PWMS			4
-
-struct pwm_lpss_chip {
-	struct pwm_chip chip;
-	void __iomem *regs;
-	const struct pwm_lpss_boardinfo *info;
-	u32 saved_ctrl[MAX_PWMS];
-};
-
 static inline struct pwm_lpss_chip *to_lpwm(struct pwm_chip *chip)
 {
 	return container_of(chip, struct pwm_lpss_chip, chip);
drivers/pwm/pwm-lpss.h +8 -1
index 7a4238ad1fcb..8f029ed263af 100644
@@ -16,7 +16,14 @@
 #include <linux/device.h>
 #include <linux/pwm.h>
 
-struct pwm_lpss_chip;
+#define MAX_PWMS			4
+
+struct pwm_lpss_chip {
+	struct pwm_chip chip;
+	void __iomem *regs;
+	const struct pwm_lpss_boardinfo *info;
+	u32 saved_ctrl[MAX_PWMS];
+};
 
 struct pwm_lpss_boardinfo {
 	unsigned long clk_rate;
Signature · 1 line
2.19.0
reviewer Reviewed-by on v4 2/3
#8 lore raw

Re: [PATCH v4 2/3] pwm: lpss: Move struct pwm_lpss_chip definition to the header file

Quoted 6 lines · Hans de Goede

Move struct pwm_lpss_chip definition from pwm-lpss.c to pwm-lpss.h, so that the pci/platform drivers can access the info member (struct pwm_lpss_boardinfo *).

This is a preparation patch for adding platform specific quirks, which the drivers need access to, to pwm_lpss_boardinfo.

Quoted 60 lines

---
Changes in v4:
-No changes in v4 of this patch-set

Changes in v3:
-There was no v3, but I accidentally put v3 in the Subject of the v2
patches, so lets skip v3

Changes in v2:

(quoted diff) +0 -2
-No changes in v2 of this patch-set
---
 drivers/pwm/pwm-lpss.c | 9 ---------
 drivers/pwm/pwm-lpss.h | 9 ++++++++-
 2 files changed, 8 insertions(+), 10 deletions(-)
drivers/pwm/pwm-lpss.c +0 -9
index 4721a264bac2..e602835fd6de 100644
@@ -32,15 +32,6 @@
 /* Size of each PWM register space if multiple */
 #define PWM_SIZE			0x400
 
-#define MAX_PWMS			4
-
-struct pwm_lpss_chip {
-	struct pwm_chip chip;
-	void __iomem *regs;
-	const struct pwm_lpss_boardinfo *info;
-	u32 saved_ctrl[MAX_PWMS];
-};
-
 static inline struct pwm_lpss_chip *to_lpwm(struct pwm_chip *chip)
 {
 	return container_of(chip, struct pwm_lpss_chip, chip);
drivers/pwm/pwm-lpss.h +8 -1
index 7a4238ad1fcb..8f029ed263af 100644
@@ -16,7 +16,14 @@
 #include <linux/device.h>
 #include <linux/pwm.h>
 
-struct pwm_lpss_chip;
+#define MAX_PWMS			4
+
+struct pwm_lpss_chip {
+	struct pwm_chip chip;
+	void __iomem *regs;
+	const struct pwm_lpss_boardinfo *info;
+	u32 saved_ctrl[MAX_PWMS];
+};
 
 struct pwm_lpss_boardinfo {
 	unsigned long clk_rate;
Signature · 1 line
2.19.0
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#9 lore raw

PATCH v4 3/3 pwm: lpss: Check PWM powerstate after resume on Cherry Trail devices

In reply to Hans de Goede

The _PS0 method for the integrated graphics on some Cherry Trail devices (observed on a HP Pavilion X2 10-p0XX) turns on the PWM chip (puts it in D0), causing an inconsistency between the state the pm-core thinks it is in (left runtime suspended as it was before the suspend/resume) and the state it actually is in.

Interestingly enough this is done on a device where the pwm controller is not used for the backlight at all, since it uses an eDP panel. On devices where the PWM is used this is not a problem since we will resume it ourselves anyways.

This inconsistency causes us to never suspend the pwm controller again, which causes the device to not be able to reach S0ix states when suspended.

This commit adds a resume-complete handler, which when we think the device is still run-time suspended checks the actual power-state and if necessary updates the rpm-core's internal state.

This fixes the Pavilion X2 10-p0XX not reaching S0ix states when suspended.

---
Changes in v4:
-Use acpi_device_get_power() instead of manually calling _PSC

Changes in v3:
-There was no v3, but I accidentally put v3 in the Subject of the v2
patches, so lets skip v3

Changes in v2:
-Do the pm_runtime_en/disable before/after checking the power-state

---
 drivers/pwm/pwm-lpss-platform.c | 25 ++++++++++++++++++++++---
 drivers/pwm/pwm-lpss.h          |  2 ++
 2 files changed, 24 insertions(+), 3 deletions(-)
drivers/pwm/pwm-lpss-platform.c +22 -3
index 7304f36ee715..b6edf8af26cc 100644
@@ -30,6 +30,7 @@ static const struct pwm_lpss_boardinfo pwm_lpss_bsw_info = {
 	.clk_rate = 19200000,
 	.npwm = 1,
 	.base_unit_bits = 16,
+	.check_power_on_resume = true,
 };
 
 /* Broxton */
@@ -74,9 +75,27 @@ static int pwm_lpss_remove_platform(struct platform_device *pdev)
 	return pwm_lpss_remove(lpwm);
 }
 
-static SIMPLE_DEV_PM_OPS(pwm_lpss_platform_pm_ops,
-			 pwm_lpss_suspend,
-			 pwm_lpss_resume);
+static void pwm_lpss_complete(struct device *dev)
+{
+	struct pwm_lpss_chip *lpwm = dev_get_drvdata(dev);
+	int ret, state;
+
+	/* The PWM may be turned on by AML code, update our state to match */
+	if (pm_runtime_suspended(dev) && lpwm->info->check_power_on_resume) {
+		pm_runtime_disable(dev);
+
+		ret = acpi_device_get_power(ACPI_COMPANION(dev), &state);
+		if (ret == 0 && state == ACPI_STATE_D0)
+			pm_runtime_set_active(dev);
+
+		pm_runtime_enable(dev);
+	}
+}
+
+static const struct dev_pm_ops pwm_lpss_platform_pm_ops = {
+	.complete = pwm_lpss_complete,
+	SET_SYSTEM_SLEEP_PM_OPS(pwm_lpss_suspend, pwm_lpss_resume)
+};
 
 static const struct acpi_device_id pwm_lpss_acpi_match[] = {
 	{ "80860F09", (unsigned long)&pwm_lpss_byt_info },
drivers/pwm/pwm-lpss.h +2 -0
index 8f029ed263af..1a2575d25bea 100644
@@ -30,6 +30,8 @@ struct pwm_lpss_boardinfo {
 	unsigned int npwm;
 	unsigned long base_unit_bits;
 	bool bypass;
+	/* Some devices have AML code messing with the state underneath us */
+	bool check_power_on_resume;
 };
 
 struct pwm_lpss_chip *pwm_lpss_probe(struct device *dev, struct resource *r,
Signature · 1 line
2.19.0
reviewer Reviewed-by on v4 3/3
#10 lore raw
Quoted 11 lines · Hans de Goede

The _PS0 method for the integrated graphics on some Cherry Trail devices (observed on a HP Pavilion X2 10-p0XX) turns on the PWM chip (puts it in D0), causing an inconsistency between the state the pm-core thinks it is in (left runtime suspended as it was before the suspend/resume) and the state it actually is in.

Interestingly enough this is done on a device where the pwm controller is not used for the backlight at all, since it uses an eDP panel. On devices where the PWM is used this is not a problem since we will resume it ourselves anyways.

Quoted 8 lines

This inconsistency causes us to never suspend the pwm controller again, which causes the device to not be able to reach S0ix states when suspended.

This commit adds a resume-complete handler, which when we think the device is still run-time suspended checks the actual power-state and if necessary updates the rpm-core's internal state.

This fixes the Pavilion X2 10-p0XX not reaching S0ix states when suspended.

Quoted 76 lines

---
Changes in v4:
-Use acpi_device_get_power() instead of manually calling _PSC

Changes in v3:
-There was no v3, but I accidentally put v3 in the Subject of the v2
patches, so lets skip v3

Changes in v2:

(quoted diff) +0 -2
-Do the pm_runtime_en/disable before/after checking the power-state
---
 drivers/pwm/pwm-lpss-platform.c | 25 ++++++++++++++++++++++---
 drivers/pwm/pwm-lpss.h          |  2 ++
 2 files changed, 24 insertions(+), 3 deletions(-)
drivers/pwm/pwm-lpss-platform.c +22 -3
index 7304f36ee715..b6edf8af26cc 100644
@@ -30,6 +30,7 @@ static const struct pwm_lpss_boardinfo pwm_lpss_bsw_info = {
 	.clk_rate = 19200000,
 	.npwm = 1,
 	.base_unit_bits = 16,
+	.check_power_on_resume = true,
 };
 
 /* Broxton */
@@ -74,9 +75,27 @@ static int pwm_lpss_remove_platform(struct platform_device *pdev)
 	return pwm_lpss_remove(lpwm);
 }
 
-static SIMPLE_DEV_PM_OPS(pwm_lpss_platform_pm_ops,
-			 pwm_lpss_suspend,
-			 pwm_lpss_resume);
+static void pwm_lpss_complete(struct device *dev)
+{
+	struct pwm_lpss_chip *lpwm = dev_get_drvdata(dev);
+	int ret, state;
+
+	/* The PWM may be turned on by AML code, update our state to match */
+	if (pm_runtime_suspended(dev) && lpwm->info->check_power_on_resume) {
+		pm_runtime_disable(dev);
+
+		ret = acpi_device_get_power(ACPI_COMPANION(dev), &state);
+		if (ret == 0 && state == ACPI_STATE_D0)
+			pm_runtime_set_active(dev);
+
+		pm_runtime_enable(dev);
+	}
+}
+
+static const struct dev_pm_ops pwm_lpss_platform_pm_ops = {
+	.complete = pwm_lpss_complete,
+	SET_SYSTEM_SLEEP_PM_OPS(pwm_lpss_suspend, pwm_lpss_resume)
+};
 
 static const struct acpi_device_id pwm_lpss_acpi_match[] = {
 	{ "80860F09", (unsigned long)&pwm_lpss_byt_info },
drivers/pwm/pwm-lpss.h +2 -0
index 8f029ed263af..1a2575d25bea 100644
@@ -30,6 +30,8 @@ struct pwm_lpss_boardinfo {
 	unsigned int npwm;
 	unsigned long base_unit_bits;
 	bool bypass;
+	/* Some devices have AML code messing with the state underneath us */
+	bool check_power_on_resume;
 };
 
 struct pwm_lpss_chip *pwm_lpss_probe(struct device *dev, struct resource *r,
Signature · 1 line
2.19.0
Signature · 2 lines
With Best Regards,
Andy Shevchenko