intel-hid: intel-vbtn: Suspend-related fix and update

PATCH In review Rafael J. Wysocki 3 participants 5 messages Aug 8, 2019 – Aug 9, 2019

Subsystems (2)

ACPI x86 platform drivers

lore

Share & tools

lore

Rafael J. Wysocki

author

2019-08-08 08:40 UTC #0 lore raw

cover intel-hid: intel-vbtn: Suspend-related fix and update

Hi,

These two patches fix a minor issue related to system suspend in the intel-hid and intel-vbtn drivers and update the suspend/resume handling in intel-hid to reduce special-casing in it somewhat.

Please refer to the changelogs for details.

Thanks,
Rafael

Rafael J. Wysocki

author

2019-08-08 08:42 UTC #1 lore raw

PATCH 1/2 intel-hid: intel-vbtn: Avoid leaking wakeup_mode set

From: Rafael J. Wysocki <[email protected]>

Both intel-hid and intel-vbtn set a wakeup_mode flag causing them to behave in a special way during system suspend and while suspended in their "prepare" PM callbacks and clear it in their "resume" PM callbacks. That may cause the wakeup_mode flag to remain set after a system suspend failure (if some other driver's "suspend" callback returns an error before the "suspend" callback of either intel-hid or intel-vbtn is invoked).

After commit 10a08fd65ec1 ("ACPI: PM: Set up EC GPE for system wakeup from drivers that need it") that also affects the "wakeup mask" bit of the EC GPE, which may not be cleared after a failing system suspend.

Fix this issue by adding "complete" PM callbacks to intel-hid and intel-vbtn to clear the wakeup_mode flag and the "wakeup mask" bit of the EC GPE if they have not been cleared earlier.

Fixes 10a08fd65ec1 ("ACPI: PM: Set up EC GPE for system wakeup from drivers that need it")

Signed-off-by Rafael J. Wysocki <[email protected]>

---

The commit pointed to by the Fixes: tag above is in linux-next only ATM.

The problem had been there before that commit, but it wasn't fixed by it and the fix depends on that commit.

---
 drivers/platform/x86/intel-hid.c  |   17 ++++++++++++-----
 drivers/platform/x86/intel-vbtn.c |   12 +++++++++---
 2 files changed, 21 insertions(+), 8 deletions(-)

drivers/platform/x86/intel-hid.c +12 -5

Index: linux-pm/drivers/platform/x86/intel-hid.c
===================================================================
@@ -262,6 +262,16 @@ static int intel_hid_pm_prepare(struct d
 	return 0;
 }
 
+static void intel_hid_pm_complete(struct device *device)
+{
+	struct intel_hid_priv *priv = dev_get_drvdata(device);
+
+	if (priv->wakeup_mode) {
+		acpi_ec_set_gpe_wake_mask(ACPI_GPE_DISABLE);
+		priv->wakeup_mode = false;
+	}
+}
+
 static int intel_hid_pl_suspend_handler(struct device *device)
 {
 	if (pm_suspend_via_firmware()) {

@@ -273,12 +283,8 @@ static int intel_hid_pl_suspend_handler(
 
 static int intel_hid_pl_resume_handler(struct device *device)
 {
-	if (device_may_wakeup(device)) {
-		struct intel_hid_priv *priv = dev_get_drvdata(device);
+	intel_hid_pm_complete(device);
 
-		acpi_ec_set_gpe_wake_mask(ACPI_GPE_DISABLE);
-		priv->wakeup_mode = false;
-	}
 	if (pm_resume_via_firmware()) {
 		intel_hid_set_enable(device, true);
 		intel_button_array_enable(device, true);

@@ -288,6 +294,7 @@ static int intel_hid_pl_resume_handler(s
 
 static const struct dev_pm_ops intel_hid_pl_pm_ops = {
 	.prepare = intel_hid_pm_prepare,
+	.complete = intel_hid_pm_complete,
 	.freeze  = intel_hid_pl_suspend_handler,
 	.thaw  = intel_hid_pl_resume_handler,
 	.restore  = intel_hid_pl_resume_handler,

drivers/platform/x86/intel-vbtn.c +9 -3

Index: linux-pm/drivers/platform/x86/intel-vbtn.c
===================================================================
@@ -210,19 +210,25 @@ static int intel_vbtn_pm_prepare(struct
 	return 0;
 }
 
-static int intel_vbtn_pm_resume(struct device *dev)
+static void intel_vbtn_pm_complete(struct device *dev)
 {
-	if (device_may_wakeup(dev)) {
-		struct intel_vbtn_priv *priv = dev_get_drvdata(dev);
+	struct intel_vbtn_priv *priv = dev_get_drvdata(dev);
 
+	if (priv->wakeup_mode) {
 		acpi_ec_set_gpe_wake_mask(ACPI_GPE_DISABLE);
 		priv->wakeup_mode = false;
 	}
+}
+
+static int intel_vbtn_pm_resume(struct device *dev)
+{
+	intel_vbtn_pm_complete(dev);
 	return 0;
 }
 
 static const struct dev_pm_ops intel_vbtn_pm_ops = {
 	.prepare = intel_vbtn_pm_prepare,
+	.complete = intel_vbtn_pm_complete,
 	.resume = intel_vbtn_pm_resume,
 	.restore = intel_vbtn_pm_resume,
 	.thaw = intel_vbtn_pm_resume,

Rafael J. Wysocki

author

2019-08-08 08:44 UTC #2 lore raw

PATCH 2/2 intel-hid: Disable button array during suspend-to-idle

In reply to Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Notice that intel_button_array_enable() never disables the power button which is the only one needed to wake up the system from suspend-to-idle, so it can be safely called during suspend-to-idle as well as during "regular" system suspend, and rearrange the code in the driver's "suspend" and "resume" callbacks accordingly.

While at it, use pm_suspend_no_platform() to check if the current suspend-resume cycle is suspend-to-idle, as that is the only case when the device should be enabled while suspended.

Signed-off-by Rafael J. Wysocki <[email protected]>

---
 drivers/platform/x86/intel-hid.c |   13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

drivers/platform/x86/intel-hid.c +7 -6

Index: linux-pm/drivers/platform/x86/intel-hid.c
===================================================================
@@ -274,10 +274,11 @@ static void intel_hid_pm_complete(struct
 
 static int intel_hid_pl_suspend_handler(struct device *device)
 {
-	if (pm_suspend_via_firmware()) {
+	intel_button_array_enable(device, false);
+
+	if (!pm_suspend_no_platform())
 		intel_hid_set_enable(device, false);
-		intel_button_array_enable(device, false);
-	}
+
 	return 0;
 }

@@ -285,10 +286,10 @@ static int intel_hid_pl_resume_handler(s
 {
 	intel_hid_pm_complete(device);
 
-	if (pm_resume_via_firmware()) {
+	if (!pm_suspend_no_platform())
 		intel_hid_set_enable(device, true);
-		intel_button_array_enable(device, true);
-	}
+
+	intel_button_array_enable(device, true);
 	return 0;
 }

Andy Shevchenko

Acked-by on the v1 series

2019-08-08 11:19 UTC #3 lore raw

In reply to Rafael J. Wysocki

Quoted 6 lines · Rafael J. Wysocki

Hi,
These two patches fix a minor issue related to system suspend in the intel-hid and intel-vbtn drivers and update the suspend/resume handling in intel-hid to reduce special-casing in it somewhat.

AFAIR the original patches go via other than PDx86 tree. Thus, while patches are looking good to me,

Acked-by Andy Shevchenko <[email protected]>

Quoted 1 line

Please refer to the changelogs for details.

Signature · 2 lines

With Best Regards,
Andy Shevchenko

Rafael J. Wysocki

maintainer

2019-08-09 08:15 UTC #4 lore raw

Quoted 10 lines · Andy Shevchenko

Quoted 6 lines · Rafael J. Wysocki
Hi,
These two patches fix a minor issue related to system suspend in the intel-hid and intel-vbtn drivers and update the suspend/resume handling in intel-hid to reduce special-casing in it somewhat.
AFAIR the original patches go via other than PDx86 tree.

That's correct.

Quoted 3 lines

Thus, while patches are looking good to me,
Acked-by Andy Shevchenko <[email protected]>

Thanks!