[RESEND PATCH v2 1/4] mfd: sec-irq: Do not enforce (incorrect) interrupt trigger type

[RESEND PATCH v2 1/4] mfd: sec-irq: Do not enforce (incorrect) interrupt trigger type

[Krzysztof Kozlowski]

From: Krzysztof Kozlowski <krzk@kernel.org>

Interrupt line can be configured on different hardware in different way,
even inverted.  Therefore driver should not enforce specific trigger
type - edge falling - but instead rely on Devicetree to configure it.

The Samsung PMIC drivers are used only on Devicetree boards.

Additionally, the PMIC datasheets describe the interrupt line as active
low with a requirement of acknowledge from the CPU therefore the edge
falling is not correct.

Marek Szyprowski reports that together with DTS change (proper level in
DTS) it fixes RTC alarm failure that he observed from time to time on
TM2e board.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com>

---

Rebased on https://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd.git/log/?h=for-mfd-next

Changes since v1:
1. Mention in commit msg that this fixes TM2e RTC alarm.
2. Add Marek's tested-by.
---
 drivers/mfd/sec-irq.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/mfd/sec-irq.c b/drivers/mfd/sec-irq.c
index e473c2fb42d5..f5f59fdc72fe 100644
--- a/drivers/mfd/sec-irq.c
+++ b/drivers/mfd/sec-irq.c
@@ -479,8 +479,7 @@ int sec_irq_init(struct sec_pmic_dev *sec_pmic)
 	}
 
 	ret = devm_regmap_add_irq_chip(sec_pmic->dev, sec_pmic->regmap_pmic,
-				       sec_pmic->irq,
-				       IRQF_TRIGGER_FALLING | IRQF_ONESHOT,
+				       sec_pmic->irq, IRQF_ONESHOT,
 				       0, sec_irq_chip, &sec_pmic->irq_data);
 	if (ret != 0) {
 		dev_err(sec_pmic->dev, "Failed to register IRQ chip: %d\n", ret);
-- 
2.27.0

[RESEND PATCH v2 2/4] mfd: max77686: Do not enforce (incorrect) interrupt trigger type

[Krzysztof Kozlowski]

From: Krzysztof Kozlowski <krzk@kernel.org>

Interrupt line can be configured on different hardware in different way,
even inverted.  Therefore driver should not enforce specific trigger
type - edge falling - but instead rely on Devicetree to configure it.

The Maxim 77686 datasheet describes the interrupt line as active low
with a requirement of acknowledge from the CPU therefore the edge
falling is not correct.

The interrupt line is shared between PMIC and RTC driver, so using level
sensitive interrupt is here especially important to avoid races.  With
an edge configuration in case if first PMIC signals interrupt followed
shortly after by the RTC, the interrupt might not be yet cleared/acked
thus the second one would not be noticed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>

---

Changes since v1:
1. Add ack.
---
 Documentation/devicetree/bindings/clock/maxim,max77686.txt | 4 ++--
 Documentation/devicetree/bindings/mfd/max77686.txt         | 2 +-
 Documentation/devicetree/bindings/regulator/max77686.txt   | 2 +-
 drivers/mfd/max77686.c                                     | 3 +--
 4 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/Documentation/devicetree/bindings/clock/maxim,max77686.txt b/Documentation/devicetree/bindings/clock/maxim,max77686.txt
index 3472b461ca93..c10849efb444 100644
--- a/Documentation/devicetree/bindings/clock/maxim,max77686.txt
+++ b/Documentation/devicetree/bindings/clock/maxim,max77686.txt
@@ -49,7 +49,7 @@ Example:
 		max77686: max77686@9 {
 			compatible = "maxim,max77686";
 			interrupt-parent = <&wakeup_eint>;
-			interrupts = <26 0>;
+			interrupts = <26 IRQ_TYPE_LEVEL_LOW>;
 			reg = <0x09>;
 			#clock-cells = <1>;
 
@@ -74,7 +74,7 @@ Example:
 		max77802: max77802@9 {
 			compatible = "maxim,max77802";
 			interrupt-parent = <&wakeup_eint>;
-			interrupts = <26 0>;
+			interrupts = <26 IRQ_TYPE_LEVEL_LOW>;
 			reg = <0x09>;
 			#clock-cells = <1>;
 
diff --git a/Documentation/devicetree/bindings/mfd/max77686.txt b/Documentation/devicetree/bindings/mfd/max77686.txt
index 42968b7144e0..4447d074894a 100644
--- a/Documentation/devicetree/bindings/mfd/max77686.txt
+++ b/Documentation/devicetree/bindings/mfd/max77686.txt
@@ -21,6 +21,6 @@ Example:
 	max77686: pmic@9 {
 		compatible = "maxim,max77686";
 		interrupt-parent = <&wakeup_eint>;
-		interrupts = <26 0>;
+		interrupts = <26 IRQ_TYPE_LEVEL_LOW>;
 		reg = <0x09>;
 	};
diff --git a/Documentation/devicetree/bindings/regulator/max77686.txt b/Documentation/devicetree/bindings/regulator/max77686.txt
index e9f7578ca09a..ff3d2dec8c4b 100644
--- a/Documentation/devicetree/bindings/regulator/max77686.txt
+++ b/Documentation/devicetree/bindings/regulator/max77686.txt
@@ -43,7 +43,7 @@ Example:
 	max77686: pmic@9 {
 		compatible = "maxim,max77686";
 		interrupt-parent = <&wakeup_eint>;
-		interrupts = <26 IRQ_TYPE_NONE>;
+		interrupts = <26 IRQ_TYPE_LEVEL_LOW>;
 		reg = <0x09>;
 
 		voltage-regulators {
diff --git a/drivers/mfd/max77686.c b/drivers/mfd/max77686.c
index 2ad554b921d9..f9e12ab2bc75 100644
--- a/drivers/mfd/max77686.c
+++ b/drivers/mfd/max77686.c
@@ -209,8 +209,7 @@ static int max77686_i2c_probe(struct i2c_client *i2c)
 
 	ret = devm_regmap_add_irq_chip(&i2c->dev, max77686->regmap,
 				       max77686->irq,
-				       IRQF_TRIGGER_FALLING | IRQF_ONESHOT |
-				       IRQF_SHARED, 0, irq_chip,
+				       IRQF_ONESHOT | IRQF_SHARED, 0, irq_chip,
 				       &max77686->irq_data);
 	if (ret < 0) {
 		dev_err(&i2c->dev, "failed to add PMIC irq chip: %d\n", ret);
-- 
2.27.0

Re: [RESEND PATCH v2 2/4] mfd: max77686: Do not enforce (incorrect) interrupt trigger type

[Lee Jones]

On Wed, 02 Jun 2021, Krzysztof Kozlowski wrote:

> From: Krzysztof Kozlowski <krzk@kernel.org>
> 
> Interrupt line can be configured on different hardware in different way,
> even inverted.  Therefore driver should not enforce specific trigger
> type - edge falling - but instead rely on Devicetree to configure it.
> 
> The Maxim 77686 datasheet describes the interrupt line as active low
> with a requirement of acknowledge from the CPU therefore the edge
> falling is not correct.
> 
> The interrupt line is shared between PMIC and RTC driver, so using level
> sensitive interrupt is here especially important to avoid races.  With
> an edge configuration in case if first PMIC signals interrupt followed
> shortly after by the RTC, the interrupt might not be yet cleared/acked
> thus the second one would not be noticed.
> 
> Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
> Acked-by: Rob Herring <robh@kernel.org>
> 
> ---
> 
> Changes since v1:
> 1. Add ack.
> ---
>  Documentation/devicetree/bindings/clock/maxim,max77686.txt | 4 ++--
>  Documentation/devicetree/bindings/mfd/max77686.txt         | 2 +-
>  Documentation/devicetree/bindings/regulator/max77686.txt   | 2 +-
>  drivers/mfd/max77686.c                                     | 3 +--
>  4 files changed, 5 insertions(+), 6 deletions(-)

Applied, thanks.

-- 
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog

[RESEND PATCH v2 3/4] mfd: max77693: Do not enforce (incorrect) interrupt trigger type

[Krzysztof Kozlowski]

From: Krzysztof Kozlowski <krzk@kernel.org>

Interrupt line can be configured on different hardware in different way,
even inverted.  Therefore driver should not enforce specific trigger
type - edge falling - but instead rely on Devicetree to configure it.

The Maxim 77693 datasheet describes the interrupt line as active low
with a requirement of acknowledge from the CPU therefore the edge
falling is not correct.

The interrupt line is shared between PMIC and RTC driver, so using level
sensitive interrupt is here especially important to avoid races.  With
an edge configuration in case if first PMIC signals interrupt followed
shortly after by the RTC, the interrupt might not be yet cleared/acked
thus the second one would not be noticed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>

---

Changes since v1:
1. Add ack.
---
 Documentation/devicetree/bindings/mfd/max77693.txt |  2 +-
 drivers/mfd/max77693.c                             | 12 ++++--------
 2 files changed, 5 insertions(+), 9 deletions(-)

diff --git a/Documentation/devicetree/bindings/mfd/max77693.txt b/Documentation/devicetree/bindings/mfd/max77693.txt
index 0ced96e16c16..1032df14498b 100644
--- a/Documentation/devicetree/bindings/mfd/max77693.txt
+++ b/Documentation/devicetree/bindings/mfd/max77693.txt
@@ -139,7 +139,7 @@ Example:
 		compatible = "maxim,max77693";
 		reg = <0x66>;
 		interrupt-parent = <&gpx1>;
-		interrupts = <5 2>;
+		interrupts = <5 IRQ_TYPE_LEVEL_LOW>;
 
 		regulators {
 			esafeout@1 {
diff --git a/drivers/mfd/max77693.c b/drivers/mfd/max77693.c
index 596ed85cab3b..4e6244e17559 100644
--- a/drivers/mfd/max77693.c
+++ b/drivers/mfd/max77693.c
@@ -222,8 +222,7 @@ static int max77693_i2c_probe(struct i2c_client *i2c,
 	}
 
 	ret = regmap_add_irq_chip(max77693->regmap, max77693->irq,
-				IRQF_ONESHOT | IRQF_SHARED |
-				IRQF_TRIGGER_FALLING, 0,
+				IRQF_ONESHOT | IRQF_SHARED, 0,
 				&max77693_led_irq_chip,
 				&max77693->irq_data_led);
 	if (ret) {
@@ -232,8 +231,7 @@ static int max77693_i2c_probe(struct i2c_client *i2c,
 	}
 
 	ret = regmap_add_irq_chip(max77693->regmap, max77693->irq,
-				IRQF_ONESHOT | IRQF_SHARED |
-				IRQF_TRIGGER_FALLING, 0,
+				IRQF_ONESHOT | IRQF_SHARED, 0,
 				&max77693_topsys_irq_chip,
 				&max77693->irq_data_topsys);
 	if (ret) {
@@ -242,8 +240,7 @@ static int max77693_i2c_probe(struct i2c_client *i2c,
 	}
 
 	ret = regmap_add_irq_chip(max77693->regmap, max77693->irq,
-				IRQF_ONESHOT | IRQF_SHARED |
-				IRQF_TRIGGER_FALLING, 0,
+				IRQF_ONESHOT | IRQF_SHARED, 0,
 				&max77693_charger_irq_chip,
 				&max77693->irq_data_chg);
 	if (ret) {
@@ -252,8 +249,7 @@ static int max77693_i2c_probe(struct i2c_client *i2c,
 	}
 
 	ret = regmap_add_irq_chip(max77693->regmap_muic, max77693->irq,
-				IRQF_ONESHOT | IRQF_SHARED |
-				IRQF_TRIGGER_FALLING, 0,
+				IRQF_ONESHOT | IRQF_SHARED, 0,
 				&max77693_muic_irq_chip,
 				&max77693->irq_data_muic);
 	if (ret) {
-- 
2.27.0

Re: [RESEND PATCH v2 3/4] mfd: max77693: Do not enforce (incorrect) interrupt trigger type

[Lee Jones]

On Wed, 02 Jun 2021, Krzysztof Kozlowski wrote:
65;6402;1c
> From: Krzysztof Kozlowski <krzk@kernel.org>
> 
> Interrupt line can be configured on different hardware in different way,
> even inverted.  Therefore driver should not enforce specific trigger
> type - edge falling - but instead rely on Devicetree to configure it.
> 
> The Maxim 77693 datasheet describes the interrupt line as active low
> with a requirement of acknowledge from the CPU therefore the edge
> falling is not correct.
> 
> The interrupt line is shared between PMIC and RTC driver, so using level
> sensitive interrupt is here especially important to avoid races.  With
> an edge configuration in case if first PMIC signals interrupt followed
> shortly after by the RTC, the interrupt might not be yet cleared/acked
> thus the second one would not be noticed.
> 
> Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
> Acked-by: Rob Herring <robh@kernel.org>
> 
> ---
> 
> Changes since v1:
> 1. Add ack.
> ---
>  Documentation/devicetree/bindings/mfd/max77693.txt |  2 +-
>  drivers/mfd/max77693.c                             | 12 ++++--------
>  2 files changed, 5 insertions(+), 9 deletions(-)

Applied, thanks.

-- 
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog

[RESEND PATCH v2 4/4] mfd: max14577: Do not enforce (incorrect) interrupt trigger type

[Krzysztof Kozlowski]

From: Krzysztof Kozlowski <krzk@kernel.org>

Interrupt line can be configured on different hardware in different way,
even inverted.  Therefore driver should not enforce specific trigger
type - edge falling - but instead rely on Devicetree to configure it.

The Maxim 14577/77836 datasheets describe the interrupt line as active
low with a requirement of acknowledge from the CPU therefore the edge
falling is not correct.

The interrupt line is shared between PMIC and charger driver, so using
level sensitive interrupt is here especially important to avoid races.
With an edge configuration in case if first PMIC signals interrupt
followed shortly after by the RTC, the interrupt might not be yet
cleared/acked thus the second one would not be noticed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>

---

Changes since v1:
1. Add Ack
---
 Documentation/devicetree/bindings/mfd/max14577.txt | 4 ++--
 drivers/mfd/max14577.c                             | 6 +++---
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/Documentation/devicetree/bindings/mfd/max14577.txt b/Documentation/devicetree/bindings/mfd/max14577.txt
index 92070b346756..be11943a0560 100644
--- a/Documentation/devicetree/bindings/mfd/max14577.txt
+++ b/Documentation/devicetree/bindings/mfd/max14577.txt
@@ -71,7 +71,7 @@ max14577@25 {
 	compatible = "maxim,max14577";
 	reg = <0x25>;
 	interrupt-parent = <&gpx1>;
-	interrupts = <5 IRQ_TYPE_NONE>;
+	interrupts = <5 IRQ_TYPE_LEVEL_LOW>;
 
 	muic: max14577-muic {
 		compatible = "maxim,max14577-muic";
@@ -106,7 +106,7 @@ max77836@25 {
 	compatible = "maxim,max77836";
 	reg = <0x25>;
 	interrupt-parent = <&gpx1>;
-	interrupts = <5 IRQ_TYPE_NONE>;
+	interrupts = <5 IRQ_TYPE_LEVEL_LOW>;
 
 	muic: max77836-muic {
 		compatible = "maxim,max77836-muic";
diff --git a/drivers/mfd/max14577.c b/drivers/mfd/max14577.c
index be185e9d5f16..6c487fa14e9c 100644
--- a/drivers/mfd/max14577.c
+++ b/drivers/mfd/max14577.c
@@ -332,7 +332,7 @@ static int max77836_init(struct max14577 *max14577)
 	}
 
 	ret = regmap_add_irq_chip(max14577->regmap_pmic, max14577->irq,
-			IRQF_TRIGGER_FALLING | IRQF_ONESHOT | IRQF_SHARED,
+			IRQF_ONESHOT | IRQF_SHARED,
 			0, &max77836_pmic_irq_chip,
 			&max14577->irq_data_pmic);
 	if (ret != 0) {
@@ -418,14 +418,14 @@ static int max14577_i2c_probe(struct i2c_client *i2c,
 		irq_chip = &max77836_muic_irq_chip;
 		mfd_devs = max77836_devs;
 		mfd_devs_size = ARRAY_SIZE(max77836_devs);
-		irq_flags = IRQF_TRIGGER_FALLING | IRQF_ONESHOT | IRQF_SHARED;
+		irq_flags = IRQF_ONESHOT | IRQF_SHARED;
 		break;
 	case MAXIM_DEVICE_TYPE_MAX14577:
 	default:
 		irq_chip = &max14577_irq_chip;
 		mfd_devs = max14577_devs;
 		mfd_devs_size = ARRAY_SIZE(max14577_devs);
-		irq_flags = IRQF_TRIGGER_FALLING | IRQF_ONESHOT;
+		irq_flags = IRQF_ONESHOT;
 		break;
 	}
 
-- 
2.27.0

Re: [RESEND PATCH v2 4/4] mfd: max14577: Do not enforce (incorrect) interrupt trigger type

[Lee Jones]

On Wed, 02 Jun 2021, Krzysztof Kozlowski wrote:

> From: Krzysztof Kozlowski <krzk@kernel.org>
> 
> Interrupt line can be configured on different hardware in different way,
> even inverted.  Therefore driver should not enforce specific trigger
> type - edge falling - but instead rely on Devicetree to configure it.
> 
> The Maxim 14577/77836 datasheets describe the interrupt line as active
> low with a requirement of acknowledge from the CPU therefore the edge
> falling is not correct.
> 
> The interrupt line is shared between PMIC and charger driver, so using
> level sensitive interrupt is here especially important to avoid races.
> With an edge configuration in case if first PMIC signals interrupt
> followed shortly after by the RTC, the interrupt might not be yet
> cleared/acked thus the second one would not be noticed.
> 
> Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
> Acked-by: Rob Herring <robh@kernel.org>
> 
> ---
> 
> Changes since v1:
> 1. Add Ack
> ---
>  Documentation/devicetree/bindings/mfd/max14577.txt | 4 ++--
>  drivers/mfd/max14577.c                             | 6 +++---
>  2 files changed, 5 insertions(+), 5 deletions(-)

Applied, thanks.

-- 
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog

Re: [RESEND PATCH v2 1/4] mfd: sec-irq: Do not enforce (incorrect) interrupt trigger type

[Krzysztof Kozlowski]

On 02/06/2021 13:04, Krzysztof Kozlowski wrote:
> From: Krzysztof Kozlowski <krzk@kernel.org>
> 
> Interrupt line can be configured on different hardware in different way,
> even inverted.  Therefore driver should not enforce specific trigger
> type - edge falling - but instead rely on Devicetree to configure it.
> 
> The Samsung PMIC drivers are used only on Devicetree boards.
> 
> Additionally, the PMIC datasheets describe the interrupt line as active
> low with a requirement of acknowledge from the CPU therefore the edge
> falling is not correct.
> 
> Marek Szyprowski reports that together with DTS change (proper level in
> DTS) it fixes RTC alarm failure that he observed from time to time on
> TM2e board.
> 
> Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com>
> 
> ---
> 
> Rebased on https://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd.git/log/?h=for-mfd-next
> 
> Changes since v1:
> 1. Mention in commit msg that this fixes TM2e RTC alarm.
> 2. Add Marek's tested-by.
> ---
>  drivers/mfd/sec-irq.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)

Dear Lee,

This was already a resend. Any comments from your side?

Best regards,
Krzysztof

Re: [RESEND PATCH v2 1/4] mfd: sec-irq: Do not enforce (incorrect) interrupt trigger type

[Lee Jones]

On Wed, 02 Jun 2021, Krzysztof Kozlowski wrote:

> From: Krzysztof Kozlowski <krzk@kernel.org>
> 
> Interrupt line can be configured on different hardware in different way,
> even inverted.  Therefore driver should not enforce specific trigger
> type - edge falling - but instead rely on Devicetree to configure it.
> 
> The Samsung PMIC drivers are used only on Devicetree boards.
> 
> Additionally, the PMIC datasheets describe the interrupt line as active
> low with a requirement of acknowledge from the CPU therefore the edge
> falling is not correct.
> 
> Marek Szyprowski reports that together with DTS change (proper level in
> DTS) it fixes RTC alarm failure that he observed from time to time on
> TM2e board.
> 
> Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com>
> 
> ---
> 
> Rebased on https://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd.git/log/?h=for-mfd-next
> 
> Changes since v1:
> 1. Mention in commit msg that this fixes TM2e RTC alarm.
> 2. Add Marek's tested-by.
> ---
>  drivers/mfd/sec-irq.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)

Applied, thanks.

-- 
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog