Skip to contentSkip to conversation

watchdog: gpio: Add "keep-armed-on-close" feature

PATCH Superseded Sylvain Lemieux 2 participants 8 messages Dec 7, 2016 – Mar 13, 2017

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 watchdog: gpio: Add "keep-armed-on-close" feature

From: Sylvain Lemieux <[email protected]>

There is a need to allow a grace period after the watchdog software client has closed. It could be used for syncing the filesystem or allow graceful termination while still providing a hardware reset in case the system has hung.

The "always-running" configuration from device-tree does not provide this since it will automatically keep the hardware watchdog alive as soon as the software client closes (i.e. keep toggling the GPIO line regardless of the state of the soft part of the watchdog).

The "keep-armed-on-close" member in the GPIO watchdog implementation indicates if an expired timeout should cause a reset.

This patch add a new "keep-armed-on-close" device-tree configuration that will keep the watchdog "armed" until the next timeout period after a close. During this period, the hardware watchdog is kept alive.

A software watchdog client that wants to provide a grace period, before a hard reset, can set the timeout before properly closing.

---
 Documentation/devicetree/bindings/watchdog/gpio-wdt.txt | 3 +++
 drivers/watchdog/gpio_wdt.c                             | 6 ++++++
 2 files changed, 9 insertions(+)
Documentation/devicetree/bindings/watchdog/gpio-wdt.txt +3 -0
index 83d2814..2669735 100644
@@ -17,6 +17,9 @@ Optional Properties:
 - always-running: If the watchdog timer cannot be disabled, add this flag to
   have the driver keep toggling the signal without a client. It will only cease
   to toggle the signal when the device is open and the timeout elapsed.
+- keep-armed-on-close: if the watchdog timer need to keep toggling the signal
+  when close, until the timeout elapsed, add this flag to have the driver
+  keep toggling the signal, until the timeout elapsed.
 - timeout-sec: Contains the watchdog timeout in seconds.
 - start-at-init: Start kicking watchdog as soon as driver is loaded.
 
drivers/watchdog/gpio_wdt.c +6 -0
index ef9ab91..ba9091a 100644
@@ -32,6 +32,7 @@ struct gpio_wdt_priv {
 	bool			active_low;
 	bool			state;
 	bool			always_running;
+	bool			keep_armed_on_close;
 	bool			armed;
 	unsigned int		hw_algo;
 	unsigned int		hw_margin;
@@ -72,6 +73,9 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 {
 	struct gpio_wdt_priv *priv = watchdog_get_drvdata(wdd);
 
+	if(priv->keep_armed_on_close)
+		return 0;
+
 	priv->armed = false;
 	if (!priv->always_running) {
 		mod_timer(&priv->timer, 0);
@@ -210,6 +214,8 @@ static int gpio_wdt_probe(struct platform_device *pdev)
 
 	priv->always_running = of_property_read_bool(pdev->dev.of_node,
 						     "always-running");
+	priv->keep_armed_on_close = of_property_read_bool(pdev->dev.of_node,
+						 "keep-armed-on-close");
 
 	watchdog_set_drvdata(&priv->wdd, priv);
 
Signature · 1 line
1.8.3.1
maintainer
#1 lore raw
Quoted 15 lines · Sylvain Lemieux

From: Sylvain Lemieux <[email protected]>

There is a need to allow a grace period after the watchdog software client has closed. It could be used for syncing the filesystem or allow graceful termination while still providing a hardware reset in case the system has hung.

The "always-running" configuration from device-tree does not provide this since it will automatically keep the hardware watchdog alive as soon as the software client closes (i.e. keep toggling the GPIO line regardless of the state of the soft part of the watchdog).

The "keep-armed-on-close" member in the GPIO watchdog implementation indicates if an expired timeout should cause a reset.

Quoted 8 lines

This patch add a new "keep-armed-on-close" device-tree configuration that will keep the watchdog "armed" until the next timeout period after a close. During this period, the hardware watchdog is kept alive.

A software watchdog client that wants to provide a grace period, before a hard reset, can set the timeout before properly closing.

Rather than changing the current gpio watchdog code, I would prefer to start with https://lkml.org/lkml/2016/1/26/517, which moves part of its code into the watchdog core - even more so here since I suspect that the suggested changes won't work anymore after the above patch is applied.

Also, isn't this quite identical to nowayout ?

Thanks,
Guenter

Quoted 53 lines
---
 Documentation/devicetree/bindings/watchdog/gpio-wdt.txt | 3 +++
 drivers/watchdog/gpio_wdt.c                             | 6 ++++++
 2 files changed, 9 insertions(+)
Documentation/devicetree/bindings/watchdog/gpio-wdt.txt +3 -0
index 83d2814..2669735 100644
@@ -17,6 +17,9 @@ Optional Properties:
 - always-running: If the watchdog timer cannot be disabled, add this flag to
   have the driver keep toggling the signal without a client. It will only cease
   to toggle the signal when the device is open and the timeout elapsed.
+- keep-armed-on-close: if the watchdog timer need to keep toggling the signal
+  when close, until the timeout elapsed, add this flag to have the driver
+  keep toggling the signal, until the timeout elapsed.
 - timeout-sec: Contains the watchdog timeout in seconds.
 - start-at-init: Start kicking watchdog as soon as driver is loaded.
 
drivers/watchdog/gpio_wdt.c +6 -0
index ef9ab91..ba9091a 100644
@@ -32,6 +32,7 @@ struct gpio_wdt_priv {
 	bool			active_low;
 	bool			state;
 	bool			always_running;
+	bool			keep_armed_on_close;
 	bool			armed;
 	unsigned int		hw_algo;
 	unsigned int		hw_margin;
@@ -72,6 +73,9 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 {
 	struct gpio_wdt_priv *priv = watchdog_get_drvdata(wdd);
 
+	if(priv->keep_armed_on_close)
+		return 0;
+
 	priv->armed = false;
 	if (!priv->always_running) {
 		mod_timer(&priv->timer, 0);
@@ -210,6 +214,8 @@ static int gpio_wdt_probe(struct platform_device *pdev)
 
 	priv->always_running = of_property_read_bool(pdev->dev.of_node,
 						     "always-running");
+	priv->keep_armed_on_close = of_property_read_bool(pdev->dev.of_node,
+						 "keep-armed-on-close");
 
 	watchdog_set_drvdata(&priv->wdd, priv);
 
Signature · 1 line
1.8.3.1
#2 lore raw

Hi Guenter,

Quoted 23 lines · Guenter Roeck
Quoted 22 lines · Sylvain Lemieux

From: Sylvain Lemieux <[email protected]>

There is a need to allow a grace period after the watchdog software client has closed. It could be used for syncing the filesystem or allow graceful termination while still providing a hardware reset in case the system has hung.

The "always-running" configuration from device-tree does not provide this since it will automatically keep the hardware watchdog alive as soon as the software client closes (i.e. keep toggling the GPIO line regardless of the state of the soft part of the watchdog).

The "keep-armed-on-close" member in the GPIO watchdog implementation indicates if an expired timeout should cause a reset.

This patch add a new "keep-armed-on-close" device-tree configuration that will keep the watchdog "armed" until the next timeout period after a close. During this period, the hardware watchdog is kept alive.

A software watchdog client that wants to provide a grace period, before a hard reset, can set the timeout before properly closing.

Quoted 7 lines
Quoted 1 line

Rather than changing the current gpio watchdog code, I would prefer to start with https://lkml.org/lkml/2016/1/26/517, which moves part of its code into the watchdog core - even more so here since I suspect that the suggested changes won't work anymore after the above patch is applied.

I agree that we should apply the "keepalive" change for this driver first; ref. https://lkml.org/lkml/2016/2/28/239 (version v8).

There is another change that was submitted to the mailing list, yesterday, that is implementing a similar functionality:
"watchdog: introduce CONFIG_WATCHDOG_OPEN_TIMEOUT"
http://www.spinics.net/lists/linux-watchdog/msg10500.html

I will try to look at both change, next week, to see if I can get the same functionality, with them, for the gpio watchdog.

Quoted 2 lines

Also, isn't this quite identical to nowayout ?

Nowayout block the watchdog driver from being stop.

In this case, we want to stop the watchdog (i.e. close) and be able to reopen it within the timeout period. If the driver is not reopen within the timeout period, the board should reset.

Regards,
Sylvain

Quoted 56 lines

Thanks,
Guenter

Quoted 53 lines
---
 Documentation/devicetree/bindings/watchdog/gpio-wdt.txt | 3 +++
 drivers/watchdog/gpio_wdt.c                             | 6 ++++++
 2 files changed, 9 insertions(+)
Documentation/devicetree/bindings/watchdog/gpio-wdt.txt +3 -0
index 83d2814..2669735 100644
@@ -17,6 +17,9 @@ Optional Properties:
 - always-running: If the watchdog timer cannot be disabled, add this flag to
   have the driver keep toggling the signal without a client. It will only cease
   to toggle the signal when the device is open and the timeout elapsed.
+- keep-armed-on-close: if the watchdog timer need to keep toggling the signal
+  when close, until the timeout elapsed, add this flag to have the driver
+  keep toggling the signal, until the timeout elapsed.
 - timeout-sec: Contains the watchdog timeout in seconds.
 - start-at-init: Start kicking watchdog as soon as driver is loaded.
 
drivers/watchdog/gpio_wdt.c +6 -0
index ef9ab91..ba9091a 100644
@@ -32,6 +32,7 @@ struct gpio_wdt_priv {
 	bool			active_low;
 	bool			state;
 	bool			always_running;
+	bool			keep_armed_on_close;
 	bool			armed;
 	unsigned int		hw_algo;
 	unsigned int		hw_margin;
@@ -72,6 +73,9 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 {
 	struct gpio_wdt_priv *priv = watchdog_get_drvdata(wdd);
 
+	if(priv->keep_armed_on_close)
+		return 0;
+
 	priv->armed = false;
 	if (!priv->always_running) {
 		mod_timer(&priv->timer, 0);
@@ -210,6 +214,8 @@ static int gpio_wdt_probe(struct platform_device *pdev)
 
 	priv->always_running = of_property_read_bool(pdev->dev.of_node,
 						     "always-running");
+	priv->keep_armed_on_close = of_property_read_bool(pdev->dev.of_node,
+						 "keep-armed-on-close");
 
 	watchdog_set_drvdata(&priv->wdd, priv);
 
Signature · 1 line
1.8.3.1
#3 lore raw

Hi Guenter,

Quoted 36 lines · Sylvain Lemieux

Hi Guenter,

Quoted 23 lines · Guenter Roeck
Quoted 22 lines · Sylvain Lemieux

From: Sylvain Lemieux <[email protected]>

There is a need to allow a grace period after the watchdog software client has closed. It could be used for syncing the filesystem or allow graceful termination while still providing a hardware reset in case the system has hung.

The "always-running" configuration from device-tree does not provide this since it will automatically keep the hardware watchdog alive as soon as the software client closes (i.e. keep toggling the GPIO line regardless of the state of the soft part of the watchdog).

The "keep-armed-on-close" member in the GPIO watchdog implementation indicates if an expired timeout should cause a reset.

This patch add a new "keep-armed-on-close" device-tree configuration that will keep the watchdog "armed" until the next timeout period after a close. During this period, the hardware watchdog is kept alive.

A software watchdog client that wants to provide a grace period, before a hard reset, can set the timeout before properly closing.

Quoted 7 lines
Quoted 1 line

Rather than changing the current gpio watchdog code, I would prefer to start with https://lkml.org/lkml/2016/1/26/517, which moves part of its code into the watchdog core - even more so here since I suspect that the suggested changes won't work anymore after the above patch is applied.

I agree that we should apply the "keepalive" change for this driver first; ref. https://lkml.org/lkml/2016/2/28/239 (version v8).

[...]

Quoted 1 line

I took the "watchdog: gpio: keepalives" patch and rebase this patch on top of it.

We have a scenario that will open and close the GPIO watchdog device a few times during the user-space initialization. This scenario is working properly with this patch on top of v4.9.

With the "watchdog: gpio: keepalives", I am getting an error when the GPIO watchdog driver is open for the 3rd time during this initialization process.

This is a short summary of stack trace:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 808 at lib/kobject.c:597 kobject_get+0x6c/0xac

kobject: '(null)' (c397f304): is not initialized, yet kobject_get() is being called.
...
[<c01a97f8>] (kobject_get) from [<c00c1278>] (chrdev_open+0x54/0x21c) [<c00c1224>] (chrdev_open) from [<c00b9df0>] (do_dentry_open +0x188/0x2d8)
...
[<c00cabcc>] (do_filp_open) from [<c00bb668>] (do_sys_open+0x114/0x1cc) [<c00bb554>] (do_sys_open) from [<c00bb74c>] (SyS_open+0x2c/0x30) [<c00bb720>] (SyS_open) from [<c000f840>] (ret_fast_syscall+0x0/0x38)

[   16.807133] ---[ end trace 7d55a0fe58991d4c ]---
[   16.807143] ------------[ cut here ]------------

Regards,
Sylvain

Quoted 70 lines
Quoted 3 lines

Also, isn't this quite identical to nowayout ?

Nowayout block the watchdog driver from being stop.

In this case, we want to stop the watchdog (i.e. close) and be able to reopen it within the timeout period. If the driver is not reopen within the timeout period, the board should reset.

Regards,
Sylvain

Quoted 56 lines

Thanks,
Guenter

Quoted 53 lines
---
 Documentation/devicetree/bindings/watchdog/gpio-wdt.txt | 3 +++
 drivers/watchdog/gpio_wdt.c                             | 6 ++++++
 2 files changed, 9 insertions(+)
Documentation/devicetree/bindings/watchdog/gpio-wdt.txt +3 -0
index 83d2814..2669735 100644
@@ -17,6 +17,9 @@ Optional Properties:
 - always-running: If the watchdog timer cannot be disabled, add this flag to
   have the driver keep toggling the signal without a client. It will only cease
   to toggle the signal when the device is open and the timeout elapsed.
+- keep-armed-on-close: if the watchdog timer need to keep toggling the signal
+  when close, until the timeout elapsed, add this flag to have the driver
+  keep toggling the signal, until the timeout elapsed.
 - timeout-sec: Contains the watchdog timeout in seconds.
 - start-at-init: Start kicking watchdog as soon as driver is loaded.
 
drivers/watchdog/gpio_wdt.c +6 -0
index ef9ab91..ba9091a 100644
@@ -32,6 +32,7 @@ struct gpio_wdt_priv {
 	bool			active_low;
 	bool			state;
 	bool			always_running;
+	bool			keep_armed_on_close;
 	bool			armed;
 	unsigned int		hw_algo;
 	unsigned int		hw_margin;
@@ -72,6 +73,9 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 {
 	struct gpio_wdt_priv *priv = watchdog_get_drvdata(wdd);
 
+	if(priv->keep_armed_on_close)
+		return 0;
+
 	priv->armed = false;
 	if (!priv->always_running) {
 		mod_timer(&priv->timer, 0);
@@ -210,6 +214,8 @@ static int gpio_wdt_probe(struct platform_device *pdev)
 
 	priv->always_running = of_property_read_bool(pdev->dev.of_node,
 						     "always-running");
+	priv->keep_armed_on_close = of_property_read_bool(pdev->dev.of_node,
+						 "keep-armed-on-close");
 
 	watchdog_set_drvdata(&priv->wdd, priv);
 
Signature · 1 line
1.8.3.1
#4 lore raw
Quoted 65 lines · Sylvain Lemieux

Hi Guenter,

Quoted 36 lines · Sylvain Lemieux

Hi Guenter,

Quoted 23 lines · Guenter Roeck
Quoted 22 lines · Sylvain Lemieux

From: Sylvain Lemieux <[email protected]>

There is a need to allow a grace period after the watchdog software client has closed. It could be used for syncing the filesystem or allow graceful termination while still providing a hardware reset in case the system has hung.

The "always-running" configuration from device-tree does not provide this since it will automatically keep the hardware watchdog alive as soon as the software client closes (i.e. keep toggling the GPIO line regardless of the state of the soft part of the watchdog).

The "keep-armed-on-close" member in the GPIO watchdog implementation indicates if an expired timeout should cause a reset.

This patch add a new "keep-armed-on-close" device-tree configuration that will keep the watchdog "armed" until the next timeout period after a close. During this period, the hardware watchdog is kept alive.

A software watchdog client that wants to provide a grace period, before a hard reset, can set the timeout before properly closing.

Quoted 7 lines
Quoted 1 line

Rather than changing the current gpio watchdog code, I would prefer to start with https://lkml.org/lkml/2016/1/26/517, which moves part of its code into the watchdog core - even more so here since I suspect that the suggested changes won't work anymore after the above patch is applied.

I agree that we should apply the "keepalive" change for this driver first; ref. https://lkml.org/lkml/2016/2/28/239 (version v8).

[...]

Quoted 1 line

I took the "watchdog: gpio: keepalives" patch and rebase this patch on top of it.

We have a scenario that will open and close the GPIO watchdog device a few times during the user-space initialization. This scenario is working properly with this patch on top of v4.9.

With the "watchdog: gpio: keepalives", I am getting an error when the GPIO watchdog driver is open for the 3rd time during this initialization process.

This is a short summary of stack trace:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 808 at lib/kobject.c:597 kobject_get+0x6c/0xac

kobject: '(null)' (c397f304): is not initialized, yet kobject_get() is being called.
...
[<c01a97f8>] (kobject_get) from [<c00c1278>] (chrdev_open+0x54/0x21c) [<c00c1224>] (chrdev_open) from [<c00b9df0>] (do_dentry_open +0x188/0x2d8)
...
[<c00cabcc>] (do_filp_open) from [<c00bb668>] (do_sys_open+0x114/0x1cc) [<c00bb554>] (do_sys_open) from [<c00bb74c>] (SyS_open+0x2c/0x30) [<c00bb720>] (SyS_open) from [<c000f840>] (ret_fast_syscall+0x0/0x38)

Quoted 6 lines
[   16.807133] ---[ end trace 7d55a0fe58991d4c ]---
[   16.807143] ------------[ cut here ]------------

Regards,
Sylvain

ping

Sylvain

Quoted 72 lines
Quoted 70 lines
Quoted 3 lines

Also, isn't this quite identical to nowayout ?

Nowayout block the watchdog driver from being stop.

In this case, we want to stop the watchdog (i.e. close) and be able to reopen it within the timeout period. If the driver is not reopen within the timeout period, the board should reset.

Regards,
Sylvain

Quoted 56 lines

Thanks,
Guenter

Quoted 53 lines
---
 Documentation/devicetree/bindings/watchdog/gpio-wdt.txt | 3 +++
 drivers/watchdog/gpio_wdt.c                             | 6 ++++++
 2 files changed, 9 insertions(+)
Documentation/devicetree/bindings/watchdog/gpio-wdt.txt +3 -0
index 83d2814..2669735 100644
@@ -17,6 +17,9 @@ Optional Properties:
 - always-running: If the watchdog timer cannot be disabled, add this flag to
   have the driver keep toggling the signal without a client. It will only cease
   to toggle the signal when the device is open and the timeout elapsed.
+- keep-armed-on-close: if the watchdog timer need to keep toggling the signal
+  when close, until the timeout elapsed, add this flag to have the driver
+  keep toggling the signal, until the timeout elapsed.
 - timeout-sec: Contains the watchdog timeout in seconds.
 - start-at-init: Start kicking watchdog as soon as driver is loaded.
 
drivers/watchdog/gpio_wdt.c +6 -0
index ef9ab91..ba9091a 100644
@@ -32,6 +32,7 @@ struct gpio_wdt_priv {
 	bool			active_low;
 	bool			state;
 	bool			always_running;
+	bool			keep_armed_on_close;
 	bool			armed;
 	unsigned int		hw_algo;
 	unsigned int		hw_margin;
@@ -72,6 +73,9 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 {
 	struct gpio_wdt_priv *priv = watchdog_get_drvdata(wdd);
 
+	if(priv->keep_armed_on_close)
+		return 0;
+
 	priv->armed = false;
 	if (!priv->always_running) {
 		mod_timer(&priv->timer, 0);
@@ -210,6 +214,8 @@ static int gpio_wdt_probe(struct platform_device *pdev)
 
 	priv->always_running = of_property_read_bool(pdev->dev.of_node,
 						     "always-running");
+	priv->keep_armed_on_close = of_property_read_bool(pdev->dev.of_node,
+						 "keep-armed-on-close");
 
 	watchdog_set_drvdata(&priv->wdd, priv);
 
Signature · 1 line
1.8.3.1
maintainer
#5 lore raw
Quoted 69 lines · Sylvain Lemieux
Quoted 68 lines · Sylvain Lemieux

Hi Guenter,

Quoted 36 lines · Sylvain Lemieux

Hi Guenter,

Quoted 23 lines · Guenter Roeck
Quoted 22 lines · Sylvain Lemieux

From: Sylvain Lemieux <[email protected]>

There is a need to allow a grace period after the watchdog software client has closed. It could be used for syncing the filesystem or allow graceful termination while still providing a hardware reset in case the system has hung.

The "always-running" configuration from device-tree does not provide this since it will automatically keep the hardware watchdog alive as soon as the software client closes (i.e. keep toggling the GPIO line regardless of the state of the soft part of the watchdog).

The "keep-armed-on-close" member in the GPIO watchdog implementation indicates if an expired timeout should cause a reset.

This patch add a new "keep-armed-on-close" device-tree configuration that will keep the watchdog "armed" until the next timeout period after a close. During this period, the hardware watchdog is kept alive.

A software watchdog client that wants to provide a grace period, before a hard reset, can set the timeout before properly closing.

Quoted 7 lines
Quoted 1 line

Rather than changing the current gpio watchdog code, I would prefer to start with https://lkml.org/lkml/2016/1/26/517, which moves part of its code into the watchdog core - even more so here since I suspect that the suggested changes won't work anymore after the above patch is applied.

I agree that we should apply the "keepalive" change for this driver first; ref. https://lkml.org/lkml/2016/2/28/239 (version v8).

[...]

Quoted 1 line

I took the "watchdog: gpio: keepalives" patch and rebase this patch on top of it.

We have a scenario that will open and close the GPIO watchdog device a few times during the user-space initialization. This scenario is working properly with this patch on top of v4.9.

With the "watchdog: gpio: keepalives", I am getting an error when the GPIO watchdog driver is open for the 3rd time during this initialization process.

This is a short summary of stack trace:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 808 at lib/kobject.c:597 kobject_get+0x6c/0xac

kobject: '(null)' (c397f304): is not initialized, yet kobject_get() is being called.
...
[<c01a97f8>] (kobject_get) from [<c00c1278>] (chrdev_open+0x54/0x21c) [<c00c1224>] (chrdev_open) from [<c00b9df0>] (do_dentry_open +0x188/0x2d8)
...
[<c00cabcc>] (do_filp_open) from [<c00bb668>] (do_sys_open+0x114/0x1cc) [<c00bb554>] (do_sys_open) from [<c00bb74c>] (SyS_open+0x2c/0x30) [<c00bb720>] (SyS_open) from [<c000f840>] (ret_fast_syscall+0x0/0x38)

[   16.807133] ---[ end trace 7d55a0fe58991d4c ]---
[   16.807143] ------------[ cut here ]------------
Quoted 5 lines
Quoted 3 lines

Regards,
Sylvain

ping

Busy, sorry.

Guenter

Quoted 76 lines

Sylvain

Quoted 72 lines
Quoted 70 lines
Quoted 3 lines

Also, isn't this quite identical to nowayout ?

Nowayout block the watchdog driver from being stop.

In this case, we want to stop the watchdog (i.e. close) and be able to reopen it within the timeout period. If the driver is not reopen within the timeout period, the board should reset.

Regards,
Sylvain

Quoted 56 lines

Thanks,
Guenter

Quoted 53 lines
---
 Documentation/devicetree/bindings/watchdog/gpio-wdt.txt | 3 +++
 drivers/watchdog/gpio_wdt.c                             | 6 ++++++
 2 files changed, 9 insertions(+)
Documentation/devicetree/bindings/watchdog/gpio-wdt.txt +3 -0
index 83d2814..2669735 100644
@@ -17,6 +17,9 @@ Optional Properties:
 - always-running: If the watchdog timer cannot be disabled, add this flag to
   have the driver keep toggling the signal without a client. It will only cease
   to toggle the signal when the device is open and the timeout elapsed.
+- keep-armed-on-close: if the watchdog timer need to keep toggling the signal
+  when close, until the timeout elapsed, add this flag to have the driver
+  keep toggling the signal, until the timeout elapsed.
 - timeout-sec: Contains the watchdog timeout in seconds.
 - start-at-init: Start kicking watchdog as soon as driver is loaded.
 
drivers/watchdog/gpio_wdt.c +6 -0
index ef9ab91..ba9091a 100644
@@ -32,6 +32,7 @@ struct gpio_wdt_priv {
 	bool			active_low;
 	bool			state;
 	bool			always_running;
+	bool			keep_armed_on_close;
 	bool			armed;
 	unsigned int		hw_algo;
 	unsigned int		hw_margin;
@@ -72,6 +73,9 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 {
 	struct gpio_wdt_priv *priv = watchdog_get_drvdata(wdd);
 
+	if(priv->keep_armed_on_close)
+		return 0;
+
 	priv->armed = false;
 	if (!priv->always_running) {
 		mod_timer(&priv->timer, 0);
@@ -210,6 +214,8 @@ static int gpio_wdt_probe(struct platform_device *pdev)
 
 	priv->always_running = of_property_read_bool(pdev->dev.of_node,
 						     "always-running");
+	priv->keep_armed_on_close = of_property_read_bool(pdev->dev.of_node,
+						 "keep-armed-on-close");
 
 	watchdog_set_drvdata(&priv->wdd, priv);
 
Signature · 1 line
1.8.3.1
#6 lore raw
Quoted 70 lines · Guenter Roeck
Quoted 69 lines · Sylvain Lemieux
Quoted 68 lines · Sylvain Lemieux

Hi Guenter,

Quoted 36 lines · Sylvain Lemieux

Hi Guenter,

Quoted 23 lines · Guenter Roeck
Quoted 22 lines · Sylvain Lemieux

From: Sylvain Lemieux <[email protected]>

There is a need to allow a grace period after the watchdog software client has closed. It could be used for syncing the filesystem or allow graceful termination while still providing a hardware reset in case the system has hung.

The "always-running" configuration from device-tree does not provide this since it will automatically keep the hardware watchdog alive as soon as the software client closes (i.e. keep toggling the GPIO line regardless of the state of the soft part of the watchdog).

The "keep-armed-on-close" member in the GPIO watchdog implementation indicates if an expired timeout should cause a reset.

This patch add a new "keep-armed-on-close" device-tree configuration that will keep the watchdog "armed" until the next timeout period after a close. During this period, the hardware watchdog is kept alive.

A software watchdog client that wants to provide a grace period, before a hard reset, can set the timeout before properly closing.

Quoted 7 lines
Quoted 1 line

Rather than changing the current gpio watchdog code, I would prefer to start with https://lkml.org/lkml/2016/1/26/517, which moves part of its code into the watchdog core - even more so here since I suspect that the suggested changes won't work anymore after the above patch is applied.

I agree that we should apply the "keepalive" change for this driver first; ref. https://lkml.org/lkml/2016/2/28/239 (version v8).

[...]

Quoted 1 line

I took the "watchdog: gpio: keepalives" patch and rebase this patch on top of it.

We have a scenario that will open and close the GPIO watchdog device a few times during the user-space initialization. This scenario is working properly with this patch on top of v4.9.

With the "watchdog: gpio: keepalives", I am getting an error when the GPIO watchdog driver is open for the 3rd time during this initialization process.

This is a short summary of stack trace:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 808 at lib/kobject.c:597 kobject_get+0x6c/0xac

kobject: '(null)' (c397f304): is not initialized, yet kobject_get() is being called.
...
[<c01a97f8>] (kobject_get) from [<c00c1278>] (chrdev_open+0x54/0x21c) [<c00c1224>] (chrdev_open) from [<c00b9df0>] (do_dentry_open +0x188/0x2d8)
...
[<c00cabcc>] (do_filp_open) from [<c00bb668>] (do_sys_open+0x114/0x1cc) [<c00bb554>] (do_sys_open) from [<c00bb74c>] (SyS_open+0x2c/0x30) [<c00bb720>] (SyS_open) from [<c000f840>] (ret_fast_syscall+0x0/0x38)

[   16.807133] ---[ end trace 7d55a0fe58991d4c ]---
[   16.807143] ------------[ cut here ]------------
Quoted 9 lines
Quoted 5 lines
Quoted 3 lines

Regards,
Sylvain

ping

Busy, sorry.

Guenter

Just doing a follow-up.

Quoted 76 lines
Quoted 76 lines

Sylvain

Quoted 72 lines
Quoted 70 lines
Quoted 3 lines

Also, isn't this quite identical to nowayout ?

Nowayout block the watchdog driver from being stop.

In this case, we want to stop the watchdog (i.e. close) and be able to reopen it within the timeout period. If the driver is not reopen within the timeout period, the board should reset.

Regards,
Sylvain

Quoted 56 lines

Thanks,
Guenter

Quoted 53 lines
---
 Documentation/devicetree/bindings/watchdog/gpio-wdt.txt | 3 +++
 drivers/watchdog/gpio_wdt.c                             | 6 ++++++
 2 files changed, 9 insertions(+)
Documentation/devicetree/bindings/watchdog/gpio-wdt.txt +3 -0
index 83d2814..2669735 100644
@@ -17,6 +17,9 @@ Optional Properties:
 - always-running: If the watchdog timer cannot be disabled, add this flag to
   have the driver keep toggling the signal without a client. It will only cease
   to toggle the signal when the device is open and the timeout elapsed.
+- keep-armed-on-close: if the watchdog timer need to keep toggling the signal
+  when close, until the timeout elapsed, add this flag to have the driver
+  keep toggling the signal, until the timeout elapsed.
 - timeout-sec: Contains the watchdog timeout in seconds.
 - start-at-init: Start kicking watchdog as soon as driver is loaded.
 
drivers/watchdog/gpio_wdt.c +6 -0
index ef9ab91..ba9091a 100644
@@ -32,6 +32,7 @@ struct gpio_wdt_priv {
 	bool			active_low;
 	bool			state;
 	bool			always_running;
+	bool			keep_armed_on_close;
 	bool			armed;
 	unsigned int		hw_algo;
 	unsigned int		hw_margin;
@@ -72,6 +73,9 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 {
 	struct gpio_wdt_priv *priv = watchdog_get_drvdata(wdd);
 
+	if(priv->keep_armed_on_close)
+		return 0;
+
 	priv->armed = false;
 	if (!priv->always_running) {
 		mod_timer(&priv->timer, 0);
@@ -210,6 +214,8 @@ static int gpio_wdt_probe(struct platform_device *pdev)
 
 	priv->always_running = of_property_read_bool(pdev->dev.of_node,
 						     "always-running");
+	priv->keep_armed_on_close = of_property_read_bool(pdev->dev.of_node,
+						 "keep-armed-on-close");
 
 	watchdog_set_drvdata(&priv->wdd, priv);
 
Signature · 1 line
1.8.3.1
#7 lore raw
In reply to Guenter Roeck

Hi Guenter,

Quoted 46 lines · Guenter Roeck
Quoted 45 lines · Sylvain Lemieux
Quoted 44 lines · Sylvain Lemieux

Hi Guenter,

Quoted 36 lines · Sylvain Lemieux

Hi Guenter,

Quoted 23 lines · Guenter Roeck
Quoted 22 lines · Sylvain Lemieux

From: Sylvain Lemieux <[email protected]>

There is a need to allow a grace period after the watchdog software client has closed. It could be used for syncing the filesystem or allow graceful termination while still providing a hardware reset in case the system has hung.

The "always-running" configuration from device-tree does not provide this since it will automatically keep the hardware watchdog alive as soon as the software client closes (i.e. keep toggling the GPIO line regardless of the state of the soft part of the watchdog).

The "keep-armed-on-close" member in the GPIO watchdog implementation indicates if an expired timeout should cause a reset.

This patch add a new "keep-armed-on-close" device-tree configuration that will keep the watchdog "armed" until the next timeout period after a close. During this period, the hardware watchdog is kept alive.

A software watchdog client that wants to provide a grace period, before a hard reset, can set the timeout before properly closing.

Quoted 7 lines
Quoted 1 line

Rather than changing the current gpio watchdog code, I would prefer to start with https://lkml.org/lkml/2016/1/26/517, which moves part of its code into the watchdog core - even more so here since I suspect that the suggested changes won't work anymore after the above patch is applied.

I agree that we should apply the "keepalive" change for this driver first; ref. https://lkml.org/lkml/2016/2/28/239 (version v8).

[...]

Quoted 1 line

I took the "watchdog: gpio: keepalives" patch and rebase this patch on top of it.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

We have a scenario that will open and close the GPIO watchdog device a few times during the user-space initialization. This scenario is working properly with this patch on top of v4.9.

With the "watchdog: gpio: keepalives", I am getting an error when the GPIO watchdog driver is open for the 3rd time during this initialization process.

[...]

I apply the "keepalive" change, for this driver, on top of 4.11-rc1.

There is an issue with this patch, and the handling of the WDOG_HW_RUNNING flag. I will submit a version 2, of this patch, rebase on top of the "keepalive" v8 patch
(https://lkml.org/lkml/2016/2/28/239).

Regards,
Sylvain

Quoted 74 lines
Quoted 74 lines
Quoted 72 lines
Quoted 70 lines
Quoted 3 lines

Also, isn't this quite identical to nowayout ?

Nowayout block the watchdog driver from being stop.

In this case, we want to stop the watchdog (i.e. close) and be able to reopen it within the timeout period. If the driver is not reopen within the timeout period, the board should reset.

Regards,
Sylvain

Quoted 56 lines

Thanks,
Guenter

Quoted 53 lines
---
 Documentation/devicetree/bindings/watchdog/gpio-wdt.txt | 3 +++
 drivers/watchdog/gpio_wdt.c                             | 6 ++++++
 2 files changed, 9 insertions(+)
Documentation/devicetree/bindings/watchdog/gpio-wdt.txt +3 -0
index 83d2814..2669735 100644
@@ -17,6 +17,9 @@ Optional Properties:
 - always-running: If the watchdog timer cannot be disabled, add this flag to
   have the driver keep toggling the signal without a client. It will only cease
   to toggle the signal when the device is open and the timeout elapsed.
+- keep-armed-on-close: if the watchdog timer need to keep toggling the signal
+  when close, until the timeout elapsed, add this flag to have the driver
+  keep toggling the signal, until the timeout elapsed.
 - timeout-sec: Contains the watchdog timeout in seconds.
 - start-at-init: Start kicking watchdog as soon as driver is loaded.
 
drivers/watchdog/gpio_wdt.c +6 -0
index ef9ab91..ba9091a 100644
@@ -32,6 +32,7 @@ struct gpio_wdt_priv {
 	bool			active_low;
 	bool			state;
 	bool			always_running;
+	bool			keep_armed_on_close;
 	bool			armed;
 	unsigned int		hw_algo;
 	unsigned int		hw_margin;
@@ -72,6 +73,9 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 {
 	struct gpio_wdt_priv *priv = watchdog_get_drvdata(wdd);
 
+	if(priv->keep_armed_on_close)
+		return 0;
+
 	priv->armed = false;
 	if (!priv->always_running) {
 		mod_timer(&priv->timer, 0);
@@ -210,6 +214,8 @@ static int gpio_wdt_probe(struct platform_device *pdev)
 
 	priv->always_running = of_property_read_bool(pdev->dev.of_node,
 						     "always-running");
+	priv->keep_armed_on_close = of_property_read_bool(pdev->dev.of_node,
+						 "keep-armed-on-close");
 
 	watchdog_set_drvdata(&priv->wdd, priv);
 
Signature · 1 line
1.8.3.1