watchdog: Introduce WDOG_HW_RUNNING flag

The WDOG_HW_RUNNING flag is expected to be set by watchdog drivers if
the hardware watchdog is running. If the flag is set, the watchdog
subsystem will ping the watchdog even if the watchdog device is closed.

The watchdog driver stop function is now optional and may be omitted
if the watchdog can not be stopped. If stopping the watchdog is not
possible but the driver implements a stop function, it is responsible
to set the WDOG_HW_RUNNING flag in its stop function.

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>

diff --git a/Documentation/watchdog/watchdog-kernel-api.txt b/Documentation/watchdog/watchdog-kernel-api.txt
index 15a02595ade1f49dc5baa0e74db98ccdb1301e75..954134a5c4a417b27d8f3ed4f1b72a061d479400 100644 (file)
--- a/Documentation/watchdog/watchdog-kernel-api.txt
+++ b/Documentation/watchdog/watchdog-kernel-api.txt
@@ -137,10 +137,10 @@ are:
 * stop: with this routine the watchdog timer device is being stopped.
   The routine needs a pointer to the watchdog timer device structure as a
   parameter. It returns zero on success or a negative errno code for failure.
-  Some watchdog timer hardware can only be started and not be stopped. The
-  driver supporting this hardware needs to make sure that a start and stop
-  routine is being provided. This can be done by using a timer in the driver
-  that regularly sends a keepalive ping to the watchdog timer hardware.
+  Some watchdog timer hardware can only be started and not be stopped.
+  If a watchdog can not be stopped, the watchdog driver must set the
+  WDOG_HW_RUNNING flag in its stop function to inform the watchdog core that
+  the watchdog is still running.
 
 Not all watchdog timer hardware supports the same functionality. That's why
 all other routines/operations are optional. They only need to be provided if
@@ -189,11 +189,19 @@ The 'ref' and 'unref' operations are no longer used and deprecated.
 The status bits should (preferably) be set with the set_bit and clear_bit alike
 bit-operations. The status bits that are defined are:
 * WDOG_ACTIVE: this status bit indicates whether or not a watchdog timer device
-  is active or not. When the watchdog is active after booting, then you should
-  set this status bit (Note: when you register the watchdog timer device with
-  this bit set, then opening /dev/watchdog will skip the start operation)
+  is active or not from user perspective. User space is expected to send
+  heartbeat requests to the driver while this flag is set.
 * WDOG_NO_WAY_OUT: this bit stores the nowayout setting for the watchdog.
   If this bit is set then the watchdog timer will not be able to stop.
+* WDOG_HW_RUNNING: Set by the watchdog driver if the hardware watchdog is
+  running. The bit must be set if the watchdog timer hardware can not be
+  stopped. The bit may also be set if the watchdog timer is running after
+  booting, before the watchdog device is opened. If set, the watchdog
+  infrastructure will send keepalives to the watchdog hardware while
+  WDOG_ACTIVE is not set.
+  Note: when you register the watchdog timer device with this bit set,
+  then opening /dev/watchdog will skip the start operation but send a keepalive
+  request instead.
 
   To set the WDOG_NO_WAY_OUT status bit (before registering your watchdog
   timer device) you can either:

diff --git a/drivers/watchdog/watchdog_dev.c b/drivers/watchdog/watchdog_dev.c
index e668a9e8b64891dd9bac383f0369e6e8818cefe8..5d3a9fa4856e0dee62e1c5db64e93cb9633642bb 100644 (file)
--- a/drivers/watchdog/watchdog_dev.c
+++ b/drivers/watchdog/watchdog_dev.c
@@ -92,7 +92,8 @@ static inline bool watchdog_need_worker(struct watchdog_device *wdd)
         *   requests.
         * - Userspace requests a longer timeout than the hardware can handle.
         */
-       return watchdog_active(wdd) && hm && t > hm;
+       return hm && ((watchdog_active(wdd) && t > hm) ||
+                     (t && !watchdog_active(wdd) && watchdog_hw_running(wdd)));
 }
 
 static long watchdog_next_keepalive(struct watchdog_device *wdd)
@@ -108,6 +109,9 @@ static long watchdog_next_keepalive(struct watchdog_device *wdd)
        hw_heartbeat_ms = min(timeout_ms, wdd->max_hw_heartbeat_ms);
        keepalive_interval = msecs_to_jiffies(hw_heartbeat_ms / 2);
 
+       if (!watchdog_active(wdd))
+               return keepalive_interval;
+
        /*
         * To ensure that the watchdog times out wdd->timeout seconds
         * after the most recent ping from userspace, the last
@@ -161,7 +165,7 @@ static int watchdog_ping(struct watchdog_device *wdd)
 {
        struct watchdog_core_data *wd_data = wdd->wd_data;
 
-       if (!watchdog_active(wdd))
+       if (!watchdog_active(wdd) && !watchdog_hw_running(wdd))
                return 0;
 
        wd_data->last_keepalive = jiffies;
@@ -178,7 +182,7 @@ static void watchdog_ping_work(struct work_struct *work)
 
        mutex_lock(&wd_data->lock);
        wdd = wd_data->wdd;
-       if (wdd && watchdog_active(wdd))
+       if (wdd && (watchdog_active(wdd) || watchdog_hw_running(wdd)))
                __watchdog_ping(wdd);
        mutex_unlock(&wd_data->lock);
 }
@@ -204,7 +208,10 @@ static int watchdog_start(struct watchdog_device *wdd)
                return 0;
 
        started_at = jiffies;
-       err = wdd->ops->start(wdd);
+       if (watchdog_hw_running(wdd) && wdd->ops->ping)
+               err = wdd->ops->ping(wdd);
+       else
+               err = wdd->ops->start(wdd);
        if (err == 0) {
                set_bit(WDOG_ACTIVE, &wdd->status);
                wd_data->last_keepalive = started_at;
@@ -228,8 +235,7 @@ static int watchdog_start(struct watchdog_device *wdd)
 
 static int watchdog_stop(struct watchdog_device *wdd)
 {
-       struct watchdog_core_data *wd_data = wdd->wd_data;
-       int err;
+       int err = 0;
 
        if (!watchdog_active(wdd))
                return 0;
@@ -243,7 +249,7 @@ static int watchdog_stop(struct watchdog_device *wdd)
        err = wdd->ops->stop(wdd);
        if (err == 0) {
                clear_bit(WDOG_ACTIVE, &wdd->status);
-               cancel_delayed_work(&wd_data->work);
+               watchdog_update_worker(wdd);
        }
 
        return err;
@@ -641,7 +647,7 @@ static int watchdog_open(struct inode *inode, struct file *file)
         * If the /dev/watchdog device is open, we don't want the module
         * to be unloaded.
         */
-       if (!try_module_get(wdd->ops->owner)) {
+       if (!watchdog_hw_running(wdd) && !try_module_get(wdd->ops->owner)) {
                err = -EBUSY;
                goto out_clear;
        }
@@ -652,7 +658,8 @@ static int watchdog_open(struct inode *inode, struct file *file)
 
        file->private_data = wd_data;
 
-       kref_get(&wd_data->kref);
+       if (!watchdog_hw_running(wdd))
+               kref_get(&wd_data->kref);
 
        /* dev/watchdog is a virtual (and thus non-seekable) filesystem */
        return nonseekable_open(inode, file);
@@ -713,15 +720,22 @@ static int watchdog_release(struct inode *inode, struct file *file)
        }
 
        cancel_delayed_work_sync(&wd_data->work);
+       watchdog_update_worker(wdd);
 
        /* make sure that /dev/watchdog can be re-opened */
        clear_bit(_WDOG_DEV_OPEN, &wd_data->status);
 
 done:
        mutex_unlock(&wd_data->lock);
-       /* Allow the owner module to be unloaded again */
-       module_put(wd_data->cdev.owner);
-       kref_put(&wd_data->kref, watchdog_core_data_release);
+       /*
+        * Allow the owner module to be unloaded again unless the watchdog
+        * is still running. If the watchdog is still running, it can not
+        * be stopped, and its driver must not be unloaded.
+        */
+       if (!watchdog_hw_running(wdd)) {
+               module_put(wdd->ops->owner);
+               kref_put(&wd_data->kref, watchdog_core_data_release);
+       }
        return 0;
 }
 
@@ -798,8 +812,20 @@ static int watchdog_cdev_register(struct watchdog_device *wdd, dev_t devno)
                        old_wd_data = NULL;
                        kref_put(&wd_data->kref, watchdog_core_data_release);
                }
+               return err;
        }
-       return err;
+
+       /*
+        * If the watchdog is running, prevent its driver from being unloaded,
+        * and schedule an immediate ping.
+        */
+       if (watchdog_hw_running(wdd)) {
+               __module_get(wdd->ops->owner);
+               kref_get(&wd_data->kref);
+               queue_delayed_work(watchdog_wq, &wd_data->work, 0);
+       }
+
+       return 0;
 }
 
 /*

diff --git a/include/linux/watchdog.h b/include/linux/watchdog.h
index 8e82daecb7d3ee60cf8e085fd72658128ae811f4..e2f45549b24304f78c9e5ea9c4fbcd8254a2603e 100644 (file)
--- a/include/linux/watchdog.h
+++ b/include/linux/watchdog.h
@@ -105,6 +105,7 @@ struct watchdog_device {
 #define WDOG_ACTIVE            0       /* Is the watchdog running/active */
 #define WDOG_NO_WAY_OUT                1       /* Is 'nowayout' feature set ? */
 #define WDOG_STOP_ON_REBOOT    2       /* Should be stopped on reboot */
+#define WDOG_HW_RUNNING                3       /* True if HW watchdog running */
        struct list_head deferred;
 };
 
@@ -117,6 +118,15 @@ static inline bool watchdog_active(struct watchdog_device *wdd)
        return test_bit(WDOG_ACTIVE, &wdd->status);
 }
 
+/*
+ * Use the following function to check whether or not the hardware watchdog
+ * is running
+ */
+static inline bool watchdog_hw_running(struct watchdog_device *wdd)
+{
+       return test_bit(WDOG_HW_RUNNING, &wdd->status);
+}
+
 /* Use the following function to set the nowayout feature */
 static inline void watchdog_set_nowayout(struct watchdog_device *wdd, bool nowayout)
 {