1The Linux WatchDog Timer Driver Core kernel API. 2=============================================== 3Last reviewed: 12-Feb-2013 4 5Wim Van Sebroeck <wim@iguana.be> 6 7Introduction 8------------ 9This document does not describe what a WatchDog Timer (WDT) Driver or Device is. 10It also does not describe the API which can be used by user space to communicate 11with a WatchDog Timer. If you want to know this then please read the following 12file: Documentation/watchdog/watchdog-api.txt . 13 14So what does this document describe? It describes the API that can be used by 15WatchDog Timer Drivers that want to use the WatchDog Timer Driver Core 16Framework. This framework provides all interfacing towards user space so that 17the same code does not have to be reproduced each time. This also means that 18a watchdog timer driver then only needs to provide the different routines 19(operations) that control the watchdog timer (WDT). 20 21The API 22------- 23Each watchdog timer driver that wants to use the WatchDog Timer Driver Core 24must #include <linux/watchdog.h> (you would have to do this anyway when 25writing a watchdog device driver). This include file contains following 26register/unregister routines: 27 28extern int watchdog_register_device(struct watchdog_device *); 29extern void watchdog_unregister_device(struct watchdog_device *); 30 31The watchdog_register_device routine registers a watchdog timer device. 32The parameter of this routine is a pointer to a watchdog_device structure. 33This routine returns zero on success and a negative errno code for failure. 34 35The watchdog_unregister_device routine deregisters a registered watchdog timer 36device. The parameter of this routine is the pointer to the registered 37watchdog_device structure. 38 39The watchdog subsystem includes an registration deferral mechanism, 40which allows you to register an watchdog as early as you wish during 41the boot process. 42 43The watchdog device structure looks like this: 44 45struct watchdog_device { 46 int id; 47 struct cdev cdev; 48 struct device *dev; 49 struct device *parent; 50 const struct watchdog_info *info; 51 const struct watchdog_ops *ops; 52 unsigned int bootstatus; 53 unsigned int timeout; 54 unsigned int min_timeout; 55 unsigned int max_timeout; 56 void *driver_data; 57 struct mutex lock; 58 unsigned long status; 59 struct list_head deferred; 60}; 61 62It contains following fields: 63* id: set by watchdog_register_device, id 0 is special. It has both a 64 /dev/watchdog0 cdev (dynamic major, minor 0) as well as the old 65 /dev/watchdog miscdev. The id is set automatically when calling 66 watchdog_register_device. 67* cdev: cdev for the dynamic /dev/watchdog<id> device nodes. This 68 field is also populated by watchdog_register_device. 69* dev: device under the watchdog class (created by watchdog_register_device). 70* parent: set this to the parent device (or NULL) before calling 71 watchdog_register_device. 72* info: a pointer to a watchdog_info structure. This structure gives some 73 additional information about the watchdog timer itself. (Like it's unique name) 74* ops: a pointer to the list of watchdog operations that the watchdog supports. 75* timeout: the watchdog timer's timeout value (in seconds). 76* min_timeout: the watchdog timer's minimum timeout value (in seconds). 77* max_timeout: the watchdog timer's maximum timeout value (in seconds). 78* bootstatus: status of the device after booting (reported with watchdog 79 WDIOF_* status bits). 80* driver_data: a pointer to the drivers private data of a watchdog device. 81 This data should only be accessed via the watchdog_set_drvdata and 82 watchdog_get_drvdata routines. 83* lock: Mutex for WatchDog Timer Driver Core internal use only. 84* status: this field contains a number of status bits that give extra 85 information about the status of the device (Like: is the watchdog timer 86 running/active, is the nowayout bit set, is the device opened via 87 the /dev/watchdog interface or not, ...). 88* deferred: entry in wtd_deferred_reg_list which is used to 89 register early initialized watchdogs. 90 91The list of watchdog operations is defined as: 92 93struct watchdog_ops { 94 struct module *owner; 95 /* mandatory operations */ 96 int (*start)(struct watchdog_device *); 97 int (*stop)(struct watchdog_device *); 98 /* optional operations */ 99 int (*ping)(struct watchdog_device *); 100 unsigned int (*status)(struct watchdog_device *); 101 int (*set_timeout)(struct watchdog_device *, unsigned int); 102 unsigned int (*get_timeleft)(struct watchdog_device *); 103 void (*ref)(struct watchdog_device *); 104 void (*unref)(struct watchdog_device *); 105 long (*ioctl)(struct watchdog_device *, unsigned int, unsigned long); 106}; 107 108It is important that you first define the module owner of the watchdog timer 109driver's operations. This module owner will be used to lock the module when 110the watchdog is active. (This to avoid a system crash when you unload the 111module and /dev/watchdog is still open). 112 113If the watchdog_device struct is dynamically allocated, just locking the module 114is not enough and a driver also needs to define the ref and unref operations to 115ensure the structure holding the watchdog_device does not go away. 116 117The simplest (and usually sufficient) implementation of this is to: 1181) Add a kref struct to the same structure which is holding the watchdog_device 1192) Define a release callback for the kref which frees the struct holding both 1203) Call kref_init on this kref *before* calling watchdog_register_device() 1214) Define a ref operation calling kref_get on this kref 1225) Define a unref operation calling kref_put on this kref 1236) When it is time to cleanup: 124 * Do not kfree() the struct holding both, the last kref_put will do this! 125 * *After* calling watchdog_unregister_device() call kref_put on the kref 126 127Some operations are mandatory and some are optional. The mandatory operations 128are: 129* start: this is a pointer to the routine that starts the watchdog timer 130 device. 131 The routine needs a pointer to the watchdog timer device structure as a 132 parameter. It returns zero on success or a negative errno code for failure. 133* stop: with this routine the watchdog timer device is being stopped. 134 The routine needs a pointer to the watchdog timer device structure as a 135 parameter. It returns zero on success or a negative errno code for failure. 136 Some watchdog timer hardware can only be started and not be stopped. The 137 driver supporting this hardware needs to make sure that a start and stop 138 routine is being provided. This can be done by using a timer in the driver 139 that regularly sends a keepalive ping to the watchdog timer hardware. 140 141Not all watchdog timer hardware supports the same functionality. That's why 142all other routines/operations are optional. They only need to be provided if 143they are supported. These optional routines/operations are: 144* ping: this is the routine that sends a keepalive ping to the watchdog timer 145 hardware. 146 The routine needs a pointer to the watchdog timer device structure as a 147 parameter. It returns zero on success or a negative errno code for failure. 148 Most hardware that does not support this as a separate function uses the 149 start function to restart the watchdog timer hardware. And that's also what 150 the watchdog timer driver core does: to send a keepalive ping to the watchdog 151 timer hardware it will either use the ping operation (when available) or the 152 start operation (when the ping operation is not available). 153 (Note: the WDIOC_KEEPALIVE ioctl call will only be active when the 154 WDIOF_KEEPALIVEPING bit has been set in the option field on the watchdog's 155 info structure). 156* status: this routine checks the status of the watchdog timer device. The 157 status of the device is reported with watchdog WDIOF_* status flags/bits. 158* set_timeout: this routine checks and changes the timeout of the watchdog 159 timer device. It returns 0 on success, -EINVAL for "parameter out of range" 160 and -EIO for "could not write value to the watchdog". On success this 161 routine should set the timeout value of the watchdog_device to the 162 achieved timeout value (which may be different from the requested one 163 because the watchdog does not necessarily has a 1 second resolution). 164 (Note: the WDIOF_SETTIMEOUT needs to be set in the options field of the 165 watchdog's info structure). 166* get_timeleft: this routines returns the time that's left before a reset. 167* ref: the operation that calls kref_get on the kref of a dynamically 168 allocated watchdog_device struct. 169* unref: the operation that calls kref_put on the kref of a dynamically 170 allocated watchdog_device struct. 171* ioctl: if this routine is present then it will be called first before we do 172 our own internal ioctl call handling. This routine should return -ENOIOCTLCMD 173 if a command is not supported. The parameters that are passed to the ioctl 174 call are: watchdog_device, cmd and arg. 175 176The status bits should (preferably) be set with the set_bit and clear_bit alike 177bit-operations. The status bits that are defined are: 178* WDOG_ACTIVE: this status bit indicates whether or not a watchdog timer device 179 is active or not. When the watchdog is active after booting, then you should 180 set this status bit (Note: when you register the watchdog timer device with 181 this bit set, then opening /dev/watchdog will skip the start operation) 182* WDOG_DEV_OPEN: this status bit shows whether or not the watchdog device 183 was opened via /dev/watchdog. 184 (This bit should only be used by the WatchDog Timer Driver Core). 185* WDOG_ALLOW_RELEASE: this bit stores whether or not the magic close character 186 has been sent (so that we can support the magic close feature). 187 (This bit should only be used by the WatchDog Timer Driver Core). 188* WDOG_NO_WAY_OUT: this bit stores the nowayout setting for the watchdog. 189 If this bit is set then the watchdog timer will not be able to stop. 190* WDOG_UNREGISTERED: this bit gets set by the WatchDog Timer Driver Core 191 after calling watchdog_unregister_device, and then checked before calling 192 any watchdog_ops, so that you can be sure that no operations (other then 193 unref) will get called after unregister, even if userspace still holds a 194 reference to /dev/watchdog 195 196 To set the WDOG_NO_WAY_OUT status bit (before registering your watchdog 197 timer device) you can either: 198 * set it statically in your watchdog_device struct with 199 .status = WATCHDOG_NOWAYOUT_INIT_STATUS, 200 (this will set the value the same as CONFIG_WATCHDOG_NOWAYOUT) or 201 * use the following helper function: 202 static inline void watchdog_set_nowayout(struct watchdog_device *wdd, int nowayout) 203 204Note: The WatchDog Timer Driver Core supports the magic close feature and 205the nowayout feature. To use the magic close feature you must set the 206WDIOF_MAGICCLOSE bit in the options field of the watchdog's info structure. 207The nowayout feature will overrule the magic close feature. 208 209To get or set driver specific data the following two helper functions should be 210used: 211 212static inline void watchdog_set_drvdata(struct watchdog_device *wdd, void *data) 213static inline void *watchdog_get_drvdata(struct watchdog_device *wdd) 214 215The watchdog_set_drvdata function allows you to add driver specific data. The 216arguments of this function are the watchdog device where you want to add the 217driver specific data to and a pointer to the data itself. 218 219The watchdog_get_drvdata function allows you to retrieve driver specific data. 220The argument of this function is the watchdog device where you want to retrieve 221data from. The function returns the pointer to the driver specific data. 222 223To initialize the timeout field, the following function can be used: 224 225extern int watchdog_init_timeout(struct watchdog_device *wdd, 226 unsigned int timeout_parm, struct device *dev); 227 228The watchdog_init_timeout function allows you to initialize the timeout field 229using the module timeout parameter or by retrieving the timeout-sec property from 230the device tree (if the module timeout parameter is invalid). Best practice is 231to set the default timeout value as timeout value in the watchdog_device and 232then use this function to set the user "preferred" timeout value. 233This routine returns zero on success and a negative errno code for failure. 234