1=============================================== 2The Linux WatchDog Timer Driver Core kernel API 3=============================================== 4 5Last reviewed: 12-Feb-2013 6 7Wim Van Sebroeck <wim@iguana.be> 8 9Introduction 10------------ 11This document does not describe what a WatchDog Timer (WDT) Driver or Device is. 12It also does not describe the API which can be used by user space to communicate 13with a WatchDog Timer. If you want to know this then please read the following 14file: Documentation/watchdog/watchdog-api.rst . 15 16So what does this document describe? It describes the API that can be used by 17WatchDog Timer Drivers that want to use the WatchDog Timer Driver Core 18Framework. This framework provides all interfacing towards user space so that 19the same code does not have to be reproduced each time. This also means that 20a watchdog timer driver then only needs to provide the different routines 21(operations) that control the watchdog timer (WDT). 22 23The API 24------- 25Each watchdog timer driver that wants to use the WatchDog Timer Driver Core 26must #include <linux/watchdog.h> (you would have to do this anyway when 27writing a watchdog device driver). This include file contains following 28register/unregister routines:: 29 30 extern int watchdog_register_device(struct watchdog_device *); 31 extern void watchdog_unregister_device(struct watchdog_device *); 32 33The watchdog_register_device routine registers a watchdog timer device. 34The parameter of this routine is a pointer to a watchdog_device structure. 35This routine returns zero on success and a negative errno code for failure. 36 37The watchdog_unregister_device routine deregisters a registered watchdog timer 38device. The parameter of this routine is the pointer to the registered 39watchdog_device structure. 40 41The watchdog subsystem includes an registration deferral mechanism, 42which allows you to register an watchdog as early as you wish during 43the boot process. 44 45The watchdog device structure looks like this:: 46 47 struct watchdog_device { 48 int id; 49 struct device *parent; 50 const struct attribute_group **groups; 51 const struct watchdog_info *info; 52 const struct watchdog_ops *ops; 53 const struct watchdog_governor *gov; 54 unsigned int bootstatus; 55 unsigned int timeout; 56 unsigned int pretimeout; 57 unsigned int min_timeout; 58 unsigned int max_timeout; 59 unsigned int min_hw_heartbeat_ms; 60 unsigned int max_hw_heartbeat_ms; 61 struct notifier_block reboot_nb; 62 struct notifier_block restart_nb; 63 void *driver_data; 64 struct watchdog_core_data *wd_data; 65 unsigned long status; 66 struct list_head deferred; 67 }; 68 69It contains following fields: 70 71* id: set by watchdog_register_device, id 0 is special. It has both a 72 /dev/watchdog0 cdev (dynamic major, minor 0) as well as the old 73 /dev/watchdog miscdev. The id is set automatically when calling 74 watchdog_register_device. 75* parent: set this to the parent device (or NULL) before calling 76 watchdog_register_device. 77* groups: List of sysfs attribute groups to create when creating the watchdog 78 device. 79* info: a pointer to a watchdog_info structure. This structure gives some 80 additional information about the watchdog timer itself. (Like its unique name) 81* ops: a pointer to the list of watchdog operations that the watchdog supports. 82* gov: a pointer to the assigned watchdog device pretimeout governor or NULL. 83* timeout: the watchdog timer's timeout value (in seconds). 84 This is the time after which the system will reboot if user space does 85 not send a heartbeat request if WDOG_ACTIVE is set. 86* pretimeout: the watchdog timer's pretimeout value (in seconds). 87* min_timeout: the watchdog timer's minimum timeout value (in seconds). 88 If set, the minimum configurable value for 'timeout'. 89* max_timeout: the watchdog timer's maximum timeout value (in seconds), 90 as seen from userspace. If set, the maximum configurable value for 91 'timeout'. Not used if max_hw_heartbeat_ms is non-zero. 92* min_hw_heartbeat_ms: Hardware limit for minimum time between heartbeats, 93 in milli-seconds. This value is normally 0; it should only be provided 94 if the hardware can not tolerate lower intervals between heartbeats. 95* max_hw_heartbeat_ms: Maximum hardware heartbeat, in milli-seconds. 96 If set, the infrastructure will send heartbeats to the watchdog driver 97 if 'timeout' is larger than max_hw_heartbeat_ms, unless WDOG_ACTIVE 98 is set and userspace failed to send a heartbeat for at least 'timeout' 99 seconds. max_hw_heartbeat_ms must be set if a driver does not implement 100 the stop function. 101* reboot_nb: notifier block that is registered for reboot notifications, for 102 internal use only. If the driver calls watchdog_stop_on_reboot, watchdog core 103 will stop the watchdog on such notifications. 104* restart_nb: notifier block that is registered for machine restart, for 105 internal use only. If a watchdog is capable of restarting the machine, it 106 should define ops->restart. Priority can be changed through 107 watchdog_set_restart_priority. 108* bootstatus: status of the device after booting (reported with watchdog 109 WDIOF_* status bits). 110* driver_data: a pointer to the drivers private data of a watchdog device. 111 This data should only be accessed via the watchdog_set_drvdata and 112 watchdog_get_drvdata routines. 113* wd_data: a pointer to watchdog core internal data. 114* status: this field contains a number of status bits that give extra 115 information about the status of the device (Like: is the watchdog timer 116 running/active, or is the nowayout bit set). 117* deferred: entry in wtd_deferred_reg_list which is used to 118 register early initialized watchdogs. 119 120The list of watchdog operations is defined as:: 121 122 struct watchdog_ops { 123 struct module *owner; 124 /* mandatory operations */ 125 int (*start)(struct watchdog_device *); 126 /* optional operations */ 127 int (*stop)(struct watchdog_device *); 128 int (*ping)(struct watchdog_device *); 129 unsigned int (*status)(struct watchdog_device *); 130 int (*set_timeout)(struct watchdog_device *, unsigned int); 131 int (*set_pretimeout)(struct watchdog_device *, unsigned int); 132 unsigned int (*get_timeleft)(struct watchdog_device *); 133 int (*restart)(struct watchdog_device *); 134 long (*ioctl)(struct watchdog_device *, unsigned int, unsigned long); 135 }; 136 137It is important that you first define the module owner of the watchdog timer 138driver's operations. This module owner will be used to lock the module when 139the watchdog is active. (This to avoid a system crash when you unload the 140module and /dev/watchdog is still open). 141 142Some operations are mandatory and some are optional. The mandatory operations 143are: 144 145* start: this is a pointer to the routine that starts the watchdog timer 146 device. 147 The routine needs a pointer to the watchdog timer device structure as a 148 parameter. It returns zero on success or a negative errno code for failure. 149 150Not all watchdog timer hardware supports the same functionality. That's why 151all other routines/operations are optional. They only need to be provided if 152they are supported. These optional routines/operations are: 153 154* stop: with this routine the watchdog timer device is being stopped. 155 156 The routine needs a pointer to the watchdog timer device structure as a 157 parameter. It returns zero on success or a negative errno code for failure. 158 Some watchdog timer hardware can only be started and not be stopped. A 159 driver supporting such hardware does not have to implement the stop routine. 160 161 If a driver has no stop function, the watchdog core will set WDOG_HW_RUNNING 162 and start calling the driver's keepalive pings function after the watchdog 163 device is closed. 164 165 If a watchdog driver does not implement the stop function, it must set 166 max_hw_heartbeat_ms. 167* ping: this is the routine that sends a keepalive ping to the watchdog timer 168 hardware. 169 170 The routine needs a pointer to the watchdog timer device structure as a 171 parameter. It returns zero on success or a negative errno code for failure. 172 173 Most hardware that does not support this as a separate function uses the 174 start function to restart the watchdog timer hardware. And that's also what 175 the watchdog timer driver core does: to send a keepalive ping to the watchdog 176 timer hardware it will either use the ping operation (when available) or the 177 start operation (when the ping operation is not available). 178 179 (Note: the WDIOC_KEEPALIVE ioctl call will only be active when the 180 WDIOF_KEEPALIVEPING bit has been set in the option field on the watchdog's 181 info structure). 182* status: this routine checks the status of the watchdog timer device. The 183 status of the device is reported with watchdog WDIOF_* status flags/bits. 184 185 WDIOF_MAGICCLOSE and WDIOF_KEEPALIVEPING are reported by the watchdog core; 186 it is not necessary to report those bits from the driver. Also, if no status 187 function is provided by the driver, the watchdog core reports the status bits 188 provided in the bootstatus variable of struct watchdog_device. 189 190* set_timeout: this routine checks and changes the timeout of the watchdog 191 timer device. It returns 0 on success, -EINVAL for "parameter out of range" 192 and -EIO for "could not write value to the watchdog". On success this 193 routine should set the timeout value of the watchdog_device to the 194 achieved timeout value (which may be different from the requested one 195 because the watchdog does not necessarily have a 1 second resolution). 196 197 Drivers implementing max_hw_heartbeat_ms set the hardware watchdog heartbeat 198 to the minimum of timeout and max_hw_heartbeat_ms. Those drivers set the 199 timeout value of the watchdog_device either to the requested timeout value 200 (if it is larger than max_hw_heartbeat_ms), or to the achieved timeout value. 201 (Note: the WDIOF_SETTIMEOUT needs to be set in the options field of the 202 watchdog's info structure). 203 204 If the watchdog driver does not have to perform any action but setting the 205 watchdog_device.timeout, this callback can be omitted. 206 207 If set_timeout is not provided but, WDIOF_SETTIMEOUT is set, the watchdog 208 infrastructure updates the timeout value of the watchdog_device internally 209 to the requested value. 210 211 If the pretimeout feature is used (WDIOF_PRETIMEOUT), then set_timeout must 212 also take care of checking if pretimeout is still valid and set up the timer 213 accordingly. This can't be done in the core without races, so it is the 214 duty of the driver. 215* set_pretimeout: this routine checks and changes the pretimeout value of 216 the watchdog. It is optional because not all watchdogs support pretimeout 217 notification. The timeout value is not an absolute time, but the number of 218 seconds before the actual timeout would happen. It returns 0 on success, 219 -EINVAL for "parameter out of range" and -EIO for "could not write value to 220 the watchdog". A value of 0 disables pretimeout notification. 221 222 (Note: the WDIOF_PRETIMEOUT needs to be set in the options field of the 223 watchdog's info structure). 224 225 If the watchdog driver does not have to perform any action but setting the 226 watchdog_device.pretimeout, this callback can be omitted. That means if 227 set_pretimeout is not provided but WDIOF_PRETIMEOUT is set, the watchdog 228 infrastructure updates the pretimeout value of the watchdog_device internally 229 to the requested value. 230 231* get_timeleft: this routines returns the time that's left before a reset. 232* restart: this routine restarts the machine. It returns 0 on success or a 233 negative errno code for failure. 234* ioctl: if this routine is present then it will be called first before we do 235 our own internal ioctl call handling. This routine should return -ENOIOCTLCMD 236 if a command is not supported. The parameters that are passed to the ioctl 237 call are: watchdog_device, cmd and arg. 238 239The status bits should (preferably) be set with the set_bit and clear_bit alike 240bit-operations. The status bits that are defined are: 241 242* WDOG_ACTIVE: this status bit indicates whether or not a watchdog timer device 243 is active or not from user perspective. User space is expected to send 244 heartbeat requests to the driver while this flag is set. 245* WDOG_NO_WAY_OUT: this bit stores the nowayout setting for the watchdog. 246 If this bit is set then the watchdog timer will not be able to stop. 247* WDOG_HW_RUNNING: Set by the watchdog driver if the hardware watchdog is 248 running. The bit must be set if the watchdog timer hardware can not be 249 stopped. The bit may also be set if the watchdog timer is running after 250 booting, before the watchdog device is opened. If set, the watchdog 251 infrastructure will send keepalives to the watchdog hardware while 252 WDOG_ACTIVE is not set. 253 Note: when you register the watchdog timer device with this bit set, 254 then opening /dev/watchdog will skip the start operation but send a keepalive 255 request instead. 256 257 To set the WDOG_NO_WAY_OUT status bit (before registering your watchdog 258 timer device) you can either: 259 260 * set it statically in your watchdog_device struct with 261 262 .status = WATCHDOG_NOWAYOUT_INIT_STATUS, 263 264 (this will set the value the same as CONFIG_WATCHDOG_NOWAYOUT) or 265 * use the following helper function:: 266 267 static inline void watchdog_set_nowayout(struct watchdog_device *wdd, 268 int nowayout) 269 270Note: 271 The WatchDog Timer Driver Core supports the magic close feature and 272 the nowayout feature. To use the magic close feature you must set the 273 WDIOF_MAGICCLOSE bit in the options field of the watchdog's info structure. 274 275The nowayout feature will overrule the magic close feature. 276 277To get or set driver specific data the following two helper functions should be 278used:: 279 280 static inline void watchdog_set_drvdata(struct watchdog_device *wdd, 281 void *data) 282 static inline void *watchdog_get_drvdata(struct watchdog_device *wdd) 283 284The watchdog_set_drvdata function allows you to add driver specific data. The 285arguments of this function are the watchdog device where you want to add the 286driver specific data to and a pointer to the data itself. 287 288The watchdog_get_drvdata function allows you to retrieve driver specific data. 289The argument of this function is the watchdog device where you want to retrieve 290data from. The function returns the pointer to the driver specific data. 291 292To initialize the timeout field, the following function can be used:: 293 294 extern int watchdog_init_timeout(struct watchdog_device *wdd, 295 unsigned int timeout_parm, 296 struct device *dev); 297 298The watchdog_init_timeout function allows you to initialize the timeout field 299using the module timeout parameter or by retrieving the timeout-sec property from 300the device tree (if the module timeout parameter is invalid). Best practice is 301to set the default timeout value as timeout value in the watchdog_device and 302then use this function to set the user "preferred" timeout value. 303This routine returns zero on success and a negative errno code for failure. 304 305To disable the watchdog on reboot, the user must call the following helper:: 306 307 static inline void watchdog_stop_on_reboot(struct watchdog_device *wdd); 308 309To disable the watchdog when unregistering the watchdog, the user must call 310the following helper. Note that this will only stop the watchdog if the 311nowayout flag is not set. 312 313:: 314 315 static inline void watchdog_stop_on_unregister(struct watchdog_device *wdd); 316 317To change the priority of the restart handler the following helper should be 318used:: 319 320 void watchdog_set_restart_priority(struct watchdog_device *wdd, int priority); 321 322User should follow the following guidelines for setting the priority: 323 324* 0: should be called in last resort, has limited restart capabilities 325* 128: default restart handler, use if no other handler is expected to be 326 available, and/or if restart is sufficient to restart the entire system 327* 255: highest priority, will preempt all other restart handlers 328 329To raise a pretimeout notification, the following function should be used:: 330 331 void watchdog_notify_pretimeout(struct watchdog_device *wdd) 332 333The function can be called in the interrupt context. If watchdog pretimeout 334governor framework (kbuild CONFIG_WATCHDOG_PRETIMEOUT_GOV symbol) is enabled, 335an action is taken by a preconfigured pretimeout governor preassigned to 336the watchdog device. If watchdog pretimeout governor framework is not 337enabled, watchdog_notify_pretimeout() prints a notification message to 338the kernel log buffer. 339 340To set the last known HW keepalive time for a watchdog, the following function 341should be used:: 342 343 int watchdog_set_last_hw_keepalive(struct watchdog_device *wdd, 344 unsigned int last_ping_ms) 345 346This function must be called immediately after watchdog registration. It 347sets the last known hardware heartbeat to have happened last_ping_ms before 348current time. Calling this is only needed if the watchdog is already running 349when probe is called, and the watchdog can only be pinged after the 350min_hw_heartbeat_ms time has passed from the last ping. 351