1 2======================================= 3Reset in QEMU: the Resettable interface 4======================================= 5 6The reset of qemu objects is handled using the resettable interface declared 7in ``include/hw/resettable.h``. 8 9This interface allows objects to be grouped (on a tree basis); so that the 10whole group can be reset consistently. Each individual member object does not 11have to care about others; in particular, problems of order (which object is 12reset first) are addressed. 13 14The main object types which implement this interface are DeviceClass 15and BusClass. 16 17Triggering reset 18---------------- 19 20This section documents the APIs which "users" of a resettable object should use 21to control it. All resettable control functions must be called while holding 22the BQL. 23 24You can apply a reset to an object using ``resettable_assert_reset()``. You need 25to call ``resettable_release_reset()`` to release the object from reset. To 26instantly reset an object, without keeping it in reset state, just call 27``resettable_reset()``. These functions take two parameters: a pointer to the 28object to reset and a reset type. 29 30The Resettable interface handles reset types with an enum ``ResetType``: 31 32``RESET_TYPE_COLD`` 33 Cold reset is supported by every resettable object. In QEMU, it means we reset 34 to the initial state corresponding to the start of QEMU; this might differ 35 from what is a real hardware cold reset. It differs from other resets (like 36 warm or bus resets) which may keep certain parts untouched. 37 38``RESET_TYPE_SNAPSHOT_LOAD`` 39 This is called for a reset which is being done to put the system into a 40 clean state prior to loading a snapshot. (This corresponds to a reset 41 with ``SHUTDOWN_CAUSE_SNAPSHOT_LOAD``.) Almost all devices should treat 42 this the same as ``RESET_TYPE_COLD``. The main exception is devices which 43 have some non-deterministic state they want to reinitialize to a different 44 value on each cold reset, such as RNG seed information, and which they 45 must not reinitialize on a snapshot-load reset. 46 47``RESET_TYPE_WAKEUP`` 48 If the machine supports waking up from a suspended state and needs to reset 49 its devices during wake-up (from the ``MachineClass::wakeup()`` method), this 50 reset type should be used for such a request. Devices can utilize this reset 51 type to differentiate the reset requested during machine wake-up from other 52 reset requests. For example, RAM content must not be lost during wake-up, and 53 memory devices like virtio-mem that provide additional RAM must not reset 54 such state during wake-ups, but might do so during cold resets. However, this 55 reset type should not be used for wake-up detection, as not every machine 56 type issues a device reset request during wake-up. 57 58``RESET_TYPE_S390_CPU_NORMAL`` 59 This is only used for S390 CPU objects; it clears interrupts, stops 60 processing, and clears the TLB, but does not touch register contents. 61 62``RESET_TYPE_S390_CPU_INITIAL`` 63 This is only used for S390 CPU objects; it does everything 64 ``RESET_TYPE_S390_CPU_NORMAL`` does and also clears the PSW, prefix, 65 FPC, timer and control registers. It does not touch gprs, fprs or acrs. 66 67Devices which implement reset methods must treat any unknown ``ResetType`` 68as equivalent to ``RESET_TYPE_COLD``; this will reduce the amount of 69existing code we need to change if we add more types in future. 70 71Calling ``resettable_reset()`` is equivalent to calling 72``resettable_assert_reset()`` then ``resettable_release_reset()``. It is 73possible to interleave multiple calls to these three functions. There may 74be several reset sources/controllers of a given object. The interface handles 75everything and the different reset controllers do not need to know anything 76about each others. The object will leave reset state only when each other 77controllers end their reset operation. This point is handled internally by 78maintaining a count of in-progress resets; it is crucial to call 79``resettable_release_reset()`` one time and only one time per 80``resettable_assert_reset()`` call. 81 82For now migration of a device or bus in reset is not supported. Care must be 83taken not to delay ``resettable_release_reset()`` after its 84``resettable_assert_reset()`` counterpart. 85 86Note that, since resettable is an interface, the API takes a simple Object as 87parameter. Still, it is a programming error to call a resettable function on a 88non-resettable object and it will trigger a run time assert error. Since most 89calls to resettable interface are done through base class functions, such an 90error is not likely to happen. 91 92For Devices and Buses, the following helper functions exist: 93 94- ``device_cold_reset()`` 95- ``bus_cold_reset()`` 96 97These are simple wrappers around resettable_reset() function; they only cast the 98Device or Bus into an Object and pass the cold reset type. When possible 99prefer to use these functions instead of ``resettable_reset()``. 100 101Device and bus functions co-exist because there can be semantic differences 102between resetting a bus and resetting the controller bridge which owns it. 103For example, consider a SCSI controller. Resetting the controller puts all 104its registers back to what reset state was as well as reset everything on the 105SCSI bus, whereas resetting just the SCSI bus only resets everything that's on 106it but not the controller. 107 108 109Multi-phase mechanism 110--------------------- 111 112This section documents the internals of the resettable interface. 113 114The resettable interface uses a multi-phase system to relieve objects and 115machines from reset ordering problems. To address this, the reset operation 116of an object is split into three well defined phases. 117 118When resetting several objects (for example the whole machine at simulation 119startup), all first phases of all objects are executed, then all second phases 120and then all third phases. 121 122The three phases are: 123 1241. The **enter** phase is executed when the object enters reset. It resets only 125 local state of the object; it must not do anything that has a side-effect 126 on other objects, such as raising or lowering a qemu_irq line or reading or 127 writing guest memory. 128 1292. The **hold** phase is executed for entry into reset, once every object in the 130 group which is being reset has had its *enter* phase executed. At this point 131 devices can do actions that affect other objects. 132 1333. The **exit** phase is executed when the object leaves the reset state. 134 Actions affecting other objects are permitted. 135 136As said in previous section, the interface maintains a count of reset. This 137count is used to ensure phases are executed only when required. *enter* and 138*hold* phases are executed only when asserting reset for the first time 139(if an object is already in reset state when calling 140``resettable_assert_reset()`` or ``resettable_reset()``, they are not 141executed). 142The *exit* phase is executed only when the last reset operation ends. Therefore 143the object does not need to care how many of reset controllers it has and how 144many of them have started a reset. 145 146 147Handling reset in a resettable object 148------------------------------------- 149 150This section documents the APIs that an implementation of a resettable object 151must provide and what functions it has access to. It is intended for people 152who want to implement or convert a class which has the resettable interface; 153for example when specializing an existing device or bus. 154 155Methods to implement 156.................... 157 158Three methods should be defined or left empty. Each method corresponds to a 159phase of the reset; they are name ``phases.enter()``, ``phases.hold()`` and 160``phases.exit()``. They all take the object as parameter. The *enter* method 161also take the reset type as second parameter. 162 163When extending an existing class, these methods may need to be extended too. 164The ``resettable_class_set_parent_phases()`` class function may be used to 165backup parent class methods. 166 167Here follows an example to implement reset for a Device which sets an IO while 168in reset. 169 170:: 171 172 static void mydev_reset_enter(Object *obj, ResetType type) 173 { 174 MyDevClass *myclass = MYDEV_GET_CLASS(obj); 175 MyDevState *mydev = MYDEV(obj); 176 /* call parent class enter phase */ 177 if (myclass->parent_phases.enter) { 178 myclass->parent_phases.enter(obj, type); 179 } 180 /* initialize local state only */ 181 mydev->var = 0; 182 } 183 184 static void mydev_reset_hold(Object *obj, ResetType type) 185 { 186 MyDevClass *myclass = MYDEV_GET_CLASS(obj); 187 MyDevState *mydev = MYDEV(obj); 188 /* call parent class hold phase */ 189 if (myclass->parent_phases.hold) { 190 myclass->parent_phases.hold(obj, type); 191 } 192 /* set an IO */ 193 qemu_set_irq(mydev->irq, 1); 194 } 195 196 static void mydev_reset_exit(Object *obj, ResetType type) 197 { 198 MyDevClass *myclass = MYDEV_GET_CLASS(obj); 199 MyDevState *mydev = MYDEV(obj); 200 /* call parent class exit phase */ 201 if (myclass->parent_phases.exit) { 202 myclass->parent_phases.exit(obj, type); 203 } 204 /* clear an IO */ 205 qemu_set_irq(mydev->irq, 0); 206 } 207 208 typedef struct MyDevClass { 209 MyParentClass parent_class; 210 /* to store eventual parent reset methods */ 211 ResettablePhases parent_phases; 212 } MyDevClass; 213 214 static void mydev_class_init(ObjectClass *class, void *data) 215 { 216 MyDevClass *myclass = MYDEV_CLASS(class); 217 ResettableClass *rc = RESETTABLE_CLASS(class); 218 resettable_class_set_parent_phases(rc, 219 mydev_reset_enter, 220 mydev_reset_hold, 221 mydev_reset_exit, 222 &myclass->parent_phases); 223 } 224 225In the above example, we override all three phases. It is possible to override 226only some of them by passing NULL instead of a function pointer to 227``resettable_class_set_parent_phases()``. For example, the following will 228only override the *enter* phase and leave *hold* and *exit* untouched:: 229 230 resettable_class_set_parent_phases(rc, mydev_reset_enter, NULL, NULL, 231 &myclass->parent_phases); 232 233This is equivalent to providing a trivial implementation of the hold and exit 234phases which does nothing but call the parent class's implementation of the 235phase. 236 237Polling the reset state 238....................... 239 240Resettable interface provides the ``resettable_is_in_reset()`` function. 241This function returns true if the object parameter is currently under reset. 242 243An object is under reset from the beginning of the *enter* phase (before 244either its children or its own enter method is called) to the *exit* 245phase. During *enter* and *hold* phase only, the function will return that the 246object is in reset. The state is changed after the *exit* is propagated to 247its children and just before calling the object's own *exit* method. 248 249This function may be used if the object behavior has to be adapted 250while in reset state. For example if a device has an irq input, 251it will probably need to ignore it while in reset; then it can for 252example check the reset state at the beginning of the irq callback. 253 254Note that until migration of the reset state is supported, an object 255should not be left in reset. So apart from being currently executing 256one of the reset phases, the only cases when this function will return 257true is if an external interaction (like changing an io) is made during 258*hold* or *exit* phase of another object in the same reset group. 259 260Helpers ``device_is_in_reset()`` and ``bus_is_in_reset()`` are also provided 261for devices and buses and should be preferred. 262 263 264Base class handling of reset 265---------------------------- 266 267This section documents parts of the reset mechanism that you only need to know 268about if you are extending it to work with a new base class other than 269DeviceClass or BusClass, or maintaining the existing code in those classes. Most 270people can ignore it. 271 272Methods to implement 273.................... 274 275There are two other methods that need to exist in a class implementing the 276interface: ``get_state()`` and ``child_foreach()``. 277 278``get_state()`` is simple. *resettable* is an interface and, as a consequence, 279does not have any class state structure. But in order to factorize the code, we 280need one. This method must return a pointer to ``ResettableState`` structure. 281The structure must be allocated by the base class; preferably it should be 282located inside the object instance structure. 283 284``child_foreach()`` is more complex. It should execute the given callback on 285every reset child of the given resettable object. All children must be 286resettable too. Additional parameters (a reset type and an opaque pointer) must 287be passed to the callback too. 288 289In ``DeviceClass`` and ``BusClass`` the ``ResettableState`` is located in the 290``DeviceState`` and ``BusState`` structures. ``child_foreach()`` is implemented 291to follow the bus hierarchy; for a bus, it calls the function on every child 292device; for a device, it calls the function on every bus child. When we reset 293the main system bus, we reset the whole machine bus tree. 294 295Changing a resettable parent 296............................ 297 298One thing which should be taken care of by the base class is handling reset 299hierarchy changes. 300 301The reset hierarchy is supposed to be static and built during machine creation. 302But there are actually some exceptions. To cope with this, the resettable API 303provides ``resettable_change_parent()``. This function allows to set, update or 304remove the parent of a resettable object after machine creation is done. As 305parameters, it takes the object being moved, the old parent if any and the new 306parent if any. 307 308This function can be used at any time when not in a reset operation. During 309a reset operation it must be used only in *hold* phase. Using it in *enter* or 310*exit* phase is an error. 311Also it should not be used during machine creation, although it is harmless to 312do so: the function is a no-op as long as old and new parent are NULL or not 313in reset. 314 315There is currently 2 cases where this function is used: 316 3171. *device hotplug*; it means a new device is introduced on a live bus. 318 3192. *hot bus change*; it means an existing live device is added, moved or 320 removed in the bus hierarchy. At the moment, it occurs only in the raspi 321 machines for changing the sdbus used by sd card. 322 323Reset of the complete system 324---------------------------- 325 326Reset of the complete system is a little complicated. The typical 327flow is: 328 3291. Code which wishes to reset the entire system does so by calling 330 ``qemu_system_reset_request()``. This schedules a reset, but the 331 reset will happen asynchronously after the function returns. 332 That makes this safe to call from, for example, device models. 333 3342. The function which is called to make the reset happen is 335 ``qemu_system_reset()``. Generally only core system code should 336 call this directly. 337 3383. ``qemu_system_reset()`` calls the ``MachineClass::reset`` method of 339 the current machine, if it has one. That method must call 340 ``qemu_devices_reset()``. If the machine has no reset method, 341 ``qemu_system_reset()`` calls ``qemu_devices_reset()`` directly. 342 3434. ``qemu_devices_reset()`` performs a reset of the system, using 344 the three-phase mechanism listed above. It resets all objects 345 that were registered with it using ``qemu_register_resettable()``. 346 It also calls all the functions registered with it using 347 ``qemu_register_reset()``. Those functions are called during the 348 "hold" phase of this reset. 349 3505. The most important object that this reset resets is the 351 'sysbus' bus. The sysbus bus is the root of the qbus tree. This 352 means that all devices on the sysbus are reset, and all their 353 child buses, and all the devices on those child buses. 354 3556. Devices which are not on the qbus tree are *not* automatically 356 reset! (The most obvious example of this is CPU objects, but 357 anything that directly inherits from ``TYPE_OBJECT`` or ``TYPE_DEVICE`` 358 rather than from ``TYPE_SYS_BUS_DEVICE`` or some other plugs-into-a-bus 359 type will be in this category.) You need to therefore arrange for these 360 to be reset in some other way (e.g. using ``qemu_register_resettable()`` 361 or ``qemu_register_reset()``). 362