linux/Documentation/video4linux/v4l2-framework.txt
<<
>>
Prefs
   1Overview of the V4L2 driver framework
   2=====================================
   3
   4This text documents the various structures provided by the V4L2 framework and
   5their relationships.
   6
   7
   8Introduction
   9------------
  10
  11The V4L2 drivers tend to be very complex due to the complexity of the
  12hardware: most devices have multiple ICs, export multiple device nodes in
  13/dev, and create also non-V4L2 devices such as DVB, ALSA, FB, I2C and input
  14(IR) devices.
  15
  16Especially the fact that V4L2 drivers have to setup supporting ICs to
  17do audio/video muxing/encoding/decoding makes it more complex than most.
  18Usually these ICs are connected to the main bridge driver through one or
  19more I2C busses, but other busses can also be used. Such devices are
  20called 'sub-devices'.
  21
  22For a long time the framework was limited to the video_device struct for
  23creating V4L device nodes and video_buf for handling the video buffers
  24(note that this document does not discuss the video_buf framework).
  25
  26This meant that all drivers had to do the setup of device instances and
  27connecting to sub-devices themselves. Some of this is quite complicated
  28to do right and many drivers never did do it correctly.
  29
  30There is also a lot of common code that could never be refactored due to
  31the lack of a framework.
  32
  33So this framework sets up the basic building blocks that all drivers
  34need and this same framework should make it much easier to refactor
  35common code into utility functions shared by all drivers.
  36
  37
  38Structure of a driver
  39---------------------
  40
  41All drivers have the following structure:
  42
  431) A struct for each device instance containing the device state.
  44
  452) A way of initializing and commanding sub-devices (if any).
  46
  473) Creating V4L2 device nodes (/dev/videoX, /dev/vbiX and /dev/radioX)
  48   and keeping track of device-node specific data.
  49
  504) Filehandle-specific structs containing per-filehandle data;
  51
  525) video buffer handling.
  53
  54This is a rough schematic of how it all relates:
  55
  56    device instances
  57      |
  58      +-sub-device instances
  59      |
  60      \-V4L2 device nodes
  61          |
  62          \-filehandle instances
  63
  64
  65Structure of the framework
  66--------------------------
  67
  68The framework closely resembles the driver structure: it has a v4l2_device
  69struct for the device instance data, a v4l2_subdev struct to refer to
  70sub-device instances, the video_device struct stores V4L2 device node data
  71and in the future a v4l2_fh struct will keep track of filehandle instances
  72(this is not yet implemented).
  73
  74The V4L2 framework also optionally integrates with the media framework. If a
  75driver sets the struct v4l2_device mdev field, sub-devices and video nodes
  76will automatically appear in the media framework as entities.
  77
  78
  79struct v4l2_device
  80------------------
  81
  82Each device instance is represented by a struct v4l2_device (v4l2-device.h).
  83Very simple devices can just allocate this struct, but most of the time you
  84would embed this struct inside a larger struct.
  85
  86You must register the device instance:
  87
  88        v4l2_device_register(struct device *dev, struct v4l2_device *v4l2_dev);
  89
  90Registration will initialize the v4l2_device struct. If the dev->driver_data
  91field is NULL, it will be linked to v4l2_dev.
  92
  93Drivers that want integration with the media device framework need to set
  94dev->driver_data manually to point to the driver-specific device structure
  95that embed the struct v4l2_device instance. This is achieved by a
  96dev_set_drvdata() call before registering the V4L2 device instance. They must
  97also set the struct v4l2_device mdev field to point to a properly initialized
  98and registered media_device instance.
  99
 100If v4l2_dev->name is empty then it will be set to a value derived from dev
 101(driver name followed by the bus_id, to be precise). If you set it up before
 102calling v4l2_device_register then it will be untouched. If dev is NULL, then
 103you *must* setup v4l2_dev->name before calling v4l2_device_register.
 104
 105You can use v4l2_device_set_name() to set the name based on a driver name and
 106a driver-global atomic_t instance. This will generate names like ivtv0, ivtv1,
 107etc. If the name ends with a digit, then it will insert a dash: cx18-0,
 108cx18-1, etc. This function returns the instance number.
 109
 110The first 'dev' argument is normally the struct device pointer of a pci_dev,
 111usb_interface or platform_device. It is rare for dev to be NULL, but it happens
 112with ISA devices or when one device creates multiple PCI devices, thus making
 113it impossible to associate v4l2_dev with a particular parent.
 114
 115You can also supply a notify() callback that can be called by sub-devices to
 116notify you of events. Whether you need to set this depends on the sub-device.
 117Any notifications a sub-device supports must be defined in a header in
 118include/media/<subdevice>.h.
 119
 120You unregister with:
 121
 122        v4l2_device_unregister(struct v4l2_device *v4l2_dev);
 123
 124If the dev->driver_data field points to v4l2_dev, it will be reset to NULL.
 125Unregistering will also automatically unregister all subdevs from the device.
 126
 127If you have a hotpluggable device (e.g. a USB device), then when a disconnect
 128happens the parent device becomes invalid. Since v4l2_device has a pointer to
 129that parent device it has to be cleared as well to mark that the parent is
 130gone. To do this call:
 131
 132        v4l2_device_disconnect(struct v4l2_device *v4l2_dev);
 133
 134This does *not* unregister the subdevs, so you still need to call the
 135v4l2_device_unregister() function for that. If your driver is not hotpluggable,
 136then there is no need to call v4l2_device_disconnect().
 137
 138Sometimes you need to iterate over all devices registered by a specific
 139driver. This is usually the case if multiple device drivers use the same
 140hardware. E.g. the ivtvfb driver is a framebuffer driver that uses the ivtv
 141hardware. The same is true for alsa drivers for example.
 142
 143You can iterate over all registered devices as follows:
 144
 145static int callback(struct device *dev, void *p)
 146{
 147        struct v4l2_device *v4l2_dev = dev_get_drvdata(dev);
 148
 149        /* test if this device was inited */
 150        if (v4l2_dev == NULL)
 151                return 0;
 152        ...
 153        return 0;
 154}
 155
 156int iterate(void *p)
 157{
 158        struct device_driver *drv;
 159        int err;
 160
 161        /* Find driver 'ivtv' on the PCI bus.
 162           pci_bus_type is a global. For USB busses use usb_bus_type. */
 163        drv = driver_find("ivtv", &pci_bus_type);
 164        /* iterate over all ivtv device instances */
 165        err = driver_for_each_device(drv, NULL, p, callback);
 166        put_driver(drv);
 167        return err;
 168}
 169
 170Sometimes you need to keep a running counter of the device instance. This is
 171commonly used to map a device instance to an index of a module option array.
 172
 173The recommended approach is as follows:
 174
 175static atomic_t drv_instance = ATOMIC_INIT(0);
 176
 177static int __devinit drv_probe(struct pci_dev *pdev,
 178                                const struct pci_device_id *pci_id)
 179{
 180        ...
 181        state->instance = atomic_inc_return(&drv_instance) - 1;
 182}
 183
 184If you have multiple device nodes then it can be difficult to know when it is
 185safe to unregister v4l2_device for hotpluggable devices. For this purpose
 186v4l2_device has refcounting support. The refcount is increased whenever
 187video_register_device is called and it is decreased whenever that device node
 188is released. When the refcount reaches zero, then the v4l2_device release()
 189callback is called. You can do your final cleanup there.
 190
 191If other device nodes (e.g. ALSA) are created, then you can increase and
 192decrease the refcount manually as well by calling:
 193
 194void v4l2_device_get(struct v4l2_device *v4l2_dev);
 195
 196or:
 197
 198int v4l2_device_put(struct v4l2_device *v4l2_dev);
 199
 200Since the initial refcount is 1 you also need to call v4l2_device_put in the
 201disconnect() callback (for USB devices) or in the remove() callback (for e.g.
 202PCI devices), otherwise the refcount will never reach 0.
 203
 204struct v4l2_subdev
 205------------------
 206
 207Many drivers need to communicate with sub-devices. These devices can do all
 208sort of tasks, but most commonly they handle audio and/or video muxing,
 209encoding or decoding. For webcams common sub-devices are sensors and camera
 210controllers.
 211
 212Usually these are I2C devices, but not necessarily. In order to provide the
 213driver with a consistent interface to these sub-devices the v4l2_subdev struct
 214(v4l2-subdev.h) was created.
 215
 216Each sub-device driver must have a v4l2_subdev struct. This struct can be
 217stand-alone for simple sub-devices or it might be embedded in a larger struct
 218if more state information needs to be stored. Usually there is a low-level
 219device struct (e.g. i2c_client) that contains the device data as setup
 220by the kernel. It is recommended to store that pointer in the private
 221data of v4l2_subdev using v4l2_set_subdevdata(). That makes it easy to go
 222from a v4l2_subdev to the actual low-level bus-specific device data.
 223
 224You also need a way to go from the low-level struct to v4l2_subdev. For the
 225common i2c_client struct the i2c_set_clientdata() call is used to store a
 226v4l2_subdev pointer, for other busses you may have to use other methods.
 227
 228Bridges might also need to store per-subdev private data, such as a pointer to
 229bridge-specific per-subdev private data. The v4l2_subdev structure provides
 230host private data for that purpose that can be accessed with
 231v4l2_get_subdev_hostdata() and v4l2_set_subdev_hostdata().
 232
 233From the bridge driver perspective you load the sub-device module and somehow
 234obtain the v4l2_subdev pointer. For i2c devices this is easy: you call
 235i2c_get_clientdata(). For other busses something similar needs to be done.
 236Helper functions exists for sub-devices on an I2C bus that do most of this
 237tricky work for you.
 238
 239Each v4l2_subdev contains function pointers that sub-device drivers can
 240implement (or leave NULL if it is not applicable). Since sub-devices can do
 241so many different things and you do not want to end up with a huge ops struct
 242of which only a handful of ops are commonly implemented, the function pointers
 243are sorted according to category and each category has its own ops struct.
 244
 245The top-level ops struct contains pointers to the category ops structs, which
 246may be NULL if the subdev driver does not support anything from that category.
 247
 248It looks like this:
 249
 250struct v4l2_subdev_core_ops {
 251        int (*g_chip_ident)(struct v4l2_subdev *sd, struct v4l2_dbg_chip_ident *chip);
 252        int (*log_status)(struct v4l2_subdev *sd);
 253        int (*init)(struct v4l2_subdev *sd, u32 val);
 254        ...
 255};
 256
 257struct v4l2_subdev_tuner_ops {
 258        ...
 259};
 260
 261struct v4l2_subdev_audio_ops {
 262        ...
 263};
 264
 265struct v4l2_subdev_video_ops {
 266        ...
 267};
 268
 269struct v4l2_subdev_pad_ops {
 270        ...
 271};
 272
 273struct v4l2_subdev_ops {
 274        const struct v4l2_subdev_core_ops  *core;
 275        const struct v4l2_subdev_tuner_ops *tuner;
 276        const struct v4l2_subdev_audio_ops *audio;
 277        const struct v4l2_subdev_video_ops *video;
 278        const struct v4l2_subdev_pad_ops *video;
 279};
 280
 281The core ops are common to all subdevs, the other categories are implemented
 282depending on the sub-device. E.g. a video device is unlikely to support the
 283audio ops and vice versa.
 284
 285This setup limits the number of function pointers while still making it easy
 286to add new ops and categories.
 287
 288A sub-device driver initializes the v4l2_subdev struct using:
 289
 290        v4l2_subdev_init(sd, &ops);
 291
 292Afterwards you need to initialize subdev->name with a unique name and set the
 293module owner. This is done for you if you use the i2c helper functions.
 294
 295If integration with the media framework is needed, you must initialize the
 296media_entity struct embedded in the v4l2_subdev struct (entity field) by
 297calling media_entity_init():
 298
 299        struct media_pad *pads = &my_sd->pads;
 300        int err;
 301
 302        err = media_entity_init(&sd->entity, npads, pads, 0);
 303
 304The pads array must have been previously initialized. There is no need to
 305manually set the struct media_entity type and name fields, but the revision
 306field must be initialized if needed.
 307
 308A reference to the entity will be automatically acquired/released when the
 309subdev device node (if any) is opened/closed.
 310
 311Don't forget to cleanup the media entity before the sub-device is destroyed:
 312
 313        media_entity_cleanup(&sd->entity);
 314
 315If the subdev driver intends to process video and integrate with the media
 316framework, it must implement format related functionality using
 317v4l2_subdev_pad_ops instead of v4l2_subdev_video_ops.
 318
 319In that case, the subdev driver may set the link_validate field to provide
 320its own link validation function. The link validation function is called for
 321every link in the pipeline where both of the ends of the links are V4L2
 322sub-devices. The driver is still responsible for validating the correctness
 323of the format configuration between sub-devices and video nodes.
 324
 325If link_validate op is not set, the default function
 326v4l2_subdev_link_validate_default() is used instead. This function ensures
 327that width, height and the media bus pixel code are equal on both source and
 328sink of the link. Subdev drivers are also free to use this function to
 329perform the checks mentioned above in addition to their own checks.
 330
 331A device (bridge) driver needs to register the v4l2_subdev with the
 332v4l2_device:
 333
 334        int err = v4l2_device_register_subdev(v4l2_dev, sd);
 335
 336This can fail if the subdev module disappeared before it could be registered.
 337After this function was called successfully the subdev->dev field points to
 338the v4l2_device.
 339
 340If the v4l2_device parent device has a non-NULL mdev field, the sub-device
 341entity will be automatically registered with the media device.
 342
 343You can unregister a sub-device using:
 344
 345        v4l2_device_unregister_subdev(sd);
 346
 347Afterwards the subdev module can be unloaded and sd->dev == NULL.
 348
 349You can call an ops function either directly:
 350
 351        err = sd->ops->core->g_chip_ident(sd, &chip);
 352
 353but it is better and easier to use this macro:
 354
 355        err = v4l2_subdev_call(sd, core, g_chip_ident, &chip);
 356
 357The macro will to the right NULL pointer checks and returns -ENODEV if subdev
 358is NULL, -ENOIOCTLCMD if either subdev->core or subdev->core->g_chip_ident is
 359NULL, or the actual result of the subdev->ops->core->g_chip_ident ops.
 360
 361It is also possible to call all or a subset of the sub-devices:
 362
 363        v4l2_device_call_all(v4l2_dev, 0, core, g_chip_ident, &chip);
 364
 365Any subdev that does not support this ops is skipped and error results are
 366ignored. If you want to check for errors use this:
 367
 368        err = v4l2_device_call_until_err(v4l2_dev, 0, core, g_chip_ident, &chip);
 369
 370Any error except -ENOIOCTLCMD will exit the loop with that error. If no
 371errors (except -ENOIOCTLCMD) occurred, then 0 is returned.
 372
 373The second argument to both calls is a group ID. If 0, then all subdevs are
 374called. If non-zero, then only those whose group ID match that value will
 375be called. Before a bridge driver registers a subdev it can set sd->grp_id
 376to whatever value it wants (it's 0 by default). This value is owned by the
 377bridge driver and the sub-device driver will never modify or use it.
 378
 379The group ID gives the bridge driver more control how callbacks are called.
 380For example, there may be multiple audio chips on a board, each capable of
 381changing the volume. But usually only one will actually be used when the
 382user want to change the volume. You can set the group ID for that subdev to
 383e.g. AUDIO_CONTROLLER and specify that as the group ID value when calling
 384v4l2_device_call_all(). That ensures that it will only go to the subdev
 385that needs it.
 386
 387If the sub-device needs to notify its v4l2_device parent of an event, then
 388it can call v4l2_subdev_notify(sd, notification, arg). This macro checks
 389whether there is a notify() callback defined and returns -ENODEV if not.
 390Otherwise the result of the notify() call is returned.
 391
 392The advantage of using v4l2_subdev is that it is a generic struct and does
 393not contain any knowledge about the underlying hardware. So a driver might
 394contain several subdevs that use an I2C bus, but also a subdev that is
 395controlled through GPIO pins. This distinction is only relevant when setting
 396up the device, but once the subdev is registered it is completely transparent.
 397
 398
 399V4L2 sub-device userspace API
 400-----------------------------
 401
 402Beside exposing a kernel API through the v4l2_subdev_ops structure, V4L2
 403sub-devices can also be controlled directly by userspace applications.
 404
 405Device nodes named v4l-subdevX can be created in /dev to access sub-devices
 406directly. If a sub-device supports direct userspace configuration it must set
 407the V4L2_SUBDEV_FL_HAS_DEVNODE flag before being registered.
 408
 409After registering sub-devices, the v4l2_device driver can create device nodes
 410for all registered sub-devices marked with V4L2_SUBDEV_FL_HAS_DEVNODE by calling
 411v4l2_device_register_subdev_nodes(). Those device nodes will be automatically
 412removed when sub-devices are unregistered.
 413
 414The device node handles a subset of the V4L2 API.
 415
 416VIDIOC_QUERYCTRL
 417VIDIOC_QUERYMENU
 418VIDIOC_G_CTRL
 419VIDIOC_S_CTRL
 420VIDIOC_G_EXT_CTRLS
 421VIDIOC_S_EXT_CTRLS
 422VIDIOC_TRY_EXT_CTRLS
 423
 424        The controls ioctls are identical to the ones defined in V4L2. They
 425        behave identically, with the only exception that they deal only with
 426        controls implemented in the sub-device. Depending on the driver, those
 427        controls can be also be accessed through one (or several) V4L2 device
 428        nodes.
 429
 430VIDIOC_DQEVENT
 431VIDIOC_SUBSCRIBE_EVENT
 432VIDIOC_UNSUBSCRIBE_EVENT
 433
 434        The events ioctls are identical to the ones defined in V4L2. They
 435        behave identically, with the only exception that they deal only with
 436        events generated by the sub-device. Depending on the driver, those
 437        events can also be reported by one (or several) V4L2 device nodes.
 438
 439        Sub-device drivers that want to use events need to set the
 440        V4L2_SUBDEV_USES_EVENTS v4l2_subdev::flags and initialize
 441        v4l2_subdev::nevents to events queue depth before registering the
 442        sub-device. After registration events can be queued as usual on the
 443        v4l2_subdev::devnode device node.
 444
 445        To properly support events, the poll() file operation is also
 446        implemented.
 447
 448Private ioctls
 449
 450        All ioctls not in the above list are passed directly to the sub-device
 451        driver through the core::ioctl operation.
 452
 453
 454I2C sub-device drivers
 455----------------------
 456
 457Since these drivers are so common, special helper functions are available to
 458ease the use of these drivers (v4l2-common.h).
 459
 460The recommended method of adding v4l2_subdev support to an I2C driver is to
 461embed the v4l2_subdev struct into the state struct that is created for each
 462I2C device instance. Very simple devices have no state struct and in that case
 463you can just create a v4l2_subdev directly.
 464
 465A typical state struct would look like this (where 'chipname' is replaced by
 466the name of the chip):
 467
 468struct chipname_state {
 469        struct v4l2_subdev sd;
 470        ...  /* additional state fields */
 471};
 472
 473Initialize the v4l2_subdev struct as follows:
 474
 475        v4l2_i2c_subdev_init(&state->sd, client, subdev_ops);
 476
 477This function will fill in all the fields of v4l2_subdev and ensure that the
 478v4l2_subdev and i2c_client both point to one another.
 479
 480You should also add a helper inline function to go from a v4l2_subdev pointer
 481to a chipname_state struct:
 482
 483static inline struct chipname_state *to_state(struct v4l2_subdev *sd)
 484{
 485        return container_of(sd, struct chipname_state, sd);
 486}
 487
 488Use this to go from the v4l2_subdev struct to the i2c_client struct:
 489
 490        struct i2c_client *client = v4l2_get_subdevdata(sd);
 491
 492And this to go from an i2c_client to a v4l2_subdev struct:
 493
 494        struct v4l2_subdev *sd = i2c_get_clientdata(client);
 495
 496Make sure to call v4l2_device_unregister_subdev(sd) when the remove() callback
 497is called. This will unregister the sub-device from the bridge driver. It is
 498safe to call this even if the sub-device was never registered.
 499
 500You need to do this because when the bridge driver destroys the i2c adapter
 501the remove() callbacks are called of the i2c devices on that adapter.
 502After that the corresponding v4l2_subdev structures are invalid, so they
 503have to be unregistered first. Calling v4l2_device_unregister_subdev(sd)
 504from the remove() callback ensures that this is always done correctly.
 505
 506
 507The bridge driver also has some helper functions it can use:
 508
 509struct v4l2_subdev *sd = v4l2_i2c_new_subdev(v4l2_dev, adapter,
 510               "module_foo", "chipid", 0x36, NULL);
 511
 512This loads the given module (can be NULL if no module needs to be loaded) and
 513calls i2c_new_device() with the given i2c_adapter and chip/address arguments.
 514If all goes well, then it registers the subdev with the v4l2_device.
 515
 516You can also use the last argument of v4l2_i2c_new_subdev() to pass an array
 517of possible I2C addresses that it should probe. These probe addresses are
 518only used if the previous argument is 0. A non-zero argument means that you
 519know the exact i2c address so in that case no probing will take place.
 520
 521Both functions return NULL if something went wrong.
 522
 523Note that the chipid you pass to v4l2_i2c_new_subdev() is usually
 524the same as the module name. It allows you to specify a chip variant, e.g.
 525"saa7114" or "saa7115". In general though the i2c driver autodetects this.
 526The use of chipid is something that needs to be looked at more closely at a
 527later date. It differs between i2c drivers and as such can be confusing.
 528To see which chip variants are supported you can look in the i2c driver code
 529for the i2c_device_id table. This lists all the possibilities.
 530
 531There are two more helper functions:
 532
 533v4l2_i2c_new_subdev_cfg: this function adds new irq and platform_data
 534arguments and has both 'addr' and 'probed_addrs' arguments: if addr is not
 5350 then that will be used (non-probing variant), otherwise the probed_addrs
 536are probed.
 537
 538For example: this will probe for address 0x10:
 539
 540struct v4l2_subdev *sd = v4l2_i2c_new_subdev_cfg(v4l2_dev, adapter,
 541               "module_foo", "chipid", 0, NULL, 0, I2C_ADDRS(0x10));
 542
 543v4l2_i2c_new_subdev_board uses an i2c_board_info struct which is passed
 544to the i2c driver and replaces the irq, platform_data and addr arguments.
 545
 546If the subdev supports the s_config core ops, then that op is called with
 547the irq and platform_data arguments after the subdev was setup. The older
 548v4l2_i2c_new_(probed_)subdev functions will call s_config as well, but with
 549irq set to 0 and platform_data set to NULL.
 550
 551struct video_device
 552-------------------
 553
 554The actual device nodes in the /dev directory are created using the
 555video_device struct (v4l2-dev.h). This struct can either be allocated
 556dynamically or embedded in a larger struct.
 557
 558To allocate it dynamically use:
 559
 560        struct video_device *vdev = video_device_alloc();
 561
 562        if (vdev == NULL)
 563                return -ENOMEM;
 564
 565        vdev->release = video_device_release;
 566
 567If you embed it in a larger struct, then you must set the release()
 568callback to your own function:
 569
 570        struct video_device *vdev = &my_vdev->vdev;
 571
 572        vdev->release = my_vdev_release;
 573
 574The release callback must be set and it is called when the last user
 575of the video device exits.
 576
 577The default video_device_release() callback just calls kfree to free the
 578allocated memory.
 579
 580You should also set these fields:
 581
 582- v4l2_dev: set to the v4l2_device parent device.
 583
 584- name: set to something descriptive and unique.
 585
 586- vfl_dir: set this to VFL_DIR_RX for capture devices (VFL_DIR_RX has value 0,
 587  so this is normally already the default), set to VFL_DIR_TX for output
 588  devices and VFL_DIR_M2M for mem2mem (codec) devices.
 589
 590- fops: set to the v4l2_file_operations struct.
 591
 592- ioctl_ops: if you use the v4l2_ioctl_ops to simplify ioctl maintenance
 593  (highly recommended to use this and it might become compulsory in the
 594  future!), then set this to your v4l2_ioctl_ops struct. The vfl_type and
 595  vfl_dir fields are used to disable ops that do not match the type/dir
 596  combination. E.g. VBI ops are disabled for non-VBI nodes, and output ops
 597  are disabled for a capture device. This makes it possible to provide
 598  just one v4l2_ioctl_ops struct for both vbi and video nodes.
 599
 600- lock: leave to NULL if you want to do all the locking in the driver.
 601  Otherwise you give it a pointer to a struct mutex_lock and before the
 602  unlocked_ioctl file operation is called this lock will be taken by the
 603  core and released afterwards. See the next section for more details.
 604
 605- queue: a pointer to the struct vb2_queue associated with this device node.
 606  If queue is non-NULL, and queue->lock is non-NULL, then queue->lock is
 607  used for the queuing ioctls (VIDIOC_REQBUFS, CREATE_BUFS, QBUF, DQBUF,
 608  QUERYBUF, PREPARE_BUF, STREAMON and STREAMOFF) instead of the lock above.
 609  That way the vb2 queuing framework does not have to wait for other ioctls.
 610  This queue pointer is also used by the vb2 helper functions to check for
 611  queuing ownership (i.e. is the filehandle calling it allowed to do the
 612  operation).
 613
 614- prio: keeps track of the priorities. Used to implement VIDIOC_G/S_PRIORITY.
 615  If left to NULL, then it will use the struct v4l2_prio_state in v4l2_device.
 616  If you want to have a separate priority state per (group of) device node(s),
 617  then you can point it to your own struct v4l2_prio_state.
 618
 619- parent: you only set this if v4l2_device was registered with NULL as
 620  the parent device struct. This only happens in cases where one hardware
 621  device has multiple PCI devices that all share the same v4l2_device core.
 622
 623  The cx88 driver is an example of this: one core v4l2_device struct, but
 624  it is used by both an raw video PCI device (cx8800) and a MPEG PCI device
 625  (cx8802). Since the v4l2_device cannot be associated with a particular
 626  PCI device it is setup without a parent device. But when the struct
 627  video_device is setup you do know which parent PCI device to use.
 628
 629- flags: optional. Set to V4L2_FL_USE_FH_PRIO if you want to let the framework
 630  handle the VIDIOC_G/S_PRIORITY ioctls. This requires that you use struct
 631  v4l2_fh. Eventually this flag will disappear once all drivers use the core
 632  priority handling. But for now it has to be set explicitly.
 633
 634If you use v4l2_ioctl_ops, then you should set .unlocked_ioctl to video_ioctl2
 635in your v4l2_file_operations struct.
 636
 637Do not use .ioctl! This is deprecated and will go away in the future.
 638
 639In some cases you want to tell the core that a function you had specified in
 640your v4l2_ioctl_ops should be ignored. You can mark such ioctls by calling this
 641function before video_device_register is called:
 642
 643void v4l2_disable_ioctl(struct video_device *vdev, unsigned int cmd);
 644
 645This tends to be needed if based on external factors (e.g. which card is
 646being used) you want to turns off certain features in v4l2_ioctl_ops without
 647having to make a new struct.
 648
 649The v4l2_file_operations struct is a subset of file_operations. The main
 650difference is that the inode argument is omitted since it is never used.
 651
 652If integration with the media framework is needed, you must initialize the
 653media_entity struct embedded in the video_device struct (entity field) by
 654calling media_entity_init():
 655
 656        struct media_pad *pad = &my_vdev->pad;
 657        int err;
 658
 659        err = media_entity_init(&vdev->entity, 1, pad, 0);
 660
 661The pads array must have been previously initialized. There is no need to
 662manually set the struct media_entity type and name fields.
 663
 664A reference to the entity will be automatically acquired/released when the
 665video device is opened/closed.
 666
 667ioctls and locking
 668------------------
 669
 670The V4L core provides optional locking services. The main service is the
 671lock field in struct video_device, which is a pointer to a mutex. If you set
 672this pointer, then that will be used by unlocked_ioctl to serialize all ioctls.
 673
 674If you are using the videobuf2 framework, then there is a second lock that you
 675can set: video_device->queue->lock. If set, then this lock will be used instead
 676of video_device->lock to serialize all queuing ioctls (see the previous section
 677for the full list of those ioctls).
 678
 679The advantage of using a different lock for the queuing ioctls is that for some
 680drivers (particularly USB drivers) certain commands such as setting controls
 681can take a long time, so you want to use a separate lock for the buffer queuing
 682ioctls. That way your VIDIOC_DQBUF doesn't stall because the driver is busy
 683changing the e.g. exposure of the webcam.
 684
 685Of course, you can always do all the locking yourself by leaving both lock
 686pointers at NULL.
 687
 688If you use the old videobuf then you must pass the video_device lock to the
 689videobuf queue initialize function: if videobuf has to wait for a frame to
 690arrive, then it will temporarily unlock the lock and relock it afterwards. If
 691your driver also waits in the code, then you should do the same to allow other
 692processes to access the device node while the first process is waiting for
 693something.
 694
 695In the case of videobuf2 you will need to implement the wait_prepare and
 696wait_finish callbacks to unlock/lock if applicable. If you use the queue->lock
 697pointer, then you can use the helper functions vb2_ops_wait_prepare/finish.
 698
 699The implementation of a hotplug disconnect should also take the lock from
 700video_device before calling v4l2_device_disconnect. If you are also using
 701video_device->queue->lock, then you have to first lock video_device->queue->lock
 702followed by video_device->lock. That way you can be sure no ioctl is running
 703when you call v4l2_device_disconnect.
 704
 705video_device registration
 706-------------------------
 707
 708Next you register the video device: this will create the character device
 709for you.
 710
 711        err = video_register_device(vdev, VFL_TYPE_GRABBER, -1);
 712        if (err) {
 713                video_device_release(vdev); /* or kfree(my_vdev); */
 714                return err;
 715        }
 716
 717If the v4l2_device parent device has a non-NULL mdev field, the video device
 718entity will be automatically registered with the media device.
 719
 720Which device is registered depends on the type argument. The following
 721types exist:
 722
 723VFL_TYPE_GRABBER: videoX for video input/output devices
 724VFL_TYPE_VBI: vbiX for vertical blank data (i.e. closed captions, teletext)
 725VFL_TYPE_RADIO: radioX for radio tuners
 726
 727The last argument gives you a certain amount of control over the device
 728device node number used (i.e. the X in videoX). Normally you will pass -1
 729to let the v4l2 framework pick the first free number. But sometimes users
 730want to select a specific node number. It is common that drivers allow
 731the user to select a specific device node number through a driver module
 732option. That number is then passed to this function and video_register_device
 733will attempt to select that device node number. If that number was already
 734in use, then the next free device node number will be selected and it
 735will send a warning to the kernel log.
 736
 737Another use-case is if a driver creates many devices. In that case it can
 738be useful to place different video devices in separate ranges. For example,
 739video capture devices start at 0, video output devices start at 16.
 740So you can use the last argument to specify a minimum device node number
 741and the v4l2 framework will try to pick the first free number that is equal
 742or higher to what you passed. If that fails, then it will just pick the
 743first free number.
 744
 745Since in this case you do not care about a warning about not being able
 746to select the specified device node number, you can call the function
 747video_register_device_no_warn() instead.
 748
 749Whenever a device node is created some attributes are also created for you.
 750If you look in /sys/class/video4linux you see the devices. Go into e.g.
 751video0 and you will see 'name' and 'index' attributes. The 'name' attribute
 752is the 'name' field of the video_device struct.
 753
 754The 'index' attribute is the index of the device node: for each call to
 755video_register_device() the index is just increased by 1. The first video
 756device node you register always starts with index 0.
 757
 758Users can setup udev rules that utilize the index attribute to make fancy
 759device names (e.g. 'mpegX' for MPEG video capture device nodes).
 760
 761After the device was successfully registered, then you can use these fields:
 762
 763- vfl_type: the device type passed to video_register_device.
 764- minor: the assigned device minor number.
 765- num: the device node number (i.e. the X in videoX).
 766- index: the device index number.
 767
 768If the registration failed, then you need to call video_device_release()
 769to free the allocated video_device struct, or free your own struct if the
 770video_device was embedded in it. The vdev->release() callback will never
 771be called if the registration failed, nor should you ever attempt to
 772unregister the device if the registration failed.
 773
 774
 775video_device cleanup
 776--------------------
 777
 778When the video device nodes have to be removed, either during the unload
 779of the driver or because the USB device was disconnected, then you should
 780unregister them:
 781
 782        video_unregister_device(vdev);
 783
 784This will remove the device nodes from sysfs (causing udev to remove them
 785from /dev).
 786
 787After video_unregister_device() returns no new opens can be done. However,
 788in the case of USB devices some application might still have one of these
 789device nodes open. So after the unregister all file operations (except
 790release, of course) will return an error as well.
 791
 792When the last user of the video device node exits, then the vdev->release()
 793callback is called and you can do the final cleanup there.
 794
 795Don't forget to cleanup the media entity associated with the video device if
 796it has been initialized:
 797
 798        media_entity_cleanup(&vdev->entity);
 799
 800This can be done from the release callback.
 801
 802
 803video_device helper functions
 804-----------------------------
 805
 806There are a few useful helper functions:
 807
 808- file/video_device private data
 809
 810You can set/get driver private data in the video_device struct using:
 811
 812void *video_get_drvdata(struct video_device *vdev);
 813void video_set_drvdata(struct video_device *vdev, void *data);
 814
 815Note that you can safely call video_set_drvdata() before calling
 816video_register_device().
 817
 818And this function:
 819
 820struct video_device *video_devdata(struct file *file);
 821
 822returns the video_device belonging to the file struct.
 823
 824The video_drvdata function combines video_get_drvdata with video_devdata:
 825
 826void *video_drvdata(struct file *file);
 827
 828You can go from a video_device struct to the v4l2_device struct using:
 829
 830struct v4l2_device *v4l2_dev = vdev->v4l2_dev;
 831
 832- Device node name
 833
 834The video_device node kernel name can be retrieved using
 835
 836const char *video_device_node_name(struct video_device *vdev);
 837
 838The name is used as a hint by userspace tools such as udev. The function
 839should be used where possible instead of accessing the video_device::num and
 840video_device::minor fields.
 841
 842
 843video buffer helper functions
 844-----------------------------
 845
 846The v4l2 core API provides a set of standard methods (called "videobuf")
 847for dealing with video buffers. Those methods allow a driver to implement
 848read(), mmap() and overlay() in a consistent way.  There are currently
 849methods for using video buffers on devices that supports DMA with
 850scatter/gather method (videobuf-dma-sg), DMA with linear access
 851(videobuf-dma-contig), and vmalloced buffers, mostly used on USB drivers
 852(videobuf-vmalloc).
 853
 854Please see Documentation/video4linux/videobuf for more information on how
 855to use the videobuf layer.
 856
 857struct v4l2_fh
 858--------------
 859
 860struct v4l2_fh provides a way to easily keep file handle specific data
 861that is used by the V4L2 framework. New drivers must use struct v4l2_fh
 862since it is also used to implement priority handling (VIDIOC_G/S_PRIORITY)
 863if the video_device flag V4L2_FL_USE_FH_PRIO is also set.
 864
 865The users of v4l2_fh (in the V4L2 framework, not the driver) know
 866whether a driver uses v4l2_fh as its file->private_data pointer by
 867testing the V4L2_FL_USES_V4L2_FH bit in video_device->flags. This bit is
 868set whenever v4l2_fh_init() is called.
 869
 870struct v4l2_fh is allocated as a part of the driver's own file handle
 871structure and file->private_data is set to it in the driver's open
 872function by the driver.
 873
 874In many cases the struct v4l2_fh will be embedded in a larger structure.
 875In that case you should call v4l2_fh_init+v4l2_fh_add in open() and
 876v4l2_fh_del+v4l2_fh_exit in release().
 877
 878Drivers can extract their own file handle structure by using the container_of
 879macro. Example:
 880
 881struct my_fh {
 882        int blah;
 883        struct v4l2_fh fh;
 884};
 885
 886...
 887
 888int my_open(struct file *file)
 889{
 890        struct my_fh *my_fh;
 891        struct video_device *vfd;
 892        int ret;
 893
 894        ...
 895
 896        my_fh = kzalloc(sizeof(*my_fh), GFP_KERNEL);
 897
 898        ...
 899
 900        v4l2_fh_init(&my_fh->fh, vfd);
 901
 902        ...
 903
 904        file->private_data = &my_fh->fh;
 905        v4l2_fh_add(&my_fh->fh);
 906        return 0;
 907}
 908
 909int my_release(struct file *file)
 910{
 911        struct v4l2_fh *fh = file->private_data;
 912        struct my_fh *my_fh = container_of(fh, struct my_fh, fh);
 913
 914        ...
 915        v4l2_fh_del(&my_fh->fh);
 916        v4l2_fh_exit(&my_fh->fh);
 917        kfree(my_fh);
 918        return 0;
 919}
 920
 921Below is a short description of the v4l2_fh functions used:
 922
 923void v4l2_fh_init(struct v4l2_fh *fh, struct video_device *vdev)
 924
 925  Initialise the file handle. This *MUST* be performed in the driver's
 926  v4l2_file_operations->open() handler.
 927
 928void v4l2_fh_add(struct v4l2_fh *fh)
 929
 930  Add a v4l2_fh to video_device file handle list. Must be called once the
 931  file handle is completely initialized.
 932
 933void v4l2_fh_del(struct v4l2_fh *fh)
 934
 935  Unassociate the file handle from video_device(). The file handle
 936  exit function may now be called.
 937
 938void v4l2_fh_exit(struct v4l2_fh *fh)
 939
 940  Uninitialise the file handle. After uninitialisation the v4l2_fh
 941  memory can be freed.
 942
 943
 944If struct v4l2_fh is not embedded, then you can use these helper functions:
 945
 946int v4l2_fh_open(struct file *filp)
 947
 948  This allocates a struct v4l2_fh, initializes it and adds it to the struct
 949  video_device associated with the file struct.
 950
 951int v4l2_fh_release(struct file *filp)
 952
 953  This deletes it from the struct video_device associated with the file
 954  struct, uninitialised the v4l2_fh and frees it.
 955
 956These two functions can be plugged into the v4l2_file_operation's open() and
 957release() ops.
 958
 959
 960Several drivers need to do something when the first file handle is opened and
 961when the last file handle closes. Two helper functions were added to check
 962whether the v4l2_fh struct is the only open filehandle of the associated
 963device node:
 964
 965int v4l2_fh_is_singular(struct v4l2_fh *fh)
 966
 967  Returns 1 if the file handle is the only open file handle, else 0.
 968
 969int v4l2_fh_is_singular_file(struct file *filp)
 970
 971  Same, but it calls v4l2_fh_is_singular with filp->private_data.
 972
 973
 974V4L2 events
 975-----------
 976
 977The V4L2 events provide a generic way to pass events to user space.
 978The driver must use v4l2_fh to be able to support V4L2 events.
 979
 980Events are defined by a type and an optional ID. The ID may refer to a V4L2
 981object such as a control ID. If unused, then the ID is 0.
 982
 983When the user subscribes to an event the driver will allocate a number of
 984kevent structs for that event. So every (type, ID) event tuple will have
 985its own set of kevent structs. This guarantees that if a driver is generating
 986lots of events of one type in a short time, then that will not overwrite
 987events of another type.
 988
 989But if you get more events of one type than the number of kevents that were
 990reserved, then the oldest event will be dropped and the new one added.
 991
 992Furthermore, the internal struct v4l2_subscribed_event has merge() and
 993replace() callbacks which drivers can set. These callbacks are called when
 994a new event is raised and there is no more room. The replace() callback
 995allows you to replace the payload of the old event with that of the new event,
 996merging any relevant data from the old payload into the new payload that
 997replaces it. It is called when this event type has only one kevent struct
 998allocated. The merge() callback allows you to merge the oldest event payload
 999into that of the second-oldest event payload. It is called when there are two
1000or more kevent structs allocated.
1001
1002This way no status information is lost, just the intermediate steps leading
1003up to that state.
1004
1005A good example of these replace/merge callbacks is in v4l2-event.c:
1006ctrls_replace() and ctrls_merge() callbacks for the control event.
1007
1008Note: these callbacks can be called from interrupt context, so they must be
1009fast.
1010
1011Useful functions:
1012
1013void v4l2_event_queue(struct video_device *vdev, const struct v4l2_event *ev)
1014
1015  Queue events to video device. The driver's only responsibility is to fill
1016  in the type and the data fields. The other fields will be filled in by
1017  V4L2.
1018
1019int v4l2_event_subscribe(struct v4l2_fh *fh,
1020                         struct v4l2_event_subscription *sub, unsigned elems,
1021                         const struct v4l2_subscribed_event_ops *ops)
1022
1023  The video_device->ioctl_ops->vidioc_subscribe_event must check the driver
1024  is able to produce events with specified event id. Then it calls
1025  v4l2_event_subscribe() to subscribe the event.
1026
1027  The elems argument is the size of the event queue for this event. If it is 0,
1028  then the framework will fill in a default value (this depends on the event
1029  type).
1030
1031  The ops argument allows the driver to specify a number of callbacks:
1032  * add:     called when a new listener gets added (subscribing to the same
1033             event twice will only cause this callback to get called once)
1034  * del:     called when a listener stops listening
1035  * replace: replace event 'old' with event 'new'.
1036  * merge:   merge event 'old' into event 'new'.
1037  All 4 callbacks are optional, if you don't want to specify any callbacks
1038  the ops argument itself maybe NULL.
1039
1040int v4l2_event_unsubscribe(struct v4l2_fh *fh,
1041                           struct v4l2_event_subscription *sub)
1042
1043  vidioc_unsubscribe_event in struct v4l2_ioctl_ops. A driver may use
1044  v4l2_event_unsubscribe() directly unless it wants to be involved in
1045  unsubscription process.
1046
1047  The special type V4L2_EVENT_ALL may be used to unsubscribe all events. The
1048  drivers may want to handle this in a special way.
1049
1050int v4l2_event_pending(struct v4l2_fh *fh)
1051
1052  Returns the number of pending events. Useful when implementing poll.
1053
1054Events are delivered to user space through the poll system call. The driver
1055can use v4l2_fh->wait (a wait_queue_head_t) as the argument for poll_wait().
1056
1057There are standard and private events. New standard events must use the
1058smallest available event type. The drivers must allocate their events from
1059their own class starting from class base. Class base is
1060V4L2_EVENT_PRIVATE_START + n * 1000 where n is the lowest available number.
1061The first event type in the class is reserved for future use, so the first
1062available event type is 'class base + 1'.
1063
1064An example on how the V4L2 events may be used can be found in the OMAP
10653 ISP driver (drivers/media/platform/omap3isp).
1066
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.