blob: 881e7f44491b70a6c803fa68f6185cb4f158b87f [file] [log] [blame]
Hans Verkuila42b57f2010-08-01 14:35:53 -03001Introduction
2============
3
4The V4L2 control API seems simple enough, but quickly becomes very hard to
5implement correctly in drivers. But much of the code needed to handle controls
6is actually not driver specific and can be moved to the V4L core framework.
7
8After all, the only part that a driver developer is interested in is:
9
101) How do I add a control?
112) How do I set the control's value? (i.e. s_ctrl)
12
13And occasionally:
14
153) How do I get the control's value? (i.e. g_volatile_ctrl)
164) How do I validate the user's proposed control value? (i.e. try_ctrl)
17
18All the rest is something that can be done centrally.
19
20The control framework was created in order to implement all the rules of the
21V4L2 specification with respect to controls in a central place. And to make
22life as easy as possible for the driver developer.
23
24Note that the control framework relies on the presence of a struct v4l2_device
25for V4L2 drivers and struct v4l2_subdev for sub-device drivers.
26
27
28Objects in the framework
29========================
30
31There are two main objects:
32
33The v4l2_ctrl object describes the control properties and keeps track of the
34control's value (both the current value and the proposed new value).
35
36v4l2_ctrl_handler is the object that keeps track of controls. It maintains a
37list of v4l2_ctrl objects that it owns and another list of references to
38controls, possibly to controls owned by other handlers.
39
40
41Basic usage for V4L2 and sub-device drivers
42===========================================
43
441) Prepare the driver:
45
461.1) Add the handler to your driver's top-level struct:
47
48 struct foo_dev {
49 ...
50 struct v4l2_ctrl_handler ctrl_handler;
51 ...
52 };
53
54 struct foo_dev *foo;
55
561.2) Initialize the handler:
57
58 v4l2_ctrl_handler_init(&foo->ctrl_handler, nr_of_controls);
59
60 The second argument is a hint telling the function how many controls this
61 handler is expected to handle. It will allocate a hashtable based on this
62 information. It is a hint only.
63
641.3) Hook the control handler into the driver:
65
661.3.1) For V4L2 drivers do this:
67
68 struct foo_dev {
69 ...
70 struct v4l2_device v4l2_dev;
71 ...
72 struct v4l2_ctrl_handler ctrl_handler;
73 ...
74 };
75
76 foo->v4l2_dev.ctrl_handler = &foo->ctrl_handler;
77
78 Where foo->v4l2_dev is of type struct v4l2_device.
79
80 Finally, remove all control functions from your v4l2_ioctl_ops:
81 vidioc_queryctrl, vidioc_querymenu, vidioc_g_ctrl, vidioc_s_ctrl,
82 vidioc_g_ext_ctrls, vidioc_try_ext_ctrls and vidioc_s_ext_ctrls.
83 Those are now no longer needed.
84
851.3.2) For sub-device drivers do this:
86
87 struct foo_dev {
88 ...
89 struct v4l2_subdev sd;
90 ...
91 struct v4l2_ctrl_handler ctrl_handler;
92 ...
93 };
94
95 foo->sd.ctrl_handler = &foo->ctrl_handler;
96
97 Where foo->sd is of type struct v4l2_subdev.
98
99 And set all core control ops in your struct v4l2_subdev_core_ops to these
100 helpers:
101
102 .queryctrl = v4l2_subdev_queryctrl,
103 .querymenu = v4l2_subdev_querymenu,
104 .g_ctrl = v4l2_subdev_g_ctrl,
105 .s_ctrl = v4l2_subdev_s_ctrl,
106 .g_ext_ctrls = v4l2_subdev_g_ext_ctrls,
107 .try_ext_ctrls = v4l2_subdev_try_ext_ctrls,
108 .s_ext_ctrls = v4l2_subdev_s_ext_ctrls,
109
110 Note: this is a temporary solution only. Once all V4L2 drivers that depend
111 on subdev drivers are converted to the control framework these helpers will
112 no longer be needed.
113
1141.4) Clean up the handler at the end:
115
116 v4l2_ctrl_handler_free(&foo->ctrl_handler);
117
118
1192) Add controls:
120
121You add non-menu controls by calling v4l2_ctrl_new_std:
122
123 struct v4l2_ctrl *v4l2_ctrl_new_std(struct v4l2_ctrl_handler *hdl,
124 const struct v4l2_ctrl_ops *ops,
125 u32 id, s32 min, s32 max, u32 step, s32 def);
126
127Menu controls are added by calling v4l2_ctrl_new_std_menu:
128
129 struct v4l2_ctrl *v4l2_ctrl_new_std_menu(struct v4l2_ctrl_handler *hdl,
130 const struct v4l2_ctrl_ops *ops,
131 u32 id, s32 max, s32 skip_mask, s32 def);
132
133These functions are typically called right after the v4l2_ctrl_handler_init:
134
135 v4l2_ctrl_handler_init(&foo->ctrl_handler, nr_of_controls);
136 v4l2_ctrl_new_std(&foo->ctrl_handler, &foo_ctrl_ops,
137 V4L2_CID_BRIGHTNESS, 0, 255, 1, 128);
138 v4l2_ctrl_new_std(&foo->ctrl_handler, &foo_ctrl_ops,
139 V4L2_CID_CONTRAST, 0, 255, 1, 128);
140 v4l2_ctrl_new_std_menu(&foo->ctrl_handler, &foo_ctrl_ops,
141 V4L2_CID_POWER_LINE_FREQUENCY,
142 V4L2_CID_POWER_LINE_FREQUENCY_60HZ, 0,
143 V4L2_CID_POWER_LINE_FREQUENCY_DISABLED);
144 ...
145 if (foo->ctrl_handler.error) {
146 int err = foo->ctrl_handler.error;
147
148 v4l2_ctrl_handler_free(&foo->ctrl_handler);
149 return err;
150 }
151
152The v4l2_ctrl_new_std function returns the v4l2_ctrl pointer to the new
153control, but if you do not need to access the pointer outside the control ops,
154then there is no need to store it.
155
156The v4l2_ctrl_new_std function will fill in most fields based on the control
157ID except for the min, max, step and default values. These are passed in the
158last four arguments. These values are driver specific while control attributes
159like type, name, flags are all global. The control's current value will be set
160to the default value.
161
162The v4l2_ctrl_new_std_menu function is very similar but it is used for menu
163controls. There is no min argument since that is always 0 for menu controls,
164and instead of a step there is a skip_mask argument: if bit X is 1, then menu
165item X is skipped.
166
167Note that if something fails, the function will return NULL or an error and
168set ctrl_handler->error to the error code. If ctrl_handler->error was already
169set, then it will just return and do nothing. This is also true for
170v4l2_ctrl_handler_init if it cannot allocate the internal data structure.
171
172This makes it easy to init the handler and just add all controls and only check
173the error code at the end. Saves a lot of repetitive error checking.
174
175It is recommended to add controls in ascending control ID order: it will be
176a bit faster that way.
177
1783) Optionally force initial control setup:
179
180 v4l2_ctrl_handler_setup(&foo->ctrl_handler);
181
182This will call s_ctrl for all controls unconditionally. Effectively this
183initializes the hardware to the default control values. It is recommended
184that you do this as this ensures that both the internal data structures and
185the hardware are in sync.
186
1874) Finally: implement the v4l2_ctrl_ops
188
189 static const struct v4l2_ctrl_ops foo_ctrl_ops = {
190 .s_ctrl = foo_s_ctrl,
191 };
192
193Usually all you need is s_ctrl:
194
195 static int foo_s_ctrl(struct v4l2_ctrl *ctrl)
196 {
197 struct foo *state = container_of(ctrl->handler, struct foo, ctrl_handler);
198
199 switch (ctrl->id) {
200 case V4L2_CID_BRIGHTNESS:
201 write_reg(0x123, ctrl->val);
202 break;
203 case V4L2_CID_CONTRAST:
204 write_reg(0x456, ctrl->val);
205 break;
206 }
207 return 0;
208 }
209
210The control ops are called with the v4l2_ctrl pointer as argument.
211The new control value has already been validated, so all you need to do is
212to actually update the hardware registers.
213
214You're done! And this is sufficient for most of the drivers we have. No need
215to do any validation of control values, or implement QUERYCTRL/QUERYMENU. And
216G/S_CTRL as well as G/TRY/S_EXT_CTRLS are automatically supported.
217
218
219==============================================================================
220
221The remainder of this document deals with more advanced topics and scenarios.
222In practice the basic usage as described above is sufficient for most drivers.
223
224===============================================================================
225
226
227Inheriting Controls
228===================
229
230When a sub-device is registered with a V4L2 driver by calling
231v4l2_device_register_subdev() and the ctrl_handler fields of both v4l2_subdev
232and v4l2_device are set, then the controls of the subdev will become
233automatically available in the V4L2 driver as well. If the subdev driver
234contains controls that already exist in the V4L2 driver, then those will be
235skipped (so a V4L2 driver can always override a subdev control).
236
237What happens here is that v4l2_device_register_subdev() calls
238v4l2_ctrl_add_handler() adding the controls of the subdev to the controls
239of v4l2_device.
240
241
242Accessing Control Values
243========================
244
245The v4l2_ctrl struct contains these two unions:
246
247 /* The current control value. */
248 union {
249 s32 val;
250 s64 val64;
251 char *string;
252 } cur;
253
254 /* The new control value. */
255 union {
256 s32 val;
257 s64 val64;
258 char *string;
259 };
260
261Within the control ops you can freely use these. The val and val64 speak for
262themselves. The string pointers point to character buffers of length
263ctrl->maximum + 1, and are always 0-terminated.
264
265In most cases 'cur' contains the current cached control value. When you create
266a new control this value is made identical to the default value. After calling
267v4l2_ctrl_handler_setup() this value is passed to the hardware. It is generally
268a good idea to call this function.
269
270Whenever a new value is set that new value is automatically cached. This means
271that most drivers do not need to implement the g_volatile_ctrl() op. The
272exception is for controls that return a volatile register such as a signal
273strength read-out that changes continuously. In that case you will need to
274implement g_volatile_ctrl like this:
275
276 static int foo_g_volatile_ctrl(struct v4l2_ctrl *ctrl)
277 {
278 switch (ctrl->id) {
279 case V4L2_CID_BRIGHTNESS:
280 ctrl->cur.val = read_reg(0x123);
281 break;
282 }
283 }
284
285The 'new value' union is not used in g_volatile_ctrl. In general controls
286that need to implement g_volatile_ctrl are read-only controls.
287
Hans Verkuil2a863792011-01-11 14:45:03 -0300288Note that if one or more controls in a control cluster are marked as volatile,
289then all the controls in the cluster are seen as volatile.
290
Hans Verkuila42b57f2010-08-01 14:35:53 -0300291To mark a control as volatile you have to set the is_volatile flag:
292
293 ctrl = v4l2_ctrl_new_std(&sd->ctrl_handler, ...);
294 if (ctrl)
295 ctrl->is_volatile = 1;
296
297For try/s_ctrl the new values (i.e. as passed by the user) are filled in and
298you can modify them in try_ctrl or set them in s_ctrl. The 'cur' union
299contains the current value, which you can use (but not change!) as well.
300
301If s_ctrl returns 0 (OK), then the control framework will copy the new final
302values to the 'cur' union.
303
304While in g_volatile/s/try_ctrl you can access the value of all controls owned
305by the same handler since the handler's lock is held. If you need to access
306the value of controls owned by other handlers, then you have to be very careful
307not to introduce deadlocks.
308
309Outside of the control ops you have to go through to helper functions to get
310or set a single control value safely in your driver:
311
312 s32 v4l2_ctrl_g_ctrl(struct v4l2_ctrl *ctrl);
313 int v4l2_ctrl_s_ctrl(struct v4l2_ctrl *ctrl, s32 val);
314
315These functions go through the control framework just as VIDIOC_G/S_CTRL ioctls
316do. Don't use these inside the control ops g_volatile/s/try_ctrl, though, that
317will result in a deadlock since these helpers lock the handler as well.
318
319You can also take the handler lock yourself:
320
321 mutex_lock(&state->ctrl_handler.lock);
322 printk(KERN_INFO "String value is '%s'\n", ctrl1->cur.string);
323 printk(KERN_INFO "Integer value is '%s'\n", ctrl2->cur.val);
324 mutex_unlock(&state->ctrl_handler.lock);
325
326
327Menu Controls
328=============
329
330The v4l2_ctrl struct contains this union:
331
332 union {
333 u32 step;
334 u32 menu_skip_mask;
335 };
336
337For menu controls menu_skip_mask is used. What it does is that it allows you
338to easily exclude certain menu items. This is used in the VIDIOC_QUERYMENU
339implementation where you can return -EINVAL if a certain menu item is not
340present. Note that VIDIOC_QUERYCTRL always returns a step value of 1 for
341menu controls.
342
343A good example is the MPEG Audio Layer II Bitrate menu control where the
344menu is a list of standardized possible bitrates. But in practice hardware
345implementations will only support a subset of those. By setting the skip
346mask you can tell the framework which menu items should be skipped. Setting
347it to 0 means that all menu items are supported.
348
349You set this mask either through the v4l2_ctrl_config struct for a custom
350control, or by calling v4l2_ctrl_new_std_menu().
351
352
353Custom Controls
354===============
355
356Driver specific controls can be created using v4l2_ctrl_new_custom():
357
358 static const struct v4l2_ctrl_config ctrl_filter = {
359 .ops = &ctrl_custom_ops,
360 .id = V4L2_CID_MPEG_CX2341X_VIDEO_SPATIAL_FILTER,
361 .name = "Spatial Filter",
362 .type = V4L2_CTRL_TYPE_INTEGER,
363 .flags = V4L2_CTRL_FLAG_SLIDER,
364 .max = 15,
365 .step = 1,
366 };
367
368 ctrl = v4l2_ctrl_new_custom(&foo->ctrl_handler, &ctrl_filter, NULL);
369
370The last argument is the priv pointer which can be set to driver-specific
371private data.
372
373The v4l2_ctrl_config struct also has fields to set the is_private and is_volatile
374flags.
375
376If the name field is not set, then the framework will assume this is a standard
377control and will fill in the name, type and flags fields accordingly.
378
379
380Active and Grabbed Controls
381===========================
382
383If you get more complex relationships between controls, then you may have to
384activate and deactivate controls. For example, if the Chroma AGC control is
385on, then the Chroma Gain control is inactive. That is, you may set it, but
386the value will not be used by the hardware as long as the automatic gain
387control is on. Typically user interfaces can disable such input fields.
388
389You can set the 'active' status using v4l2_ctrl_activate(). By default all
390controls are active. Note that the framework does not check for this flag.
391It is meant purely for GUIs. The function is typically called from within
392s_ctrl.
393
394The other flag is the 'grabbed' flag. A grabbed control means that you cannot
395change it because it is in use by some resource. Typical examples are MPEG
396bitrate controls that cannot be changed while capturing is in progress.
397
398If a control is set to 'grabbed' using v4l2_ctrl_grab(), then the framework
399will return -EBUSY if an attempt is made to set this control. The
400v4l2_ctrl_grab() function is typically called from the driver when it
401starts or stops streaming.
402
403
404Control Clusters
405================
406
407By default all controls are independent from the others. But in more
408complex scenarios you can get dependencies from one control to another.
409In that case you need to 'cluster' them:
410
411 struct foo {
412 struct v4l2_ctrl_handler ctrl_handler;
413#define AUDIO_CL_VOLUME (0)
414#define AUDIO_CL_MUTE (1)
415 struct v4l2_ctrl *audio_cluster[2];
416 ...
417 };
418
419 state->audio_cluster[AUDIO_CL_VOLUME] =
420 v4l2_ctrl_new_std(&state->ctrl_handler, ...);
421 state->audio_cluster[AUDIO_CL_MUTE] =
422 v4l2_ctrl_new_std(&state->ctrl_handler, ...);
423 v4l2_ctrl_cluster(ARRAY_SIZE(state->audio_cluster), state->audio_cluster);
424
425From now on whenever one or more of the controls belonging to the same
426cluster is set (or 'gotten', or 'tried'), only the control ops of the first
427control ('volume' in this example) is called. You effectively create a new
428composite control. Similar to how a 'struct' works in C.
429
430So when s_ctrl is called with V4L2_CID_AUDIO_VOLUME as argument, you should set
431all two controls belonging to the audio_cluster:
432
433 static int foo_s_ctrl(struct v4l2_ctrl *ctrl)
434 {
435 struct foo *state = container_of(ctrl->handler, struct foo, ctrl_handler);
436
437 switch (ctrl->id) {
438 case V4L2_CID_AUDIO_VOLUME: {
439 struct v4l2_ctrl *mute = ctrl->cluster[AUDIO_CL_MUTE];
440
441 write_reg(0x123, mute->val ? 0 : ctrl->val);
442 break;
443 }
444 case V4L2_CID_CONTRAST:
445 write_reg(0x456, ctrl->val);
446 break;
447 }
448 return 0;
449 }
450
451In the example above the following are equivalent for the VOLUME case:
452
453 ctrl == ctrl->cluster[AUDIO_CL_VOLUME] == state->audio_cluster[AUDIO_CL_VOLUME]
454 ctrl->cluster[AUDIO_CL_MUTE] == state->audio_cluster[AUDIO_CL_MUTE]
455
456Note that controls in a cluster may be NULL. For example, if for some
457reason mute was never added (because the hardware doesn't support that
458particular feature), then mute will be NULL. So in that case we have a
459cluster of 2 controls, of which only 1 is actually instantiated. The
460only restriction is that the first control of the cluster must always be
461present, since that is the 'master' control of the cluster. The master
462control is the one that identifies the cluster and that provides the
463pointer to the v4l2_ctrl_ops struct that is used for that cluster.
464
465Obviously, all controls in the cluster array must be initialized to either
466a valid control or to NULL.
467
Hans Verkuil2a863792011-01-11 14:45:03 -0300468In rare cases you might want to know which controls of a cluster actually
469were set explicitly by the user. For this you can check the 'is_new' flag of
470each control. For example, in the case of a volume/mute cluster the 'is_new'
471flag of the mute control would be set if the user called VIDIOC_S_CTRL for
472mute only. If the user would call VIDIOC_S_EXT_CTRLS for both mute and volume
473controls, then the 'is_new' flag would be 1 for both controls.
474
475The 'is_new' flag is always 1 when called from v4l2_ctrl_handler_setup().
476
Hans Verkuila42b57f2010-08-01 14:35:53 -0300477
478VIDIOC_LOG_STATUS Support
479=========================
480
481This ioctl allow you to dump the current status of a driver to the kernel log.
482The v4l2_ctrl_handler_log_status(ctrl_handler, prefix) can be used to dump the
483value of the controls owned by the given handler to the log. You can supply a
484prefix as well. If the prefix didn't end with a space, then ': ' will be added
485for you.
486
487
488Different Handlers for Different Video Nodes
489============================================
490
491Usually the V4L2 driver has just one control handler that is global for
492all video nodes. But you can also specify different control handlers for
493different video nodes. You can do that by manually setting the ctrl_handler
494field of struct video_device.
495
496That is no problem if there are no subdevs involved but if there are, then
497you need to block the automatic merging of subdev controls to the global
498control handler. You do that by simply setting the ctrl_handler field in
499struct v4l2_device to NULL. Now v4l2_device_register_subdev() will no longer
500merge subdev controls.
501
502After each subdev was added, you will then have to call v4l2_ctrl_add_handler
503manually to add the subdev's control handler (sd->ctrl_handler) to the desired
504control handler. This control handler may be specific to the video_device or
505for a subset of video_device's. For example: the radio device nodes only have
506audio controls, while the video and vbi device nodes share the same control
507handler for the audio and video controls.
508
509If you want to have one handler (e.g. for a radio device node) have a subset
510of another handler (e.g. for a video device node), then you should first add
511the controls to the first handler, add the other controls to the second
512handler and finally add the first handler to the second. For example:
513
514 v4l2_ctrl_new_std(&radio_ctrl_handler, &radio_ops, V4L2_CID_AUDIO_VOLUME, ...);
515 v4l2_ctrl_new_std(&radio_ctrl_handler, &radio_ops, V4L2_CID_AUDIO_MUTE, ...);
516 v4l2_ctrl_new_std(&video_ctrl_handler, &video_ops, V4L2_CID_BRIGHTNESS, ...);
517 v4l2_ctrl_new_std(&video_ctrl_handler, &video_ops, V4L2_CID_CONTRAST, ...);
518 v4l2_ctrl_add_handler(&video_ctrl_handler, &radio_ctrl_handler);
519
520Or you can add specific controls to a handler:
521
522 volume = v4l2_ctrl_new_std(&video_ctrl_handler, &ops, V4L2_CID_AUDIO_VOLUME, ...);
523 v4l2_ctrl_new_std(&video_ctrl_handler, &ops, V4L2_CID_BRIGHTNESS, ...);
524 v4l2_ctrl_new_std(&video_ctrl_handler, &ops, V4L2_CID_CONTRAST, ...);
525 v4l2_ctrl_add_ctrl(&radio_ctrl_handler, volume);
526
527What you should not do is make two identical controls for two handlers.
528For example:
529
530 v4l2_ctrl_new_std(&radio_ctrl_handler, &radio_ops, V4L2_CID_AUDIO_MUTE, ...);
531 v4l2_ctrl_new_std(&video_ctrl_handler, &video_ops, V4L2_CID_AUDIO_MUTE, ...);
532
533This would be bad since muting the radio would not change the video mute
534control. The rule is to have one control for each hardware 'knob' that you
535can twiddle.
536
537
538Finding Controls
539================
540
541Normally you have created the controls yourself and you can store the struct
542v4l2_ctrl pointer into your own struct.
543
544But sometimes you need to find a control from another handler that you do
545not own. For example, if you have to find a volume control from a subdev.
546
547You can do that by calling v4l2_ctrl_find:
548
549 struct v4l2_ctrl *volume;
550
551 volume = v4l2_ctrl_find(sd->ctrl_handler, V4L2_CID_AUDIO_VOLUME);
552
553Since v4l2_ctrl_find will lock the handler you have to be careful where you
554use it. For example, this is not a good idea:
555
556 struct v4l2_ctrl_handler ctrl_handler;
557
558 v4l2_ctrl_new_std(&ctrl_handler, &video_ops, V4L2_CID_BRIGHTNESS, ...);
559 v4l2_ctrl_new_std(&ctrl_handler, &video_ops, V4L2_CID_CONTRAST, ...);
560
561...and in video_ops.s_ctrl:
562
563 case V4L2_CID_BRIGHTNESS:
564 contrast = v4l2_find_ctrl(&ctrl_handler, V4L2_CID_CONTRAST);
565 ...
566
567When s_ctrl is called by the framework the ctrl_handler.lock is already taken, so
568attempting to find another control from the same handler will deadlock.
569
570It is recommended not to use this function from inside the control ops.
571
572
573Inheriting Controls
574===================
575
576When one control handler is added to another using v4l2_ctrl_add_handler, then
577by default all controls from one are merged to the other. But a subdev might
578have low-level controls that make sense for some advanced embedded system, but
579not when it is used in consumer-level hardware. In that case you want to keep
580those low-level controls local to the subdev. You can do this by simply
581setting the 'is_private' flag of the control to 1:
582
583 static const struct v4l2_ctrl_config ctrl_private = {
584 .ops = &ctrl_custom_ops,
585 .id = V4L2_CID_...,
586 .name = "Some Private Control",
587 .type = V4L2_CTRL_TYPE_INTEGER,
588 .max = 15,
589 .step = 1,
590 .is_private = 1,
591 };
592
593 ctrl = v4l2_ctrl_new_custom(&foo->ctrl_handler, &ctrl_private, NULL);
594
595These controls will now be skipped when v4l2_ctrl_add_handler is called.
596
597
598V4L2_CTRL_TYPE_CTRL_CLASS Controls
599==================================
600
601Controls of this type can be used by GUIs to get the name of the control class.
602A fully featured GUI can make a dialog with multiple tabs with each tab
603containing the controls belonging to a particular control class. The name of
604each tab can be found by querying a special control with ID <control class | 1>.
605
606Drivers do not have to care about this. The framework will automatically add
607a control of this type whenever the first control belonging to a new control
608class is added.
609
610
611Differences from the Spec
612=========================
613
614There are a few places where the framework acts slightly differently from the
615V4L2 Specification. Those differences are described in this section. We will
616have to see whether we need to adjust the spec or not.
617
6181) It is no longer required to have all controls contained in a
619v4l2_ext_control array be from the same control class. The framework will be
620able to handle any type of control in the array. You need to set ctrl_class
621to 0 in order to enable this. If ctrl_class is non-zero, then it will still
622check that all controls belong to that control class.
623
624If you set ctrl_class to 0 and count to 0, then it will only return an error
625if there are no controls at all.
626
6272) Clarified the way error_idx works. For get and set it will be equal to
628count if nothing was done yet. If it is less than count then only the controls
629up to error_idx-1 were successfully applied.
630
6313) When attempting to read a button control the framework will return -EACCES
632instead of -EINVAL as stated in the spec. It seems to make more sense since
633button controls are write-only controls.
634
6354) Attempting to write to a read-only control will return -EACCES instead of
636-EINVAL as the spec says.
637
6385) The spec does not mention what should happen when you try to set/get a
639control class controls. ivtv currently returns -EINVAL (indicating that the
640control ID does not exist) while the framework will return -EACCES, which
641makes more sense.
642
643
644Proposals for Extensions
645========================
646
647Some ideas for future extensions to the spec:
648
6491) Add a V4L2_CTRL_FLAG_HEX to have values shown as hexadecimal instead of
650decimal. Useful for e.g. video_mute_yuv.
651
6522) It is possible to mark in the controls array which controls have been
653successfully written and which failed by for example adding a bit to the
654control ID. Not sure if it is worth the effort, though.
655
6563) Trying to set volatile inactive controls should result in -EACCESS.
657
6584) Add a new flag to mark volatile controls. Any application that wants
659to store the state of the controls can then skip volatile inactive controls.
660Currently it is not possible to detect such controls.