1 <!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook V3.1//EN"[]>
5 <title>Video4Linux Programming</title>
9 <firstname>Alan</firstname>
10 <surname>Cox</surname>
13 <email>alan@redhat.com</email>
21 <holder>Alan Cox</holder>
26 This documentation is free software; you can redistribute
27 it and/or modify it under the terms of the GNU General Public
28 License as published by the Free Software Foundation; either
29 version 2 of the License, or (at your option) any later
34 This program is distributed in the hope that it will be
35 useful, but WITHOUT ANY WARRANTY; without even the implied
36 warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
37 See the GNU General Public License for more details.
41 You should have received a copy of the GNU General Public
42 License along with this program; if not, write to the Free
43 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston,
48 For more details see the file COPYING in the source
49 distribution of Linux.
57 <title>Introduction</title>
59 Parts of this document first appeared in Linux Magazine under a
60 ninety day exclusivity.
63 Video4Linux is intended to provide a common programming interface
64 for the many TV and capture cards now on the market, as well as
65 parallel port and USB video cameras. Radio, teletext decoders and
66 vertical blanking data interfaces are also provided.
70 <title>Radio Devices</title>
72 There are a wide variety of radio interfaces available for PC's, and these
73 are generally very simple to program. The biggest problem with supporting
74 such devices is normally extracting documentation from the vendor.
77 The radio interface supports a simple set of control ioctls standardised
78 across all radio and tv interfaces. It does not support read or write, which
79 are used for video streams. The reason radio cards do not allow you to read
80 the audio stream into an application is that without exception they provide
81 a connection on to a soundcard. Soundcards can be used to read the radio
84 <sect1 id="registerradio">
85 <title>Registering Radio Devices</title>
87 The Video4linux core provides an interface for registering devices. The
88 first step in writing our radio card driver is to register it.
93 static struct video_device my_radio
104 NULL, /* no special init function */
105 NULL /* no private data */
111 This declares our video4linux device driver interface. The VID_TYPE_ value
112 defines what kind of an interface we are, and defines basic capabilities.
115 The only defined value relevant for a radio card is VID_TYPE_TUNER which
116 indicates that the device can be tuned. Clearly our radio is going to have some
117 way to change channel so it is tuneable.
120 The VID_HARDWARE_ types are unique to each device. Numbers are assigned by
121 <email>alan@redhat.com</email> when device drivers are going to be released. Until then you
122 can pull a suitably large number out of your hat and use it. 10000 should be
123 safe for a very long time even allowing for the huge number of vendors
124 making new and different radio cards at the moment.
127 We declare an open and close routine, but we do not need read or write,
128 which are used to read and write video data to or from the card itself. As
129 we have no read or write there is no poll function.
132 The private initialise function is run when the device is registered. In
133 this driver we've already done all the work needed. The final pointer is a
134 private data pointer that can be used by the device driver to attach and
135 retrieve private data structures. We set this field "priv" to NULL for
139 Having the structure defined is all very well but we now need to register it
145 static int io = 0x320;
147 int __init myradio_init(struct video_init *v)
149 if(!request_region(io, MY_IO_SIZE, "myradio"))
152 "myradio: port 0x%03X is in use.\n", io);
156 if(video_device_register(&my_radio, VFL_TYPE_RADIO)==-1) {
157 release_region(io, MY_IO_SIZE);
165 The first stage of the initialisation, as is normally the case, is to check
166 that the I/O space we are about to fiddle with doesn't belong to some other
167 driver. If it is we leave well alone. If the user gives the address of the
168 wrong device then we will spot this. These policies will generally avoid
169 crashing the machine.
172 Now we ask the Video4Linux layer to register the device for us. We hand it
173 our carefully designed video_device structure and also tell it which group
174 of devices we want it registered with. In this case VFL_TYPE_RADIO.
177 The types available are
179 <table frame="all"><title>Device Types</title>
180 <tgroup cols="3" align="left">
183 <entry>VFL_TYPE_RADIO</><entry>/dev/radio{n}</><entry>
185 Radio devices are assigned in this block. As with all of these
186 selections the actual number assignment is done by the video layer
187 accordijng to what is free.</entry>
189 <entry>VFL_TYPE_GRABBER</><entry>/dev/video{n}</><entry>
190 Video capture devices and also -- counter-intuitively for the name --
191 hardware video playback devices such as MPEG2 cards.</entry>
193 <entry>VFL_TYPE_VBI</><entry>/dev/vbi{n}</><entry>
194 The VBI devices capture the hidden lines on a television picture
195 that carry further information like closed caption data, teletext
196 (primarily in Europe) and now Intercast and the ATVEC internet
197 television encodings.</entry>
199 <entry>VFL_TYPE_VTX</><entry>/dev/vtx[n}</><entry>
200 VTX is 'Videotext' also known as 'Teletext'. This is a system for
201 sending numbered, 40x25, mostly textual page images over the hidden
202 lines. Unlike the /dev/vbi interfaces, this is for 'smart' decoder
203 chips. (The use of the word smart here has to be taken in context,
204 the smartest teletext chips are fairly dumb pieces of technology).
211 We are most definitely a radio.
214 Finally we allocate our I/O space so that nobody treads on us and return 0
215 to signify general happiness with the state of the universe.
218 <sect1 id="openradio">
219 <title>Opening And Closing The Radio</title>
222 The functions we declared in our video_device are mostly very simple.
223 Firstly we can drop in what is basically standard code for open and close.
228 static int users = 0;
230 static int radio_open(stuct video_device *dev, int flags)
240 At open time we need to do nothing but check if someone else is also using
241 the radio card. If nobody is using it we make a note that we are using it,
242 then we ensure that nobody unloads our driver on us.
247 static int radio_close(struct video_device *dev)
254 At close time we simply need to reduce the user count and allow the module
255 to become unloadable.
258 If you are sharp you will have noticed neither the open nor the close
259 routines attempt to reset or change the radio settings. This is intentional.
260 It allows an application to set up the radio and exit. It avoids a user
261 having to leave an application running all the time just to listen to the
265 <sect1 id="ioctlradio">
266 <title>The Ioctl Interface</title>
268 This leaves the ioctl routine, without which the driver will not be
269 terribly useful to anyone.
274 static int radio_ioctl(struct video_device *dev, unsigned int cmd, void *arg)
280 struct video_capability v;
281 v.type = VID_TYPE_TUNER;
288 strcpy(v.name, "My Radio");
289 if(copy_to_user(arg, &v, sizeof(v)))
296 VIDIOCGCAP is the first ioctl all video4linux devices must support. It
297 allows the applications to find out what sort of a card they have found and
298 to figure out what they want to do about it. The fields in the structure are
300 <table frame="all"><title>struct video_capability fields</title>
301 <tgroup cols="2" align="left">
304 <entry>name</><entry>The device text name. This is intended for the user.</>
306 <entry>channels</><entry>The number of different channels you can tune on
307 this card. It could even by zero for a card that has
308 no tuning capability. For our simple FM radio it is 1.
309 An AM/FM radio would report 2.</entry>
311 <entry>audios</><entry>The number of audio inputs on this device. For our
312 radio there is only one audio input.</entry>
314 <entry>minwidth,minheight</><entry>The smallest size the card is capable of capturing
315 images in. We set these to zero. Radios do not
316 capture pictures</entry>
318 <entry>maxwidth,maxheight</><entry>The largest image size the card is capable of
319 capturing. For our radio we report 0.
322 <entry>type</><entry>This reports the capabilities of the device, and
323 matches the field we filled in in the struct
324 video_device when registering.</entry>
330 Having filled in the fields, we use copy_to_user to copy the structure into
331 the users buffer. If the copy fails we return an EFAULT to the application
332 so that it knows it tried to feed us garbage.
335 The next pair of ioctl operations select which tuner is to be used and let
336 the application find the tuner properties. We have only a single FM band
337 tuner in our example device.
344 struct video_tuner v;
345 if(copy_from_user(&v, arg, sizeof(v))!=0)
349 v.rangelow=(87*16000);
350 v.rangehigh=(108*16000);
351 v.flags = VIDEO_TUNER_LOW;
352 v.mode = VIDEO_MODE_AUTO;
354 strcpy(v.name, "FM");
355 if(copy_to_user(&v, arg, sizeof(v))!=0)
362 The VIDIOCGTUNER ioctl allows applications to query a tuner. The application
363 sets the tuner field to the tuner number it wishes to query. The query does
364 not change the tuner that is being used, it merely enquires about the tuner
368 We have exactly one tuner so after copying the user buffer to our temporary
369 structure we complain if they asked for a tuner other than tuner 0.
372 The video_tuner structure has the following fields
374 <table frame="all"><title>struct video_tuner fields</title>
375 <tgroup cols="2" align="left">
378 <entry>int tuner</><entry>The number of the tuner in question</entry>
380 <entry>char name[32]</><entry>A text description of this tuner. "FM" will do fine.
381 This is intended for the application.</entry>
384 <entry>Tuner capability flags</entry>
387 <entry>u16 mode</><entry>The current reception mode</entry>
390 <entry>u16 signal</><entry>The signal strength scaled between 0 and 65535. If
391 a device cannot tell the signal strength it should
392 report 65535. Many simple cards contain only a
393 signal/no signal bit. Such cards will report either
397 <entry>u32 rangelow, rangehigh</><entry>
398 The range of frequencies supported by the radio
399 or TV. It is scaled according to the VIDEO_TUNER_LOW
407 <table frame="all"><title>struct video_tuner flags</title>
408 <tgroup cols="2" align="left">
411 <entry>VIDEO_TUNER_PAL</><entry>A PAL TV tuner</entry>
413 <entry>VIDEO_TUNER_NTSC</><entry>An NTSC (US) TV tuner</entry>
415 <entry>VIDEO_TUNER_SECAM</><entry>A SECAM (French) TV tuner</entry>
417 <entry>VIDEO_TUNER_LOW</><entry>
418 The tuner frequency is scaled in 1/16th of a KHz
419 steps. If not it is in 1/16th of a MHz steps
422 <entry>VIDEO_TUNER_NORM</><entry>The tuner can set its format</entry>
424 <entry>VIDEO_TUNER_STEREO_ON</><entry>The tuner is currently receiving a stereo signal</entry>
430 <table frame="all"><title>struct video_tuner modes</title>
431 <tgroup cols="2" align="left">
434 <entry>VIDEO_MODE_PAL</><entry>PAL Format</entry>
436 <entry>VIDEO_MODE_NTSC</><entry>NTSC Format (USA)</entry>
438 <entry>VIDEO_MODE_SECAM</><entry>French Format</entry>
440 <entry>VIDEO_MODE_AUTO</><entry>A device that does not need to do
441 TV format switching</entry>
447 The settings for the radio card are thus fairly simple. We report that we
448 are a tuner called "FM" for FM radio. In order to get the best tuning
449 resolution we report VIDEO_TUNER_LOW and select tuning to 1/16th of KHz. Its
450 unlikely our card can do that resolution but it is a fair bet the card can
451 do better than 1/16th of a MHz. VIDEO_TUNER_LOW is appropriate to almost all
455 We report that the tuner automatically handles deciding what format it is
456 receiving - true enough as it only handles FM radio. Our example card is
457 also incapable of detecting stereo or signal strengths so it reports a
458 strength of 0xFFFF (maximum) and no stereo detected.
461 To finish off we set the range that can be tuned to be 87-108Mhz, the normal
462 FM broadcast radio range. It is important to find out what the card is
463 actually capable of tuning. It is easy enough to simply use the FM broadcast
464 range. Unfortunately if you do this you will discover the FM broadcast
465 ranges in the USA, Europe and Japan are all subtly different and some users
466 cannot receive all the stations they wish.
469 The application also needs to be able to set the tuner it wishes to use. In
470 our case, with a single tuner this is rather simple to arrange.
476 struct video_tuner v;
477 if(copy_from_user(&v, arg, sizeof(v)))
486 We copy the user supplied structure into kernel memory so we can examine it.
487 If the user has selected a tuner other than zero we reject the request. If
488 they wanted tuner 0 then, surprisingly enough, that is the current tuner already.
491 The next two ioctls we need to provide are to get and set the frequency of
492 the radio. These both use an unsigned long argument which is the frequency.
493 The scale of the frequency depends on the VIDEO_TUNER_LOW flag as I
494 mentioned earlier on. Since we have VIDEO_TUNER_LOW set this will be in
499 static unsigned long current_freq;
504 if(copy_to_user(arg, &current_freq,
505 sizeof(unsigned long))
511 Querying the frequency in our case is relatively simple. Our radio card is
512 too dumb to let us query the signal strength so we remember our setting if
513 we know it. All we have to do is copy it to the user.
521 if(copy_from_user(arg, &freq,
522 sizeof(unsigned long))!=0)
524 if(hardware_set_freq(freq)<0)
532 Setting the frequency is a little more complex. We begin by copying the
533 desired frequency into kernel space. Next we call a hardware specific routine
534 to set the radio up. This might be as simple as some scaling and a few
535 writes to an I/O port. For most radio cards it turns out a good deal more
536 complicated and may involve programming things like a phase locked loop on
537 the card. This is what documentation is for.
540 The final set of operations we need to provide for our radio are the
541 volume controls. Not all radio cards can even do volume control. After all
542 there is a perfectly good volume control on the sound card. We will assume
543 our radio card has a simple 4 step volume control.
546 There are two ioctls with audio we need to support
550 static int current_volume=0;
554 struct video_audio v;
555 if(copy_from_user(&v, arg, sizeof(v)))
559 v.volume = 16384*current_volume;
561 strcpy(v.name, "Radio");
562 v.mode = VIDEO_SOUND_MONO;
567 if(copy_to_user(arg. &v, sizeof(v)))
574 Much like the tuner we start by copying the user structure into kernel
575 space. Again we check if the user has asked for a valid audio input. We have
576 only input 0 and we punt if they ask for another input.
579 Then we fill in the video_audio structure. This has the following format
581 <table frame="all"><title>struct video_audio fields</title>
582 <tgroup cols="2" align="left">
585 <entry>audio</><entry>The input the user wishes to query</>
587 <entry>volume</><entry>The volume setting on a scale of 0-65535</>
589 <entry>base</><entry>The base level on a scale of 0-65535</>
591 <entry>treble</><entry>The treble level on a scale of 0-65535</>
593 <entry>flags</><entry>The features this audio device supports
596 <entry>name</><entry>A text name to display to the user. We picked
597 "Radio" as it explains things quite nicely.</>
599 <entry>mode</><entry>The current reception mode for the audio
601 We report MONO because our card is too stupid to know if it is in
605 <entry>balance</><entry>The stereo balance on a scale of 0-65535, 32768 is
608 <entry>step</><entry>The step by which the volume control jumps. This is
609 used to help make it easy for applications to set
616 <table frame="all"><title>struct video_audio flags</title>
617 <tgroup cols="2" align="left">
620 <entry>VIDEO_AUDIO_MUTE</><entry>The audio is currently muted. We
621 could fake this in our driver but we
622 choose not to bother.</entry>
624 <entry>VIDEO_AUDIO_MUTABLE</><entry>The input has a mute option</entry>
626 <entry>VIDEO_AUDIO_TREBLE</><entry>The input has a treble control</entry>
628 <entry>VIDEO_AUDIO_BASS</><entry>The input has a base control</entry>
634 <table frame="all"><title>struct video_audio modes</title>
635 <tgroup cols="2" align="left">
638 <entry>VIDEO_SOUND_MONO</><entry>Mono sound</entry>
640 <entry>VIDEO_SOUND_STEREO</><entry>Stereo sound</entry>
642 <entry>VIDEO_SOUND_LANG1</><entry>Alternative language 1 (TV specific)</entry>
644 <entry>VIDEO_SOUND_LANG2</><entry>Alternative language 2 (TV specific)</entry>
650 Having filled in the structure we copy it back to user space.
653 The VIDIOCSAUDIO ioctl allows the user to set the audio parameters in the
654 video_audio structure. The driver does its best to honour the request.
660 struct video_audio v;
661 if(copy_from_user(&v, arg, sizeof(v)))
665 current_volume = v/16384;
666 hardware_set_volume(current_volume);
672 In our case there is very little that the user can set. The volume is
673 basically the limit. Note that we could pretend to have a mute feature
680 struct video_audio v;
681 if(copy_from_user(&v, arg, sizeof(v)))
685 current_volume = v/16384;
686 if(v.flags&VIDEO_AUDIO_MUTE)
687 hardware_set_volume(0);
689 hardware_set_volume(current_volume);
690 current_muted = v.flags &
697 This with the corresponding changes to the VIDIOCGAUDIO code to report the
698 state of the mute flag we save and to report the card has a mute function,
699 will allow applications to use a mute facility with this card. It is
700 questionable whether this is a good idea however. User applications can already
701 fake this themselves and kernel space is precious.
704 We now have a working radio ioctl handler. So we just wrap up the function
715 and pass the Video4Linux layer back an error so that it knows we did not
716 understand the request we got passed.
719 <sect1 id="modradio">
720 <title>Module Wrapper</title>
722 Finally we add in the usual module wrapping and the driver is done.
728 static int io = 0x300;
736 MODULE_AUTHOR("Alan Cox");
737 MODULE_DESCRIPTION("A driver for an imaginary radio card.");
738 module_param(io, int, 0444);
739 MODULE_PARM_DESC(io, "I/O address of the card.");
741 static int __init init(void)
746 "You must set an I/O address with io=0x???\n");
749 return myradio_init(NULL);
752 static void __exit cleanup(void)
754 video_unregister_device(&my_radio);
755 release_region(io, MY_IO_SIZE);
759 module_exit(cleanup);
763 In this example we set the IO base by default if the driver is compiled into
764 the kernel: you can still set it using "my_radio.irq" if this file is called <filename>my_radio.c</filename>. For the module we require the
765 user sets the parameter. We set io to a nonsense port (-1) so that we can
766 tell if the user supplied an io parameter or not.
769 We use MODULE_ defines to give an author for the card driver and a
770 description. We also use them to declare that io is an integer and it is the
771 address of the card, and can be read by anyone from sysfs.
774 The clean-up routine unregisters the video_device we registered, and frees
775 up the I/O space. Note that the unregister takes the actual video_device
776 structure as its argument. Unlike the file operations structure which can be
777 shared by all instances of a device a video_device structure as an actual
778 instance of the device. If you are registering multiple radio devices you
779 need to fill in one structure per device (most likely by setting up a
780 template and copying it to each of the actual device structures).
785 <title>Video Capture Devices</title>
786 <sect1 id="introvid">
787 <title>Video Capture Device Types</title>
789 The video capture devices share the same interfaces as radio devices. In
790 order to explain the video capture interface I will use the example of a
791 camera that has no tuners or audio input. This keeps the example relatively
792 clean. To get both combine the two driver examples.
795 Video capture devices divide into four categories. A little technology
796 backgrounder. Full motion video even at television resolution (which is
797 actually fairly low) is pretty resource-intensive. You are continually
798 passing megabytes of data every second from the capture card to the display.
799 several alternative approaches have emerged because copying this through the
800 processor and the user program is a particularly bad idea .
803 The first is to add the television image onto the video output directly.
804 This is also how some 3D cards work. These basic cards can generally drop the
805 video into any chosen rectangle of the display. Cards like this, which
806 include most mpeg1 cards that used the feature connector, aren't very
807 friendly in a windowing environment. They don't understand windows or
808 clipping. The video window is always on the top of the display.
811 Chroma keying is a technique used by cards to get around this. It is an old
812 television mixing trick where you mark all the areas you wish to replace
813 with a single clear colour that isn't used in the image - TV people use an
814 incredibly bright blue while computing people often use a particularly
815 virulent purple. Bright blue occurs on the desktop. Anyone with virulent
816 purple windows has another problem besides their TV overlay.
819 The third approach is to copy the data from the capture card to the video
820 card, but to do it directly across the PCI bus. This relieves the processor
821 from doing the work but does require some smartness on the part of the video
822 capture chip, as well as a suitable video card. Programming this kind of
823 card and more so debugging it can be extremely tricky. There are some quite
824 complicated interactions with the display and you may also have to cope with
825 various chipset bugs that show up when PCI cards start talking to each
829 To keep our example fairly simple we will assume a card that supports
830 overlaying a flat rectangular image onto the frame buffer output, and which
831 can also capture stuff into processor memory.
835 <title>Registering Video Capture Devices</title>
837 This time we need to add more functions for our camera device.
840 static struct video_device my_camera
843 VID_TYPE_OVERLAY|VID_TYPE_SCALES|\
844 VID_TYPE_CAPTURE|VID_TYPE_CHROMAKEY,
845 VID_HARDWARE_MYCAMERA,
848 camera_read, /* no read */
850 camera_poll, /* no poll */
852 NULL, /* no special init function */
853 NULL /* no private data */
857 We need a read() function which is used for capturing data from
858 the card, and we need a poll function so that a driver can wait for the next
859 frame to be captured.
862 We use the extra video capability flags that did not apply to the
863 radio interface. The video related flags are
865 <table frame="all"><title>Capture Capabilities</title>
866 <tgroup cols="2" align="left">
869 <entry>VID_TYPE_CAPTURE</><entry>We support image capture</>
871 <entry>VID_TYPE_TELETEXT</><entry>A teletext capture device (vbi{n])</>
873 <entry>VID_TYPE_OVERLAY</><entry>The image can be directly overlaid onto the
876 <entry>VID_TYPE_CHROMAKEY</><entry>Chromakey can be used to select which parts
877 of the image to display</>
879 <entry>VID_TYPE_CLIPPING</><entry>It is possible to give the board a list of
880 rectangles to draw around. </>
882 <entry>VID_TYPE_FRAMERAM</><entry>The video capture goes into the video memory
883 and actually changes it. Applications need
884 to know this so they can clean up after the
887 <entry>VID_TYPE_SCALES</><entry>The image can be scaled to various sizes,
888 rather than being a single fixed size.</>
890 <entry>VID_TYPE_MONOCHROME</><entry>The capture will be monochrome. This isn't a
891 complete answer to the question since a mono
892 camera on a colour capture card will still
893 produce mono output.</>
895 <entry>VID_TYPE_SUBCAPTURE</><entry>The card allows only part of its field of
896 view to be captured. This enables
897 applications to avoid copying all of a large
898 image into memory when only some section is
905 We set VID_TYPE_CAPTURE so that we are seen as a capture card,
906 VID_TYPE_CHROMAKEY so the application knows it is time to draw in virulent
907 purple, and VID_TYPE_SCALES because we can be resized.
910 Our setup is fairly similar. This time we also want an interrupt line
911 for the 'frame captured' signal. Not all cards have this so some of them
912 cannot handle poll().
917 static int io = 0x320;
920 int __init mycamera_init(struct video_init *v)
922 if(!request_region(io, MY_IO_SIZE, "mycamera"))
925 "mycamera: port 0x%03X is in use.\n", io);
929 if(video_device_register(&my_camera,
930 VFL_TYPE_GRABBER)==-1) {
931 release_region(io, MY_IO_SIZE);
939 This is little changed from the needs of the radio card. We specify
940 VFL_TYPE_GRABBER this time as we want to be allocated a /dev/video name.
944 <title>Opening And Closing The Capture Device</title>
948 static int users = 0;
950 static int camera_open(stuct video_device *dev, int flags)
954 if(request_irq(irq, camera_irq, 0, "camera", dev)<0)
961 static int camera_close(struct video_device *dev)
968 The open and close routines are also quite similar. The only real change is
969 that we now request an interrupt for the camera device interrupt line. If we
970 cannot get the interrupt we report EBUSY to the application and give up.
974 <title>Interrupt Handling</title>
976 Our example handler is for an ISA bus device. If it was PCI you would be
977 able to share the interrupt and would have set SA_SHIRQ to indicate a
978 shared IRQ. We pass the device pointer as the interrupt routine argument. We
979 don't need to since we only support one card but doing this will make it
980 easier to upgrade the driver for multiple devices in the future.
983 Our interrupt routine needs to do little if we assume the card can simply
984 queue one frame to be read after it captures it.
989 static struct wait_queue *capture_wait;
990 static int capture_ready = 0;
992 static void camera_irq(int irq, void *dev_id,
993 struct pt_regs *regs)
996 wake_up_interruptible(&capture_wait);
1000 The interrupt handler is nice and simple for this card as we are assuming
1001 the card is buffering the frame for us. This means we have little to do but
1002 wake up anybody interested. We also set a capture_ready flag, as we may
1003 capture a frame before an application needs it. In this case we need to know
1004 that a frame is ready. If we had to collect the frame on the interrupt life
1005 would be more complex.
1008 The two new routines we need to supply are camera_read which returns a
1009 frame, and camera_poll which waits for a frame to become ready.
1014 static int camera_poll(struct video_device *dev,
1015 struct file *file, struct poll_table *wait)
1017 poll_wait(file, &capture_wait, wait);
1019 return POLLIN|POLLRDNORM;
1025 Our wait queue for polling is the capture_wait queue. This will cause the
1026 task to be woken up by our camera_irq routine. We check capture_read to see
1027 if there is an image present and if so report that it is readable.
1031 <title>Reading The Video Image</title>
1035 static long camera_read(struct video_device *dev, char *buf,
1036 unsigned long count)
1038 struct wait_queue wait = { current, NULL };
1043 add_wait_queue(&capture_wait, &wait);
1045 while(!capture_ready)
1047 if(file->flags&O_NDELAY)
1049 remove_wait_queue(&capture_wait, &wait);
1050 current->state = TASK_RUNNING;
1051 return -EWOULDBLOCK;
1053 if(signal_pending(current))
1055 remove_wait_queue(&capture_wait, &wait);
1056 current->state = TASK_RUNNING;
1057 return -ERESTARTSYS;
1060 current->state = TASK_INTERRUPTIBLE;
1062 remove_wait_queue(&capture_wait, &wait);
1063 current->state = TASK_RUNNING;
1067 The first thing we have to do is to ensure that the application waits until
1068 the next frame is ready. The code here is almost identical to the mouse code
1069 we used earlier in this chapter. It is one of the common building blocks of
1070 Linux device driver code and probably one which you will find occurs in any
1074 We wait for a frame to be ready, or for a signal to interrupt our waiting. If a
1075 signal occurs we need to return from the system call so that the signal can
1076 be sent to the application itself. We also check to see if the user actually
1077 wanted to avoid waiting - ie if they are using non-blocking I/O and have other things
1081 Next we copy the data from the card to the user application. This is rarely
1082 as easy as our example makes out. We will add capture_w, and capture_h here
1083 to hold the width and height of the captured image. We assume the card only
1084 supports 24bit RGB for now.
1093 len = capture_w * 3 * capture_h; /* 24bit RGB */
1096 len=count; /* Doesn't all fit */
1098 for(i=0; i<len; i++)
1100 put_user(inb(io+IMAGE_DATA), ptr);
1104 hardware_restart_capture();
1111 For a real hardware device you would try to avoid the loop with put_user().
1112 Each call to put_user() has a time overhead checking whether the accesses to user
1113 space are allowed. It would be better to read a line into a temporary buffer
1114 then copy this to user space in one go.
1117 Having captured the image and put it into user space we can kick the card to
1118 get the next frame acquired.
1122 <title>Video Ioctl Handling</title>
1124 As with the radio driver the major control interface is via the ioctl()
1125 function. Video capture devices support the same tuner calls as a radio
1126 device and also support additional calls to control how the video functions
1127 are handled. In this simple example the card has no tuners to avoid making
1134 static int camera_ioctl(struct video_device *dev, unsigned int cmd, void *arg)
1140 struct video_capability v;
1141 v.type = VID_TYPE_CAPTURE|\
1142 VID_TYPE_CHROMAKEY|\
1151 strcpy(v.name, "My Camera");
1152 if(copy_to_user(arg, &v, sizeof(v)))
1160 The first ioctl we must support and which all video capture and radio
1161 devices are required to support is VIDIOCGCAP. This behaves exactly the same
1162 as with a radio device. This time, however, we report the extra capabilities
1163 we outlined earlier on when defining our video_dev structure.
1166 We now set the video flags saying that we support overlay, capture,
1167 scaling and chromakey. We also report size limits - our smallest image is
1168 16x16 pixels, our largest is 640x480.
1171 To keep things simple we report no audio and no tuning capabilities at all.
1177 struct video_channel v;
1178 if(copy_from_user(&v, arg, sizeof(v)))
1184 v.type = VIDEO_TYPE_CAMERA;
1185 v.norm = VIDEO_MODE_AUTO;
1186 strcpy(v.name, "Camera Input");break;
1187 if(copy_to_user(&v, arg, sizeof(v)))
1195 This follows what is very much the standard way an ioctl handler looks
1196 in Linux. We copy the data into a kernel space variable and we check that the
1197 request is valid (in this case that the input is 0). Finally we copy the
1198 camera info back to the user.
1201 The VIDIOCGCHAN ioctl allows a user to ask about video channels (that is
1202 inputs to the video card). Our example card has a single camera input. The
1203 fields in the structure are
1205 <table frame="all"><title>struct video_channel fields</title>
1206 <tgroup cols="2" align="left">
1210 <entry>channel</><entry>The channel number we are selecting</entry>
1212 <entry>name</><entry>The name for this channel. This is intended
1213 to describe the port to the user.
1214 Appropriate names are therefore things like
1215 "Camera" "SCART input"</entry>
1217 <entry>flags</><entry>Channel properties</entry>
1219 <entry>type</><entry>Input type</entry>
1221 <entry>norm</><entry>The current television encoding being used
1222 if relevant for this channel.
1228 <table frame="all"><title>struct video_channel flags</title>
1229 <tgroup cols="2" align="left">
1232 <entry>VIDEO_VC_TUNER</><entry>Channel has a tuner.</entry>
1234 <entry>VIDEO_VC_AUDIO</><entry>Channel has audio.</entry>
1239 <table frame="all"><title>struct video_channel types</title>
1240 <tgroup cols="2" align="left">
1243 <entry>VIDEO_TYPE_TV</><entry>Television input.</entry>
1245 <entry>VIDEO_TYPE_CAMERA</><entry>Fixed camera input.</entry>
1247 <entry>0</><entry>Type is unknown.</entry>
1252 <table frame="all"><title>struct video_channel norms</title>
1253 <tgroup cols="2" align="left">
1256 <entry>VIDEO_MODE_PAL</><entry>PAL encoded Television</entry>
1258 <entry>VIDEO_MODE_NTSC</><entry>NTSC (US) encoded Television</entry>
1260 <entry>VIDEO_MODE_SECAM</><entry>SECAM (French) Television </entry>
1262 <entry>VIDEO_MODE_AUTO</><entry>Automatic switching, or format does not
1269 The corresponding VIDIOCSCHAN ioctl allows a user to change channel and to
1270 request the norm is changed - for example to switch between a PAL or an NTSC
1278 struct video_channel v;
1279 if(copy_from_user(&v, arg, sizeof(v)))
1283 if(v.norm != VIDEO_MODE_AUTO)
1291 The implementation of this call in our driver is remarkably easy. Because we
1292 are assuming fixed format hardware we need only check that the user has not
1293 tried to change anything.
1296 The user also needs to be able to configure and adjust the picture they are
1297 seeing. This is much like adjusting a television set. A user application
1298 also needs to know the palette being used so that it knows how to display
1299 the image that has been captured. The VIDIOCGPICT and VIDIOCSPICT ioctl
1300 calls provide this information.
1307 struct video_picture v;
1308 v.brightness = hardware_brightness();
1309 v.hue = hardware_hue();
1310 v.colour = hardware_saturation();
1311 v.contrast = hardware_brightness();
1313 v.whiteness = 32768;
1314 v.depth = 24; /* 24bit */
1315 v.palette = VIDEO_PALETTE_RGB24;
1316 if(copy_to_user(&v, arg,
1325 The brightness, hue, color, and contrast provide the picture controls that
1326 are akin to a conventional television. Whiteness provides additional
1327 control for greyscale images. All of these values are scaled between 0-65535
1328 and have 32768 as the mid point setting. The scaling means that applications
1329 do not have to worry about the capability range of the hardware but can let
1330 it make a best effort attempt.
1333 Our depth is 24, as this is in bits. We will be returning RGB24 format. This
1334 has one byte of red, then one of green, then one of blue. This then repeats
1335 for every other pixel in the image. The other common formats the interface
1338 <table frame="all"><title>Framebuffer Encodings</title>
1339 <tgroup cols="2" align="left">
1342 <entry>GREY</><entry>Linear greyscale. This is for simple cameras and the
1345 <entry>RGB565</><entry>The top 5 bits hold 32 red levels, the next six bits
1346 hold green and the low 5 bits hold blue. </>
1348 <entry>RGB555</><entry>The top bit is clear. The red green and blue levels
1349 each occupy five bits.</>
1355 Additional modes are support for YUV capture formats. These are common for
1356 TV and video conferencing applications.
1359 The VIDIOCSPICT ioctl allows a user to set some of the picture parameters.
1360 Exactly which ones are supported depends heavily on the card itself. It is
1361 possible to support many modes and effects in software. In general doing
1362 this in the kernel is a bad idea. Video capture is a performance-sensitive
1363 application and the programs can often do better if they aren't being
1364 'helped' by an overkeen driver writer. Thus for our device we will report
1365 RGB24 only and refuse to allow a change.
1372 struct video_picture v;
1373 if(copy_from_user(&v, arg, sizeof(v)))
1376 v.palette != VIDEO_PALETTE_RGB24)
1378 set_hardware_brightness(v.brightness);
1379 set_hardware_hue(v.hue);
1380 set_hardware_saturation(v.colour);
1381 set_hardware_brightness(v.contrast);
1388 We check the user has not tried to change the palette or the depth. We do
1389 not want to carry out some of the changes and then return an error. This may
1390 confuse the application which will be assuming no change occurred.
1393 In much the same way as you need to be able to set the picture controls to
1394 get the right capture images, many cards need to know what they are
1395 displaying onto when generating overlay output. In some cases getting this
1396 wrong even makes a nasty mess or may crash the computer. For that reason
1397 the VIDIOCSBUF ioctl used to set up the frame buffer information may well
1398 only be usable by root.
1401 We will assume our card is one of the old ISA devices with feature connector
1402 and only supports a couple of standard video modes. Very common for older
1403 cards although the PCI devices are way smarter than this.
1408 static struct video_buffer capture_fb;
1412 if(copy_to_user(arg, &capture_fb,
1413 sizeof(capture_fb)))
1422 We keep the frame buffer information in the format the ioctl uses. This
1423 makes it nice and easy to work with in the ioctl calls.
1429 struct video_buffer v;
1431 if(!capable(CAP_SYS_ADMIN))
1434 if(copy_from_user(&v, arg, sizeof(v)))
1436 if(v.width!=320 && v.width!=640)
1438 if(v.height!=200 && v.height!=240
1439 && v.height!=400
1440 && v.height !=480)
1442 memcpy(&capture_fb, &v, sizeof(v));
1443 hardware_set_fb(&v);
1451 The capable() function checks a user has the required capability. The Linux
1452 operating system has a set of about 30 capabilities indicating privileged
1453 access to services. The default set up gives the superuser (uid 0) all of
1454 them and nobody else has any.
1457 We check that the user has the SYS_ADMIN capability, that is they are
1458 allowed to operate as the machine administrator. We don't want anyone but
1459 the administrator making a mess of the display.
1462 Next we check for standard PC video modes (320 or 640 wide with either
1463 EGA or VGA depths). If the mode is not a standard video mode we reject it as
1464 not supported by our card. If the mode is acceptable we save it so that
1465 VIDIOCFBUF will give the right answer next time it is called. The
1466 hardware_set_fb() function is some undescribed card specific function to
1467 program the card for the desired mode.
1470 Before the driver can display an overlay window it needs to know where the
1471 window should be placed, and also how large it should be. If the card
1472 supports clipping it needs to know which rectangles to omit from the
1473 display. The video_window structure is used to describe the way the image
1474 should be displayed.
1476 <table frame="all"><title>struct video_window fields</title>
1477 <tgroup cols="2" align="left">
1480 <entry>width</><entry>The width in pixels of the desired image. The card
1481 may use a smaller size if this size is not available</>
1483 <entry>height</><entry>The height of the image. The card may use a smaller
1484 size if this size is not available.</>
1486 <entry>x</><entry> The X position of the top left of the window. This
1487 is in pixels relative to the left hand edge of the
1488 picture. Not all cards can display images aligned on
1489 any pixel boundary. If the position is unsuitable
1490 the card adjusts the image right and reduces the
1493 <entry>y</><entry> The Y position of the top left of the window. This
1494 is counted in pixels relative to the top edge of the
1495 picture. As with the width if the card cannot
1496 display starting on this line it will adjust the
1499 <entry>chromakey</><entry>The colour (expressed in RGB32 format) for the
1500 chromakey colour if chroma keying is being used. </>
1502 <entry>clips</><entry>An array of rectangles that must not be drawn
1505 <entry>clipcount</><entry>The number of clips in this array.</>
1511 Each clip is a struct video_clip which has the following fields
1513 <table frame="all"><title>video_clip fields</title>
1514 <tgroup cols="2" align="left">
1517 <entry>x, y</><entry>Co-ordinates relative to the display</>
1519 <entry>width, height</><entry>Width and height in pixels</>
1521 <entry>next</><entry>A spare field for the application to use</>
1527 The driver is required to ensure it always draws in the area requested or a smaller area, and that it never draws in any of the areas that are clipped.
1528 This may well mean it has to leave alone. small areas the application wished to be
1532 Our example card uses chromakey so does not have to address most of the
1533 clipping. We will add a video_window structure to our global variables to
1534 remember our parameters, as we did with the frame buffer.
1541 if(copy_to_user(arg, &capture_win,
1542 sizeof(capture_win)))
1550 struct video_window v;
1551 if(copy_from_user(&v, arg, sizeof(v)))
1553 if(v.width > 640 || v.height > 480)
1555 if(v.width < 16 || v.height < 16)
1557 hardware_set_key(v.chromakey);
1558 hardware_set_window(v);
1559 memcpy(&capture_win, &v, sizeof(v));
1560 capture_w = v.width;
1561 capture_h = v.height;
1568 Because we are using Chromakey our setup is fairly simple. Mostly we have to
1569 check the values are sane and load them into the capture card.
1572 With all the setup done we can now turn on the actual capture/overlay. This
1573 is done with the VIDIOCCAPTURE ioctl. This takes a single integer argument
1574 where 0 is on and 1 is off.
1582 if(get_user(v, (int *)arg))
1585 hardware_capture_off();
1588 if(capture_fb.width == 0
1591 hardware_capture_on();
1599 We grab the flag from user space and either enable or disable according to
1600 its value. There is one small corner case we have to consider here. Suppose
1601 that the capture was requested before the video window or the frame buffer
1602 had been set up. In those cases there will be unconfigured fields in our
1603 card data, as well as unconfigured hardware settings. We check for this case and
1604 return an error if the frame buffer or the capture window width is zero.
1610 return -ENOIOCTLCMD;
1616 We don't need to support any other ioctls, so if we get this far, it is time
1617 to tell the video layer that we don't now what the user is talking about.
1621 <title>Other Functionality</title>
1623 The Video4Linux layer supports additional features, including a high
1624 performance mmap() based capture mode and capturing part of the image.
1625 These features are out of the scope of the book. You should however have enough
1626 example code to implement most simple video4linux devices for radio and TV
1632 <title>Known Bugs And Assumptions</title>
1635 <varlistentry><term>Multiple Opens</term>
1638 The driver assumes multiple opens should not be allowed. A driver
1639 can work around this but not cleanly.
1641 </listitem></varlistentry>
1643 <varlistentry><term>API Deficiencies</term>
1646 The existing API poorly reflects compression capable devices. There
1647 are plans afoot to merge V4L, V4L2 and some other ideas into a
1650 </listitem></varlistentry>
1656 <chapter id="pubfunctions">
1657 <title>Public Functions Provided</title>
1658 !Edrivers/media/video/videodev.c