| blob | fdff984a516188266df0845e3e3e1b181d9972e1 |
6 <bookinfo>
9 <authorgroup>
10 <author>
13 <affiliation>
14 <address>
16 </address>
17 </affiliation>
18 </author>
19 </authorgroup>
21 <copyright>
24 </copyright>
26 <legalnotice>
27 <para>
28 This documentation is free software; you can redistribute
29 it and/or modify it under the terms of the GNU General Public
30 License as published by the Free Software Foundation; either
31 version 2 of the License, or (at your option) any later
32 version.
33 </para>
35 <para>
36 This program is distributed in the hope that it will be
37 useful, but WITHOUT ANY WARRANTY; without even the implied
38 warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
39 See the GNU General Public License for more details.
40 </para>
42 <para>
43 You should have received a copy of the GNU General Public
44 License along with this program; if not, write to the Free
47 </para>
49 <para>
50 For more details see the file COPYING in the source
51 distribution of Linux.
52 </para>
53 </legalnotice>
54 </bookinfo>
56 <toc></toc>
60 <para>
61 Parts of this document first appeared in Linux Magazine under a
62 ninety day exclusivity.
63 </para>
64 <para>
65 Video4Linux is intended to provide a common programming interface
66 for the many TV and capture cards now on the market, as well as
67 parallel port and USB video cameras. Radio, teletext decoders and
68 vertical blanking data interfaces are also provided.
69 </para>
70 </chapter>
73 <para>
74 There are a wide variety of radio interfaces available for PC's, and these
75 are generally very simple to program. The biggest problem with supporting
76 such devices is normally extracting documentation from the vendor.
77 </para>
78 <para>
79 The radio interface supports a simple set of control ioctls standardised
80 across all radio and tv interfaces. It does not support read or write, which
81 are used for video streams. The reason radio cards do not allow you to read
82 the audio stream into an application is that without exception they provide
83 a connection on to a soundcard. Soundcards can be used to read the radio
84 data just fine.
85 </para>
88 <para>
89 The Video4linux core provides an interface for registering devices. The
90 first step in writing our radio card driver is to register it.
91 </para>
92 <programlisting>
95 static struct video_device my_radio
96 {
97 "My radio",
98 VID_TYPE_TUNER,
99 VID_HARDWARE_MYRADIO,
100 radio_open.
101 radio_close,
102 NULL, /* no read */
103 NULL, /* no write */
104 NULL, /* no poll */
105 radio_ioctl,
106 NULL, /* no special init function */
107 NULL /* no private data */
108 };
111 </programlisting>
112 <para>
113 This declares our video4linux device driver interface. The VID_TYPE_ value
114 defines what kind of an interface we are, and defines basic capabilities.
115 </para>
116 <para>
117 The only defined value relevant for a radio card is VID_TYPE_TUNER which
118 indicates that the device can be tuned. Clearly our radio is going to have some
119 way to change channel so it is tuneable.
120 </para>
121 <para>
122 The VID_HARDWARE_ types are unique to each device. Numbers are assigned by
124 can pull a suitably large number out of your hat and use it. 10000 should be
125 safe for a very long time even allowing for the huge number of vendors
126 making new and different radio cards at the moment.
127 </para>
128 <para>
129 We declare an open and close routine, but we do not need read or write,
130 which are used to read and write video data to or from the card itself. As
131 we have no read or write there is no poll function.
132 </para>
133 <para>
134 The private initialise function is run when the device is registered. In
135 this driver we've already done all the work needed. The final pointer is a
136 private data pointer that can be used by the device driver to attach and
137 retrieve private data structures. We set this field "priv" to NULL for
138 the moment.
139 </para>
140 <para>
141 Having the structure defined is all very well but we now need to register it
142 with the kernel.
143 </para>
144 <programlisting>
147 static int io = 0x320;
149 int __init myradio_init(struct video_init *v)
150 {
151 if(!request_region(io, MY_IO_SIZE, "myradio"))
152 {
153 printk(KERN_ERR
154 "myradio: port 0x%03X is in use.\n", io);
155 return -EBUSY;
156 }
159 release_region(io, MY_IO_SIZE);
160 return -EINVAL;
161 }
162 return 0;
163 }
165 </programlisting>
166 <para>
167 The first stage of the initialisation, as is normally the case, is to check
168 that the I/O space we are about to fiddle with doesn't belong to some other
169 driver. If it is we leave well alone. If the user gives the address of the
170 wrong device then we will spot this. These policies will generally avoid
171 crashing the machine.
172 </para>
173 <para>
174 Now we ask the Video4Linux layer to register the device for us. We hand it
175 our carefully designed video_device structure and also tell it which group
176 of devices we want it registered with. In this case VFL_TYPE_RADIO.
177 </para>
178 <para>
179 The types available are
180 </para>
183 <tbody>
184 <row>
187 Radio devices are assigned in this block. As with all of these
188 selections the actual number assignment is done by the video layer
189 accordijng to what is free.</entry>
190 </row><row>
192 Video capture devices and also -- counter-intuitively for the name --
193 hardware video playback devices such as MPEG2 cards.</entry>
194 </row><row>
196 The VBI devices capture the hidden lines on a television picture
197 that carry further information like closed caption data, teletext
198 (primarily in Europe) and now Intercast and the ATVEC internet
199 television encodings.</entry>
200 </row><row>
202 VTX is 'Videotext' also known as 'Teletext'. This is a system for
203 sending numbered, 40x25, mostly textual page images over the hidden
204 lines. Unlike the /dev/vbi interfaces, this is for 'smart' decoder
205 chips. (The use of the word smart here has to be taken in context,
206 the smartest teletext chips are fairly dumb pieces of technology).
207 </entry>
208 </row>
209 </tbody>
210 </tgroup>
211 </table>
212 <para>
213 We are most definitely a radio.
214 </para>
215 <para>
216 Finally we allocate our I/O space so that nobody treads on us and return 0
217 to signify general happiness with the state of the universe.
218 </para>
219 </sect1>
223 <para>
224 The functions we declared in our video_device are mostly very simple.
225 Firstly we can drop in what is basically standard code for open and close.
226 </para>
227 <programlisting>
230 static int users = 0;
232 static int radio_open(struct video_device *dev, int flags)
233 {
234 if(users)
235 return -EBUSY;
236 users++;
237 return 0;
238 }
240 </programlisting>
241 <para>
242 At open time we need to do nothing but check if someone else is also using
243 the radio card. If nobody is using it we make a note that we are using it,
244 then we ensure that nobody unloads our driver on us.
245 </para>
246 <programlisting>
249 static int radio_close(struct video_device *dev)
250 {
251 users--;
252 }
254 </programlisting>
255 <para>
256 At close time we simply need to reduce the user count and allow the module
257 to become unloadable.
258 </para>
259 <para>
260 If you are sharp you will have noticed neither the open nor the close
261 routines attempt to reset or change the radio settings. This is intentional.
262 It allows an application to set up the radio and exit. It avoids a user
263 having to leave an application running all the time just to listen to the
264 radio.
265 </para>
266 </sect1>
269 <para>
270 This leaves the ioctl routine, without which the driver will not be
271 terribly useful to anyone.
272 </para>
273 <programlisting>
276 static int radio_ioctl(struct video_device *dev, unsigned int cmd, void *arg)
277 {
278 switch(cmd)
279 {
280 case VIDIOCGCAP:
281 {
282 struct video_capability v;
283 v.type = VID_TYPE_TUNER;
284 v.channels = 1;
285 v.audios = 1;
286 v.maxwidth = 0;
287 v.minwidth = 0;
288 v.maxheight = 0;
289 v.minheight = 0;
290 strcpy(v.name, "My Radio");
291 if(copy_to_user(arg, &v, sizeof(v)))
292 return -EFAULT;
293 return 0;
294 }
296 </programlisting>
297 <para>
298 VIDIOCGCAP is the first ioctl all video4linux devices must support. It
299 allows the applications to find out what sort of a card they have found and
300 to figure out what they want to do about it. The fields in the structure are
301 </para>
304 <tbody>
305 <row>
307 </row><row>
309 this card. It could even by zero for a card that has
310 no tuning capability. For our simple FM radio it is 1.
312 </row><row>
314 radio there is only one audio input.</entry>
315 </row><row>
317 images in. We set these to zero. Radios do not
318 capture pictures</entry>
319 </row><row>
321 capturing. For our radio we report 0.
322 </entry>
323 </row><row>
325 matches the field we filled in in the struct
326 video_device when registering.</entry>
327 </row>
328 </tbody>
329 </tgroup>
330 </table>
331 <para>
332 Having filled in the fields, we use copy_to_user to copy the structure into
333 the users buffer. If the copy fails we return an EFAULT to the application
334 so that it knows it tried to feed us garbage.
335 </para>
336 <para>
337 The next pair of ioctl operations select which tuner is to be used and let
338 the application find the tuner properties. We have only a single FM band
339 tuner in our example device.
340 </para>
341 <programlisting>
344 case VIDIOCGTUNER:
345 {
346 struct video_tuner v;
348 return -EFAULT;
349 if(v.tuner)
350 return -EINVAL;
353 v.flags = VIDEO_TUNER_LOW;
354 v.mode = VIDEO_MODE_AUTO;
355 v.signal = 0xFFFF;
356 strcpy(v.name, "FM");
358 return -EFAULT;
359 return 0;
360 }
362 </programlisting>
363 <para>
364 The VIDIOCGTUNER ioctl allows applications to query a tuner. The application
365 sets the tuner field to the tuner number it wishes to query. The query does
366 not change the tuner that is being used, it merely enquires about the tuner
367 in question.
368 </para>
369 <para>
370 We have exactly one tuner so after copying the user buffer to our temporary
371 structure we complain if they asked for a tuner other than tuner 0.
372 </para>
373 <para>
374 The video_tuner structure has the following fields
375 </para>
378 <tbody>
379 <row>
381 </row><row>
383 This is intended for the application.</entry>
384 </row><row>
387 </row>
388 <row>
391 </row><row>
393 a device cannot tell the signal strength it should
394 report 65535. Many simple cards contain only a
395 signal/no signal bit. Such cards will report either
398 </row><row>
400 The range of frequencies supported by the radio
401 or TV. It is scaled according to the VIDEO_TUNER_LOW
402 flag.</entry>
404 </row>
405 </tbody>
406 </tgroup>
407 </table>
411 <tbody>
412 <row>
414 </row><row>
416 </row><row>
418 </row><row>
422 </entry>
423 </row><row>
425 </row><row>
426 <entry>VIDEO_TUNER_STEREO_ON</entry><entry>The tuner is currently receiving a stereo signal</entry>
427 </row>
428 </tbody>
429 </tgroup>
430 </table>
434 <tbody>
435 <row>
437 </row><row>
439 </row><row>
441 </row><row>
443 TV format switching</entry>
444 </row>
445 </tbody>
446 </tgroup>
447 </table>
448 <para>
449 The settings for the radio card are thus fairly simple. We report that we
450 are a tuner called "FM" for FM radio. In order to get the best tuning
452 unlikely our card can do that resolution but it is a fair bet the card can
454 radio usage.
455 </para>
456 <para>
457 We report that the tuner automatically handles deciding what format it is
458 receiving - true enough as it only handles FM radio. Our example card is
459 also incapable of detecting stereo or signal strengths so it reports a
460 strength of 0xFFFF (maximum) and no stereo detected.
461 </para>
462 <para>
464 FM broadcast radio range. It is important to find out what the card is
465 actually capable of tuning. It is easy enough to simply use the FM broadcast
466 range. Unfortunately if you do this you will discover the FM broadcast
467 ranges in the USA, Europe and Japan are all subtly different and some users
468 cannot receive all the stations they wish.
469 </para>
470 <para>
471 The application also needs to be able to set the tuner it wishes to use. In
472 our case, with a single tuner this is rather simple to arrange.
473 </para>
474 <programlisting>
476 case VIDIOCSTUNER:
477 {
478 struct video_tuner v;
479 if(copy_from_user(&v, arg, sizeof(v)))
480 return -EFAULT;
481 if(v.tuner != 0)
482 return -EINVAL;
483 return 0;
484 }
486 </programlisting>
487 <para>
488 We copy the user supplied structure into kernel memory so we can examine it.
489 If the user has selected a tuner other than zero we reject the request. If
490 they wanted tuner 0 then, surprisingly enough, that is the current tuner already.
491 </para>
492 <para>
493 The next two ioctls we need to provide are to get and set the frequency of
494 the radio. These both use an unsigned long argument which is the frequency.
495 The scale of the frequency depends on the VIDEO_TUNER_LOW flag as I
496 mentioned earlier on. Since we have VIDEO_TUNER_LOW set this will be in
498 </para>
499 <programlisting>
501 static unsigned long current_freq;
505 case VIDIOCGFREQ:
506 if(copy_to_user(arg, &current_freq,
507 sizeof(unsigned long))
508 return -EFAULT;
509 return 0;
511 </programlisting>
512 <para>
513 Querying the frequency in our case is relatively simple. Our radio card is
514 too dumb to let us query the signal strength so we remember our setting if
515 we know it. All we have to do is copy it to the user.
516 </para>
517 <programlisting>
520 case VIDIOCSFREQ:
521 {
522 u32 freq;
523 if(copy_from_user(arg, &freq,
524 sizeof(unsigned long))!=0)
525 return -EFAULT;
527 return -EINVAL;
528 current_freq = freq;
529 return 0;
530 }
532 </programlisting>
533 <para>
534 Setting the frequency is a little more complex. We begin by copying the
535 desired frequency into kernel space. Next we call a hardware specific routine
536 to set the radio up. This might be as simple as some scaling and a few
537 writes to an I/O port. For most radio cards it turns out a good deal more
538 complicated and may involve programming things like a phase locked loop on
539 the card. This is what documentation is for.
540 </para>
541 <para>
542 The final set of operations we need to provide for our radio are the
543 volume controls. Not all radio cards can even do volume control. After all
544 there is a perfectly good volume control on the sound card. We will assume
545 our radio card has a simple 4 step volume control.
546 </para>
547 <para>
548 There are two ioctls with audio we need to support
549 </para>
550 <programlisting>
552 static int current_volume=0;
554 case VIDIOCGAUDIO:
555 {
556 struct video_audio v;
557 if(copy_from_user(&v, arg, sizeof(v)))
558 return -EFAULT;
559 if(v.audio != 0)
560 return -EINVAL;
561 v.volume = 16384*current_volume;
562 v.step = 16384;
563 strcpy(v.name, "Radio");
564 v.mode = VIDEO_SOUND_MONO;
565 v.balance = 0;
566 v.base = 0;
567 v.treble = 0;
569 if(copy_to_user(arg. &v, sizeof(v)))
570 return -EFAULT;
571 return 0;
572 }
574 </programlisting>
575 <para>
576 Much like the tuner we start by copying the user structure into kernel
577 space. Again we check if the user has asked for a valid audio input. We have
578 only input 0 and we punt if they ask for another input.
579 </para>
580 <para>
581 Then we fill in the video_audio structure. This has the following format
582 </para>
585 <tbody>
586 <row>
588 </row><row>
590 </row><row>
592 </row><row>
594 </row><row>
596 </entry>
597 </row><row>
600 </row><row>
603 We report MONO because our card is too stupid to know if it is in
604 mono or stereo.
605 </entry>
606 </row><row>
608 middle.</entry>
609 </row><row>
611 used to help make it easy for applications to set
612 slider behaviour.</entry>
613 </row>
614 </tbody>
615 </tgroup>
616 </table>
620 <tbody>
621 <row>
623 could fake this in our driver but we
624 choose not to bother.</entry>
625 </row><row>
627 </row><row>
629 </row><row>
631 </row>
632 </tbody>
633 </tgroup>
634 </table>
638 <tbody>
639 <row>
641 </row><row>
643 </row><row>
645 </row><row>
647 </row>
648 </tbody>
649 </tgroup>
650 </table>
651 <para>
652 Having filled in the structure we copy it back to user space.
653 </para>
654 <para>
655 The VIDIOCSAUDIO ioctl allows the user to set the audio parameters in the
656 video_audio structure. The driver does its best to honour the request.
657 </para>
658 <programlisting>
660 case VIDIOCSAUDIO:
661 {
662 struct video_audio v;
663 if(copy_from_user(&v, arg, sizeof(v)))
664 return -EFAULT;
665 if(v.audio)
666 return -EINVAL;
667 current_volume = v/16384;
668 hardware_set_volume(current_volume);
669 return 0;
670 }
672 </programlisting>
673 <para>
674 In our case there is very little that the user can set. The volume is
675 basically the limit. Note that we could pretend to have a mute feature
676 by rewriting this to
677 </para>
678 <programlisting>
680 case VIDIOCSAUDIO:
681 {
682 struct video_audio v;
683 if(copy_from_user(&v, arg, sizeof(v)))
684 return -EFAULT;
685 if(v.audio)
686 return -EINVAL;
687 current_volume = v/16384;
688 if(v.flags&VIDEO_AUDIO_MUTE)
689 hardware_set_volume(0);
690 else
691 hardware_set_volume(current_volume);
692 current_muted = v.flags &
693 VIDEO_AUDIO_MUTE;
694 return 0;
695 }
697 </programlisting>
698 <para>
699 This with the corresponding changes to the VIDIOCGAUDIO code to report the
700 state of the mute flag we save and to report the card has a mute function,
701 will allow applications to use a mute facility with this card. It is
702 questionable whether this is a good idea however. User applications can already
703 fake this themselves and kernel space is precious.
704 </para>
705 <para>
706 We now have a working radio ioctl handler. So we just wrap up the function
707 </para>
708 <programlisting>
711 }
712 return -ENOIOCTLCMD;
713 }
715 </programlisting>
716 <para>
717 and pass the Video4Linux layer back an error so that it knows we did not
718 understand the request we got passed.
719 </para>
720 </sect1>
723 <para>
724 Finally we add in the usual module wrapping and the driver is done.
725 </para>
726 <programlisting>
728 #ifndef MODULE
730 static int io = 0x300;
732 #else
734 static int io = -1;
736 #endif
738 MODULE_AUTHOR("Alan Cox");
739 MODULE_DESCRIPTION("A driver for an imaginary radio card.");
740 module_param(io, int, 0444);
741 MODULE_PARM_DESC(io, "I/O address of the card.");
743 static int __init init(void)
744 {
745 if(io==-1)
746 {
747 printk(KERN_ERR
748 "You must set an I/O address with io=0x???\n");
749 return -EINVAL;
750 }
751 return myradio_init(NULL);
752 }
754 static void __exit cleanup(void)
755 {
756 video_unregister_device(&my_radio);
757 release_region(io, MY_IO_SIZE);
758 }
760 module_init(init);
761 module_exit(cleanup);
763 </programlisting>
764 <para>
765 In this example we set the IO base by default if the driver is compiled into
766 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
767 user sets the parameter. We set io to a nonsense port (-1) so that we can
768 tell if the user supplied an io parameter or not.
769 </para>
770 <para>
771 We use MODULE_ defines to give an author for the card driver and a
772 description. We also use them to declare that io is an integer and it is the
773 address of the card, and can be read by anyone from sysfs.
774 </para>
775 <para>
776 The clean-up routine unregisters the video_device we registered, and frees
777 up the I/O space. Note that the unregister takes the actual video_device
778 structure as its argument. Unlike the file operations structure which can be
779 shared by all instances of a device a video_device structure as an actual
780 instance of the device. If you are registering multiple radio devices you
781 need to fill in one structure per device (most likely by setting up a
782 template and copying it to each of the actual device structures).
783 </para>
784 </sect1>
785 </chapter>
786 <chapter>
790 <para>
791 The video capture devices share the same interfaces as radio devices. In
792 order to explain the video capture interface I will use the example of a
793 camera that has no tuners or audio input. This keeps the example relatively
794 clean. To get both combine the two driver examples.
795 </para>
796 <para>
797 Video capture devices divide into four categories. A little technology
798 backgrounder. Full motion video even at television resolution (which is
799 actually fairly low) is pretty resource-intensive. You are continually
800 passing megabytes of data every second from the capture card to the display.
801 several alternative approaches have emerged because copying this through the
802 processor and the user program is a particularly bad idea .
803 </para>
804 <para>
805 The first is to add the television image onto the video output directly.
806 This is also how some 3D cards work. These basic cards can generally drop the
807 video into any chosen rectangle of the display. Cards like this, which
808 include most mpeg1 cards that used the feature connector, aren't very
809 friendly in a windowing environment. They don't understand windows or
810 clipping. The video window is always on the top of the display.
811 </para>
812 <para>
813 Chroma keying is a technique used by cards to get around this. It is an old
814 television mixing trick where you mark all the areas you wish to replace
815 with a single clear colour that isn't used in the image - TV people use an
816 incredibly bright blue while computing people often use a particularly
817 virulent purple. Bright blue occurs on the desktop. Anyone with virulent
818 purple windows has another problem besides their TV overlay.
819 </para>
820 <para>
821 The third approach is to copy the data from the capture card to the video
822 card, but to do it directly across the PCI bus. This relieves the processor
823 from doing the work but does require some smartness on the part of the video
824 capture chip, as well as a suitable video card. Programming this kind of
825 card and more so debugging it can be extremely tricky. There are some quite
826 complicated interactions with the display and you may also have to cope with
827 various chipset bugs that show up when PCI cards start talking to each
828 other.
829 </para>
830 <para>
831 To keep our example fairly simple we will assume a card that supports
832 overlaying a flat rectangular image onto the frame buffer output, and which
833 can also capture stuff into processor memory.
834 </para>
835 </sect1>
838 <para>
839 This time we need to add more functions for our camera device.
840 </para>
841 <programlisting>
842 static struct video_device my_camera
843 {
844 "My Camera",
845 VID_TYPE_OVERLAY|VID_TYPE_SCALES|\
846 VID_TYPE_CAPTURE|VID_TYPE_CHROMAKEY,
847 VID_HARDWARE_MYCAMERA,
848 camera_open.
849 camera_close,
850 camera_read, /* no read */
851 NULL, /* no write */
852 camera_poll, /* no poll */
853 camera_ioctl,
854 NULL, /* no special init function */
855 NULL /* no private data */
856 };
857 </programlisting>
858 <para>
859 We need a read() function which is used for capturing data from
860 the card, and we need a poll function so that a driver can wait for the next
861 frame to be captured.
862 </para>
863 <para>
864 We use the extra video capability flags that did not apply to the
865 radio interface. The video related flags are
866 </para>
869 <tbody>
870 <row>
872 </row><row>
874 </row><row>
876 frame buffer</entry>
877 </row><row>
879 of the image to display</entry>
880 </row><row>
882 rectangles to draw around. </entry>
883 </row><row>
885 and actually changes it. Applications need
886 to know this so they can clean up after the
887 card</entry>
888 </row><row>
890 rather than being a single fixed size.</entry>
891 </row><row>
893 complete answer to the question since a mono
894 camera on a colour capture card will still
895 produce mono output.</entry>
896 </row><row>
898 view to be captured. This enables
899 applications to avoid copying all of a large
900 image into memory when only some section is
901 relevant.</entry>
902 </row>
903 </tbody>
904 </tgroup>
905 </table>
906 <para>
907 We set VID_TYPE_CAPTURE so that we are seen as a capture card,
908 VID_TYPE_CHROMAKEY so the application knows it is time to draw in virulent
909 purple, and VID_TYPE_SCALES because we can be resized.
910 </para>
911 <para>
912 Our setup is fairly similar. This time we also want an interrupt line
913 for the 'frame captured' signal. Not all cards have this so some of them
914 cannot handle poll().
915 </para>
916 <programlisting>
919 static int io = 0x320;
920 static int irq = 11;
922 int __init mycamera_init(struct video_init *v)
923 {
924 if(!request_region(io, MY_IO_SIZE, "mycamera"))
925 {
926 printk(KERN_ERR
927 "mycamera: port 0x%03X is in use.\n", io);
928 return -EBUSY;
929 }
931 if(video_device_register(&my_camera,
932 VFL_TYPE_GRABBER)==-1) {
933 release_region(io, MY_IO_SIZE);
934 return -EINVAL;
935 }
936 return 0;
937 }
939 </programlisting>
940 <para>
941 This is little changed from the needs of the radio card. We specify
942 VFL_TYPE_GRABBER this time as we want to be allocated a /dev/video name.
943 </para>
944 </sect1>
947 <programlisting>
950 static int users = 0;
952 static int camera_open(struct video_device *dev, int flags)
953 {
954 if(users)
955 return -EBUSY;
957 return -EBUSY;
958 users++;
959 return 0;
960 }
963 static int camera_close(struct video_device *dev)
964 {
965 users--;
966 free_irq(irq, dev);
967 }
968 </programlisting>
969 <para>
970 The open and close routines are also quite similar. The only real change is
971 that we now request an interrupt for the camera device interrupt line. If we
972 cannot get the interrupt we report EBUSY to the application and give up.
973 </para>
974 </sect1>
977 <para>
978 Our example handler is for an ISA bus device. If it was PCI you would be
979 able to share the interrupt and would have set SA_SHIRQ to indicate a
980 shared IRQ. We pass the device pointer as the interrupt routine argument. We
981 don't need to since we only support one card but doing this will make it
982 easier to upgrade the driver for multiple devices in the future.
983 </para>
984 <para>
985 Our interrupt routine needs to do little if we assume the card can simply
986 queue one frame to be read after it captures it.
987 </para>
988 <programlisting>
991 static struct wait_queue *capture_wait;
992 static int capture_ready = 0;
994 static void camera_irq(int irq, void *dev_id,
995 struct pt_regs *regs)
996 {
997 capture_ready=1;
998 wake_up_interruptible(&capture_wait);
999 }
1000 </programlisting>
1001 <para>
1002 The interrupt handler is nice and simple for this card as we are assuming
1003 the card is buffering the frame for us. This means we have little to do but
1004 wake up anybody interested. We also set a capture_ready flag, as we may
1005 capture a frame before an application needs it. In this case we need to know
1006 that a frame is ready. If we had to collect the frame on the interrupt life
1007 would be more complex.
1008 </para>
1009 <para>
1010 The two new routines we need to supply are camera_read which returns a
1011 frame, and camera_poll which waits for a frame to become ready.
1012 </para>
1013 <programlisting>
1016 static int camera_poll(struct video_device *dev,
1017 struct file *file, struct poll_table *wait)
1018 {
1019 poll_wait(file, &capture_wait, wait);
1020 if(capture_read)
1021 return POLLIN|POLLRDNORM;
1022 return 0;
1023 }
1025 </programlisting>
1026 <para>
1027 Our wait queue for polling is the capture_wait queue. This will cause the
1028 task to be woken up by our camera_irq routine. We check capture_read to see
1029 if there is an image present and if so report that it is readable.
1030 </para>
1031 </sect1>
1034 <programlisting>
1037 static long camera_read(struct video_device *dev, char *buf,
1038 unsigned long count)
1039 {
1040 struct wait_queue wait = { current, NULL };
1041 u8 *ptr;
1042 int len;
1043 int i;
1047 while(!capture_ready)
1048 {
1050 {
1052 current->state = TASK_RUNNING;
1053 return -EWOULDBLOCK;
1054 }
1055 if(signal_pending(current))
1056 {
1058 current->state = TASK_RUNNING;
1059 return -ERESTARTSYS;
1060 }
1061 schedule();
1062 current->state = TASK_INTERRUPTIBLE;
1063 }
1065 current->state = TASK_RUNNING;
1067 </programlisting>
1068 <para>
1069 The first thing we have to do is to ensure that the application waits until
1070 the next frame is ready. The code here is almost identical to the mouse code
1071 we used earlier in this chapter. It is one of the common building blocks of
1072 Linux device driver code and probably one which you will find occurs in any
1073 drivers you write.
1074 </para>
1075 <para>
1076 We wait for a frame to be ready, or for a signal to interrupt our waiting. If a
1077 signal occurs we need to return from the system call so that the signal can
1078 be sent to the application itself. We also check to see if the user actually
1079 wanted to avoid waiting - ie if they are using non-blocking I/O and have other things
1080 to get on with.
1081 </para>
1082 <para>
1083 Next we copy the data from the card to the user application. This is rarely
1084 as easy as our example makes out. We will add capture_w, and capture_h here
1085 to hold the width and height of the captured image. We assume the card only
1086 supports 24bit RGB for now.
1087 </para>
1088 <programlisting>
1092 capture_ready = 0;
1094 ptr=(u8 *)buf;
1097 if(len>count)
1098 len=count; /* Doesn't all fit */
1101 {
1102 put_user(inb(io+IMAGE_DATA), ptr);
1103 ptr++;
1104 }
1106 hardware_restart_capture();
1108 return i;
1109 }
1111 </programlisting>
1112 <para>
1113 For a real hardware device you would try to avoid the loop with put_user().
1114 Each call to put_user() has a time overhead checking whether the accesses to user
1115 space are allowed. It would be better to read a line into a temporary buffer
1116 then copy this to user space in one go.
1117 </para>
1118 <para>
1119 Having captured the image and put it into user space we can kick the card to
1120 get the next frame acquired.
1121 </para>
1122 </sect1>
1125 <para>
1126 As with the radio driver the major control interface is via the ioctl()
1127 function. Video capture devices support the same tuner calls as a radio
1128 device and also support additional calls to control how the video functions
1129 are handled. In this simple example the card has no tuners to avoid making
1130 the code complex.
1131 </para>
1132 <programlisting>
1136 static int camera_ioctl(struct video_device *dev, unsigned int cmd, void *arg)
1137 {
1138 switch(cmd)
1139 {
1140 case VIDIOCGCAP:
1141 {
1142 struct video_capability v;
1143 v.type = VID_TYPE_CAPTURE|\
1144 VID_TYPE_CHROMAKEY|\
1145 VID_TYPE_SCALES|\
1146 VID_TYPE_OVERLAY;
1147 v.channels = 1;
1148 v.audios = 0;
1149 v.maxwidth = 640;
1150 v.minwidth = 16;
1151 v.maxheight = 480;
1152 v.minheight = 16;
1153 strcpy(v.name, "My Camera");
1154 if(copy_to_user(arg, &v, sizeof(v)))
1155 return -EFAULT;
1156 return 0;
1157 }
1160 </programlisting>
1161 <para>
1162 The first ioctl we must support and which all video capture and radio
1163 devices are required to support is VIDIOCGCAP. This behaves exactly the same
1164 as with a radio device. This time, however, we report the extra capabilities
1165 we outlined earlier on when defining our video_dev structure.
1166 </para>
1167 <para>
1168 We now set the video flags saying that we support overlay, capture,
1169 scaling and chromakey. We also report size limits - our smallest image is
1171 </para>
1172 <para>
1173 To keep things simple we report no audio and no tuning capabilities at all.
1174 </para>
1175 <programlisting>
1177 case VIDIOCGCHAN:
1178 {
1179 struct video_channel v;
1180 if(copy_from_user(&v, arg, sizeof(v)))
1181 return -EFAULT;
1182 if(v.channel != 0)
1183 return -EINVAL;
1184 v.flags = 0;
1185 v.tuners = 0;
1186 v.type = VIDEO_TYPE_CAMERA;
1187 v.norm = VIDEO_MODE_AUTO;
1188 strcpy(v.name, "Camera Input");break;
1189 if(copy_to_user(&v, arg, sizeof(v)))
1190 return -EFAULT;
1191 return 0;
1192 }
1195 </programlisting>
1196 <para>
1197 This follows what is very much the standard way an ioctl handler looks
1198 in Linux. We copy the data into a kernel space variable and we check that the
1199 request is valid (in this case that the input is 0). Finally we copy the
1200 camera info back to the user.
1201 </para>
1202 <para>
1203 The VIDIOCGCHAN ioctl allows a user to ask about video channels (that is
1204 inputs to the video card). Our example card has a single camera input. The
1205 fields in the structure are
1206 </para>
1209 <tbody>
1210 <row>
1213 </row><row>
1215 to describe the port to the user.
1216 Appropriate names are therefore things like
1218 </row><row>
1220 </row><row>
1222 </row><row>
1224 if relevant for this channel.
1225 </entry>
1226 </row>
1227 </tbody>
1228 </tgroup>
1229 </table>
1232 <tbody>
1233 <row>
1235 </row><row>
1237 </row>
1238 </tbody>
1239 </tgroup>
1240 </table>
1243 <tbody>
1244 <row>
1246 </row><row>
1248 </row><row>
1250 </row>
1251 </tbody>
1252 </tgroup>
1253 </table>
1256 <tbody>
1257 <row>
1259 </row><row>
1261 </row><row>
1263 </row><row>
1265 matter</entry>
1266 </row>
1267 </tbody>
1268 </tgroup>
1269 </table>
1270 <para>
1271 The corresponding VIDIOCSCHAN ioctl allows a user to change channel and to
1272 request the norm is changed - for example to switch between a PAL or an NTSC
1273 format camera.
1274 </para>
1275 <programlisting>
1278 case VIDIOCSCHAN:
1279 {
1280 struct video_channel v;
1281 if(copy_from_user(&v, arg, sizeof(v)))
1282 return -EFAULT;
1283 if(v.channel != 0)
1284 return -EINVAL;
1285 if(v.norm != VIDEO_MODE_AUTO)
1286 return -EINVAL;
1287 return 0;
1288 }
1291 </programlisting>
1292 <para>
1293 The implementation of this call in our driver is remarkably easy. Because we
1294 are assuming fixed format hardware we need only check that the user has not
1295 tried to change anything.
1296 </para>
1297 <para>
1298 The user also needs to be able to configure and adjust the picture they are
1299 seeing. This is much like adjusting a television set. A user application
1300 also needs to know the palette being used so that it knows how to display
1301 the image that has been captured. The VIDIOCGPICT and VIDIOCSPICT ioctl
1302 calls provide this information.
1303 </para>
1304 <programlisting>
1307 case VIDIOCGPICT
1308 {
1309 struct video_picture v;
1310 v.brightness = hardware_brightness();
1311 v.hue = hardware_hue();
1312 v.colour = hardware_saturation();
1313 v.contrast = hardware_brightness();
1314 /* Not settable */
1315 v.whiteness = 32768;
1317 v.palette = VIDEO_PALETTE_RGB24;
1318 if(copy_to_user(&v, arg,
1319 sizeof(v)))
1320 return -EFAULT;
1321 return 0;
1322 }
1325 </programlisting>
1326 <para>
1327 The brightness, hue, color, and contrast provide the picture controls that
1328 are akin to a conventional television. Whiteness provides additional
1330 and have 32768 as the mid point setting. The scaling means that applications
1331 do not have to worry about the capability range of the hardware but can let
1332 it make a best effort attempt.
1333 </para>
1334 <para>
1335 Our depth is 24, as this is in bits. We will be returning RGB24 format. This
1336 has one byte of red, then one of green, then one of blue. This then repeats
1337 for every other pixel in the image. The other common formats the interface
1338 defines are
1339 </para>
1342 <tbody>
1343 <row>
1345 like</entry>
1346 </row><row>
1349 </row><row>
1351 each occupy five bits.</entry>
1352 </row>
1353 </tbody>
1354 </tgroup>
1355 </table>
1356 <para>
1357 Additional modes are support for YUV capture formats. These are common for
1358 TV and video conferencing applications.
1359 </para>
1360 <para>
1361 The VIDIOCSPICT ioctl allows a user to set some of the picture parameters.
1362 Exactly which ones are supported depends heavily on the card itself. It is
1363 possible to support many modes and effects in software. In general doing
1364 this in the kernel is a bad idea. Video capture is a performance-sensitive
1365 application and the programs can often do better if they aren't being
1366 'helped' by an overkeen driver writer. Thus for our device we will report
1367 RGB24 only and refuse to allow a change.
1368 </para>
1369 <programlisting>
1372 case VIDIOCSPICT:
1373 {
1374 struct video_picture v;
1375 if(copy_from_user(&v, arg, sizeof(v)))
1376 return -EFAULT;
1377 if(v.depth!=24 ||
1378 v.palette != VIDEO_PALETTE_RGB24)
1379 return -EINVAL;
1380 set_hardware_brightness(v.brightness);
1381 set_hardware_hue(v.hue);
1382 set_hardware_saturation(v.colour);
1383 set_hardware_brightness(v.contrast);
1384 return 0;
1385 }
1388 </programlisting>
1389 <para>
1390 We check the user has not tried to change the palette or the depth. We do
1391 not want to carry out some of the changes and then return an error. This may
1392 confuse the application which will be assuming no change occurred.
1393 </para>
1394 <para>
1395 In much the same way as you need to be able to set the picture controls to
1396 get the right capture images, many cards need to know what they are
1397 displaying onto when generating overlay output. In some cases getting this
1398 wrong even makes a nasty mess or may crash the computer. For that reason
1399 the VIDIOCSBUF ioctl used to set up the frame buffer information may well
1400 only be usable by root.
1401 </para>
1402 <para>
1403 We will assume our card is one of the old ISA devices with feature connector
1404 and only supports a couple of standard video modes. Very common for older
1405 cards although the PCI devices are way smarter than this.
1406 </para>
1407 <programlisting>
1410 static struct video_buffer capture_fb;
1412 case VIDIOCGFBUF:
1413 {
1414 if(copy_to_user(arg, &capture_fb,
1415 sizeof(capture_fb)))
1416 return -EFAULT;
1417 return 0;
1419 }
1422 </programlisting>
1423 <para>
1424 We keep the frame buffer information in the format the ioctl uses. This
1425 makes it nice and easy to work with in the ioctl calls.
1426 </para>
1427 <programlisting>
1429 case VIDIOCSFBUF:
1430 {
1431 struct video_buffer v;
1433 if(!capable(CAP_SYS_ADMIN))
1434 return -EPERM;
1436 if(copy_from_user(&v, arg, sizeof(v)))
1437 return -EFAULT;
1439 return -EINVAL;
1443 return -EINVAL;
1445 hardware_set_fb(&v);
1446 return 0;
1447 }
1451 </programlisting>
1452 <para>
1453 The capable() function checks a user has the required capability. The Linux
1454 operating system has a set of about 30 capabilities indicating privileged
1455 access to services. The default set up gives the superuser (uid 0) all of
1456 them and nobody else has any.
1457 </para>
1458 <para>
1459 We check that the user has the SYS_ADMIN capability, that is they are
1460 allowed to operate as the machine administrator. We don't want anyone but
1461 the administrator making a mess of the display.
1462 </para>
1463 <para>
1465 EGA or VGA depths). If the mode is not a standard video mode we reject it as
1466 not supported by our card. If the mode is acceptable we save it so that
1467 VIDIOCFBUF will give the right answer next time it is called. The
1468 hardware_set_fb() function is some undescribed card specific function to
1469 program the card for the desired mode.
1470 </para>
1471 <para>
1472 Before the driver can display an overlay window it needs to know where the
1473 window should be placed, and also how large it should be. If the card
1474 supports clipping it needs to know which rectangles to omit from the
1475 display. The video_window structure is used to describe the way the image
1476 should be displayed.
1477 </para>
1480 <tbody>
1481 <row>
1483 may use a smaller size if this size is not available</entry>
1484 </row><row>
1486 size if this size is not available.</entry>
1487 </row><row>
1489 is in pixels relative to the left hand edge of the
1490 picture. Not all cards can display images aligned on
1491 any pixel boundary. If the position is unsuitable
1492 the card adjusts the image right and reduces the
1493 width.</entry>
1494 </row><row>
1496 is counted in pixels relative to the top edge of the
1497 picture. As with the width if the card cannot
1498 display starting on this line it will adjust the
1499 values.</entry>
1500 </row><row>
1502 chromakey colour if chroma keying is being used. </entry>
1503 </row><row>
1505 over.</entry>
1506 </row><row>
1508 </row>
1509 </tbody>
1510 </tgroup>
1511 </table>
1512 <para>
1513 Each clip is a struct video_clip which has the following fields
1514 </para>
1517 <tbody>
1518 <row>
1520 </row><row>
1522 </row><row>
1524 </row>
1525 </tbody>
1526 </tgroup>
1527 </table>
1528 <para>
1529 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.
1530 This may well mean it has to leave alone. small areas the application wished to be
1531 drawn.
1532 </para>
1533 <para>
1534 Our example card uses chromakey so does not have to address most of the
1535 clipping. We will add a video_window structure to our global variables to
1536 remember our parameters, as we did with the frame buffer.
1537 </para>
1538 <programlisting>
1541 case VIDIOCGWIN:
1542 {
1543 if(copy_to_user(arg, &capture_win,
1544 sizeof(capture_win)))
1545 return -EFAULT;
1546 return 0;
1547 }
1550 case VIDIOCSWIN:
1551 {
1552 struct video_window v;
1553 if(copy_from_user(&v, arg, sizeof(v)))
1554 return -EFAULT;
1556 return -EINVAL;
1558 return -EINVAL;
1559 hardware_set_key(v.chromakey);
1560 hardware_set_window(v);
1562 capture_w = v.width;
1563 capture_h = v.height;
1564 return 0;
1565 }
1568 </programlisting>
1569 <para>
1570 Because we are using Chromakey our setup is fairly simple. Mostly we have to
1571 check the values are sane and load them into the capture card.
1572 </para>
1573 <para>
1574 With all the setup done we can now turn on the actual capture/overlay. This
1575 is done with the VIDIOCCAPTURE ioctl. This takes a single integer argument
1577 </para>
1578 <programlisting>
1581 case VIDIOCCAPTURE:
1582 {
1583 int v;
1584 if(get_user(v, (int *)arg))
1585 return -EFAULT;
1586 if(v==0)
1587 hardware_capture_off();
1588 else
1589 {
1590 if(capture_fb.width == 0
1591 || capture_w == 0)
1592 return -EINVAL;
1593 hardware_capture_on();
1594 }
1595 return 0;
1596 }
1599 </programlisting>
1600 <para>
1601 We grab the flag from user space and either enable or disable according to
1602 its value. There is one small corner case we have to consider here. Suppose
1603 that the capture was requested before the video window or the frame buffer
1604 had been set up. In those cases there will be unconfigured fields in our
1605 card data, as well as unconfigured hardware settings. We check for this case and
1606 return an error if the frame buffer or the capture window width is zero.
1607 </para>
1608 <programlisting>
1611 default:
1612 return -ENOIOCTLCMD;
1613 }
1614 }
1615 </programlisting>
1616 <para>
1618 We don't need to support any other ioctls, so if we get this far, it is time
1619 to tell the video layer that we don't now what the user is talking about.
1620 </para>
1621 </sect1>
1624 <para>
1625 The Video4Linux layer supports additional features, including a high
1626 performance mmap() based capture mode and capturing part of the image.
1627 These features are out of the scope of the book. You should however have enough
1628 example code to implement most simple video4linux devices for radio and TV
1629 cards.
1630 </para>
1631 </sect1>
1632 </chapter>
1635 <para>
1636 <variablelist>
1638 <listitem>
1639 <para>
1640 The driver assumes multiple opens should not be allowed. A driver
1641 can work around this but not cleanly.
1642 </para>
1643 </listitem></varlistentry>
1646 <listitem>
1647 <para>
1648 The existing API poorly reflects compression capable devices. There
1649 are plans afoot to merge V4L, V4L2 and some other ideas into a
1650 better interface.
1651 </para>
1652 </listitem></varlistentry>
1653 </variablelist>
1655 </para>
1656 </chapter>
1660 !Edrivers/media/video/videodev.c
1661 </chapter>
1663 </book>