1=======================
2A Linux CD-ROM standard
3=======================
4
5:Author: David van Leeuwen <david@ElseWare.cistron.nl>
6:Date: 12 March 1999
7:Updated by: Erik Andersen (andersee@debian.org)
8:Updated by: Jens Axboe (axboe@image.dk)
9
10
11Introduction
12============
13
14Linux is probably the Unix-like operating system that supports
15the widest variety of hardware devices. The reasons for this are
16presumably
17
18- The large list of hardware devices available for the many platforms
19  that Linux now supports (i.e., i386-PCs, Sparc Suns, etc.)
20- The open design of the operating system, such that anybody can write a
21  driver for Linux.
22- There is plenty of source code around as examples of how to write a driver.
23
24The openness of Linux, and the many different types of available
25hardware has allowed Linux to support many different hardware devices.
26Unfortunately, the very openness that has allowed Linux to support
27all these different devices has also allowed the behavior of each
28device driver to differ significantly from one device to another.
29This divergence of behavior has been very significant for CD-ROM
30devices; the way a particular drive reacts to a `standard` *ioctl()*
31call varies greatly from one device driver to another. To avoid making
32their drivers totally inconsistent, the writers of Linux CD-ROM
33drivers generally created new device drivers by understanding, copying,
34and then changing an existing one. Unfortunately, this practice did not
35maintain uniform behavior across all the Linux CD-ROM drivers.
36
37This document describes an effort to establish Uniform behavior across
38all the different CD-ROM device drivers for Linux. This document also
39defines the various *ioctl()'s*, and how the low-level CD-ROM device
40drivers should implement them. Currently (as of the Linux 2.1.\ *x*
41development kernels) several low-level CD-ROM device drivers, including
42both IDE/ATAPI and SCSI, now use this Uniform interface.
43
44When the CD-ROM was developed, the interface between the CD-ROM drive
45and the computer was not specified in the standards. As a result, many
46different CD-ROM interfaces were developed. Some of them had their
47own proprietary design (Sony, Mitsumi, Panasonic, Philips), other
48manufacturers adopted an existing electrical interface and changed
49the functionality (CreativeLabs/SoundBlaster, Teac, Funai) or simply
50adapted their drives to one or more of the already existing electrical
51interfaces (Aztech, Sanyo, Funai, Vertos, Longshine, Optics Storage and
52most of the `NoName` manufacturers). In cases where a new drive really
53brought its own interface or used its own command set and flow control
54scheme, either a separate driver had to be written, or an existing
55driver had to be enhanced. History has delivered us CD-ROM support for
56many of these different interfaces. Nowadays, almost all new CD-ROM
57drives are either IDE/ATAPI or SCSI, and it is very unlikely that any
58manufacturer will create a new interface. Even finding drives for the
59old proprietary interfaces is getting difficult.
60
61When (in the 1.3.70's) I looked at the existing software interface,
62which was expressed through `cdrom.h`, it appeared to be a rather wild
63set of commands and data formats [#f1]_. It seemed that many
64features of the software interface had been added to accommodate the
65capabilities of a particular drive, in an *ad hoc* manner. More
66importantly, it appeared that the behavior of the `standard` commands
67was different for most of the different drivers: e. g., some drivers
68close the tray if an *open()* call occurs when the tray is open, while
69others do not. Some drivers lock the door upon opening the device, to
70prevent an incoherent file system, but others don't, to allow software
71ejection. Undoubtedly, the capabilities of the different drives vary,
72but even when two drives have the same capability their drivers'
73behavior was usually different.
74
75.. [#f1]
76   I cannot recollect what kernel version I looked at, then,
77   presumably 1.2.13 and 1.3.34 --- the latest kernel that I was
78   indirectly involved in.
79
80I decided to start a discussion on how to make all the Linux CD-ROM
81drivers behave more uniformly. I began by contacting the developers of
82the many CD-ROM drivers found in the Linux kernel. Their reactions
83encouraged me to write the Uniform CD-ROM Driver which this document is
84intended to describe. The implementation of the Uniform CD-ROM Driver is
85in the file `cdrom.c`. This driver is intended to be an additional software
86layer that sits on top of the low-level device drivers for each CD-ROM drive.
87By adding this additional layer, it is possible to have all the different
88CD-ROM devices behave **exactly** the same (insofar as the underlying
89hardware will allow).
90
91The goal of the Uniform CD-ROM Driver is **not** to alienate driver developers
92whohave not yet taken steps to support this effort. The goal of Uniform CD-ROM
93Driver is simply to give people writing application programs for CD-ROM drives
94**one** Linux CD-ROM interface with consistent behavior for all
95CD-ROM devices. In addition, this also provides a consistent interface
96between the low-level device driver code and the Linux kernel. Care
97is taken that 100% compatibility exists with the data structures and
98programmer's interface defined in `cdrom.h`. This guide was written to
99help CD-ROM driver developers adapt their code to use the Uniform CD-ROM
100Driver code defined in `cdrom.c`.
101
102Personally, I think that the most important hardware interfaces are
103the IDE/ATAPI drives and, of course, the SCSI drives, but as prices
104of hardware drop continuously, it is also likely that people may have
105more than one CD-ROM drive, possibly of mixed types. It is important
106that these drives behave in the same way. In December 1994, one of the
107cheapest CD-ROM drives was a Philips cm206, a double-speed proprietary
108drive. In the months that I was busy writing a Linux driver for it,
109proprietary drives became obsolete and IDE/ATAPI drives became the
110standard. At the time of the last update to this document (November
1111997) it is becoming difficult to even **find** anything less than a
11216 speed CD-ROM drive, and 24 speed drives are common.
113
114.. _cdrom_api:
115
116Standardizing through another software level
117============================================
118
119At the time this document was conceived, all drivers directly
120implemented the CD-ROM *ioctl()* calls through their own routines. This
121led to the danger of different drivers forgetting to do important things
122like checking that the user was giving the driver valid data. More
123importantly, this led to the divergence of behavior, which has already
124been discussed.
125
126For this reason, the Uniform CD-ROM Driver was created to enforce consistent
127CD-ROM drive behavior, and to provide a common set of services to the various
128low-level CD-ROM device drivers. The Uniform CD-ROM Driver now provides another
129software-level, that separates the *ioctl()* and *open()* implementation
130from the actual hardware implementation. Note that this effort has
131made few changes which will affect a user's application programs. The
132greatest change involved moving the contents of the various low-level
133CD-ROM drivers\' header files to the kernel's cdrom directory. This was
134done to help ensure that the user is only presented with only one cdrom
135interface, the interface defined in `cdrom.h`.
136
137CD-ROM drives are specific enough (i. e., different from other
138block-devices such as floppy or hard disc drives), to define a set
139of common **CD-ROM device operations**, *<cdrom-device>_dops*.
140These operations are different from the classical block-device file
141operations, *<block-device>_fops*.
142
143The routines for the Uniform CD-ROM Driver interface level are implemented
144in the file `cdrom.c`. In this file, the Uniform CD-ROM Driver interfaces
145with the kernel as a block device by registering the following general
146*struct file_operations*::
147
148	struct file_operations cdrom_fops = {
149		NULL,			/* lseek */
150		block _read ,		/* read--general block-dev read */
151		block _write,		/* write--general block-dev write */
152		NULL,			/* readdir */
153		NULL,			/* select */
154		cdrom_ioctl,		/* ioctl */
155		NULL,			/* mmap */
156		cdrom_open,		/* open */
157		cdrom_release,		/* release */
158		NULL,			/* fsync */
159		NULL,			/* fasync */
160		NULL			/* revalidate */
161	};
162
163Every active CD-ROM device shares this *struct*. The routines
164declared above are all implemented in `cdrom.c`, since this file is the
165place where the behavior of all CD-ROM-devices is defined and
166standardized. The actual interface to the various types of CD-ROM
167hardware is still performed by various low-level CD-ROM-device
168drivers. These routines simply implement certain **capabilities**
169that are common to all CD-ROM (and really, all removable-media
170devices).
171
172Registration of a low-level CD-ROM device driver is now done through
173the general routines in `cdrom.c`, not through the Virtual File System
174(VFS) any more. The interface implemented in `cdrom.c` is carried out
175through two general structures that contain information about the
176capabilities of the driver, and the specific drives on which the
177driver operates. The structures are:
178
179cdrom_device_ops
180  This structure contains information about the low-level driver for a
181  CD-ROM device. This structure is conceptually connected to the major
182  number of the device (although some drivers may have different
183  major numbers, as is the case for the IDE driver).
184
185cdrom_device_info
186  This structure contains information about a particular CD-ROM drive,
187  such as its device name, speed, etc. This structure is conceptually
188  connected to the minor number of the device.
189
190Registering a particular CD-ROM drive with the Uniform CD-ROM Driver
191is done by the low-level device driver though a call to::
192
193	register_cdrom(struct cdrom_device_info * <device>_info)
194
195The device information structure, *<device>_info*, contains all the
196information needed for the kernel to interface with the low-level
197CD-ROM device driver. One of the most important entries in this
198structure is a pointer to the *cdrom_device_ops* structure of the
199low-level driver.
200
201The device operations structure, *cdrom_device_ops*, contains a list
202of pointers to the functions which are implemented in the low-level
203device driver. When `cdrom.c` accesses a CD-ROM device, it does it
204through the functions in this structure. It is impossible to know all
205the capabilities of future CD-ROM drives, so it is expected that this
206list may need to be expanded from time to time as new technologies are
207developed. For example, CD-R and CD-R/W drives are beginning to become
208popular, and support will soon need to be added for them. For now, the
209current *struct* is::
210
211	struct cdrom_device_ops {
212		int (*open)(struct cdrom_device_info *, int)
213		void (*release)(struct cdrom_device_info *);
214		int (*drive_status)(struct cdrom_device_info *, int);
215		unsigned int (*check_events)(struct cdrom_device_info *,
216					     unsigned int, int);
217		int (*media_changed)(struct cdrom_device_info *, int);
218		int (*tray_move)(struct cdrom_device_info *, int);
219		int (*lock_door)(struct cdrom_device_info *, int);
220		int (*select_speed)(struct cdrom_device_info *, int);
221		int (*get_last_session) (struct cdrom_device_info *,
222					 struct cdrom_multisession *);
223		int (*get_mcn)(struct cdrom_device_info *, struct cdrom_mcn *);
224		int (*reset)(struct cdrom_device_info *);
225		int (*audio_ioctl)(struct cdrom_device_info *,
226				   unsigned int, void *);
227		const int capability;		/* capability flags */
228		int (*generic_packet)(struct cdrom_device_info *,
229				      struct packet_command *);
230	};
231
232When a low-level device driver implements one of these capabilities,
233it should add a function pointer to this *struct*. When a particular
234function is not implemented, however, this *struct* should contain a
235NULL instead. The *capability* flags specify the capabilities of the
236CD-ROM hardware and/or low-level CD-ROM driver when a CD-ROM drive
237is registered with the Uniform CD-ROM Driver.
238
239Note that most functions have fewer parameters than their
240*blkdev_fops* counterparts. This is because very little of the
241information in the structures *inode* and *file* is used. For most
242drivers, the main parameter is the *struct* *cdrom_device_info*, from
243which the major and minor number can be extracted. (Most low-level
244CD-ROM drivers don't even look at the major and minor number though,
245since many of them only support one device.) This will be available
246through *dev* in *cdrom_device_info* described below.
247
248The drive-specific, minor-like information that is registered with
249`cdrom.c`, currently contains the following fields::
250
251  struct cdrom_device_info {
252	const struct cdrom_device_ops * ops;	/* device operations for this major */
253	struct list_head list;			/* linked list of all device_info */
254	struct gendisk * disk;			/* matching block layer disk */
255	void *  handle;				/* driver-dependent data */
256
257	int mask;				/* mask of capability: disables them */
258	int speed;				/* maximum speed for reading data */
259	int capacity;				/* number of discs in a jukebox */
260
261	unsigned int options:30;		/* options flags */
262	unsigned mc_flags:2;			/*  media-change buffer flags */
263	unsigned int vfs_events;		/*  cached events for vfs path */
264	unsigned int ioctl_events;		/*  cached events for ioctl path */
265	int use_count;				/*  number of times device is opened */
266	char name[20];				/*  name of the device type */
267
268	__u8 sanyo_slot : 2;			/*  Sanyo 3-CD changer support */
269	__u8 keeplocked : 1;			/*  CDROM_LOCKDOOR status */
270	__u8 reserved : 5;			/*  not used yet */
271	int cdda_method;			/*  see CDDA_* flags */
272	__u8 last_sense;			/*  saves last sense key */
273	__u8 media_written;			/*  dirty flag, DVD+RW bookkeeping */
274	unsigned short mmc3_profile;		/*  current MMC3 profile */
275	int for_data;				/*  unknown:TBD */
276	int (*exit)(struct cdrom_device_info *);/*  unknown:TBD */
277	int mrw_mode_page;			/*  which MRW mode page is in use */
278  };
279
280Using this *struct*, a linked list of the registered minor devices is
281built, using the *next* field. The device number, the device operations
282struct and specifications of properties of the drive are stored in this
283structure.
284
285The *mask* flags can be used to mask out some of the capabilities listed
286in *ops->capability*, if a specific drive doesn't support a feature
287of the driver. The value *speed* specifies the maximum head-rate of the
288drive, measured in units of normal audio speed (176kB/sec raw data or
289150kB/sec file system data). The parameters are declared *const*
290because they describe properties of the drive, which don't change after
291registration.
292
293A few registers contain variables local to the CD-ROM drive. The
294flags *options* are used to specify how the general CD-ROM routines
295should behave. These various flags registers should provide enough
296flexibility to adapt to the different users' wishes (and **not** the
297`arbitrary` wishes of the author of the low-level device driver, as is
298the case in the old scheme). The register *mc_flags* is used to buffer
299the information from *media_changed()* to two separate queues. Other
300data that is specific to a minor drive, can be accessed through *handle*,
301which can point to a data structure specific to the low-level driver.
302The fields *use_count*, *next*, *options* and *mc_flags* need not be
303initialized.
304
305The intermediate software layer that `cdrom.c` forms will perform some
306additional bookkeeping. The use count of the device (the number of
307processes that have the device opened) is registered in *use_count*. The
308function *cdrom_ioctl()* will verify the appropriate user-memory regions
309for read and write, and in case a location on the CD is transferred,
310it will `sanitize` the format by making requests to the low-level
311drivers in a standard format, and translating all formats between the
312user-software and low level drivers. This relieves much of the drivers'
313memory checking and format checking and translation. Also, the necessary
314structures will be declared on the program stack.
315
316The implementation of the functions should be as defined in the
317following sections. Two functions **must** be implemented, namely
318*open()* and *release()*. Other functions may be omitted, their
319corresponding capability flags will be cleared upon registration.
320Generally, a function returns zero on success and negative on error. A
321function call should return only after the command has completed, but of
322course waiting for the device should not use processor time.
323
324::
325
326	int open(struct cdrom_device_info *cdi, int purpose)
327
328*Open()* should try to open the device for a specific *purpose*, which
329can be either:
330
331- Open for reading data, as done by `mount()` (2), or the
332  user commands `dd` or `cat`.
333- Open for *ioctl* commands, as done by audio-CD playing programs.
334
335Notice that any strategic code (closing tray upon *open()*, etc.) is
336done by the calling routine in `cdrom.c`, so the low-level routine
337should only be concerned with proper initialization, such as spinning
338up the disc, etc.
339
340::
341
342	void release(struct cdrom_device_info *cdi)
343
344Device-specific actions should be taken such as spinning down the device.
345However, strategic actions such as ejection of the tray, or unlocking
346the door, should be left over to the general routine *cdrom_release()*.
347This is the only function returning type *void*.
348
349.. _cdrom_drive_status:
350
351::
352
353	int drive_status(struct cdrom_device_info *cdi, int slot_nr)
354
355The function *drive_status*, if implemented, should provide
356information on the status of the drive (not the status of the disc,
357which may or may not be in the drive). If the drive is not a changer,
358*slot_nr* should be ignored. In `cdrom.h` the possibilities are listed::
359
360
361	CDS_NO_INFO		/* no information available */
362	CDS_NO_DISC		/* no disc is inserted, tray is closed */
363	CDS_TRAY_OPEN		/* tray is opened */
364	CDS_DRIVE_NOT_READY	/* something is wrong, tray is moving? */
365	CDS_DISC_OK		/* a disc is loaded and everything is fine */
366
367::
368
369	int tray_move(struct cdrom_device_info *cdi, int position)
370
371This function, if implemented, should control the tray movement. (No
372other function should control this.) The parameter *position* controls
373the desired direction of movement:
374
375- 0 Close tray
376- 1 Open tray
377
378This function returns 0 upon success, and a non-zero value upon
379error. Note that if the tray is already in the desired position, no
380action need be taken, and the return value should be 0.
381
382::
383
384	int lock_door(struct cdrom_device_info *cdi, int lock)
385
386This function (and no other code) controls locking of the door, if the
387drive allows this. The value of *lock* controls the desired locking
388state:
389
390- 0 Unlock door, manual opening is allowed
391- 1 Lock door, tray cannot be ejected manually
392
393This function returns 0 upon success, and a non-zero value upon
394error. Note that if the door is already in the requested state, no
395action need be taken, and the return value should be 0.
396
397::
398
399	int select_speed(struct cdrom_device_info *cdi, int speed)
400
401Some CD-ROM drives are capable of changing their head-speed. There
402are several reasons for changing the speed of a CD-ROM drive. Badly
403pressed CD-ROM s may benefit from less-than-maximum head rate. Modern
404CD-ROM drives can obtain very high head rates (up to *24x* is
405common). It has been reported that these drives can make reading
406errors at these high speeds, reducing the speed can prevent data loss
407in these circumstances. Finally, some of these drives can
408make an annoyingly loud noise, which a lower speed may reduce.
409
410This function specifies the speed at which data is read or audio is
411played back. The value of *speed* specifies the head-speed of the
412drive, measured in units of standard cdrom speed (176kB/sec raw data
413or 150kB/sec file system data). So to request that a CD-ROM drive
414operate at 300kB/sec you would call the CDROM_SELECT_SPEED *ioctl*
415with *speed=2*. The special value `0` means `auto-selection`, i. e.,
416maximum data-rate or real-time audio rate. If the drive doesn't have
417this `auto-selection` capability, the decision should be made on the
418current disc loaded and the return value should be positive. A negative
419return value indicates an error.
420
421::
422
423	int get_last_session(struct cdrom_device_info *cdi,
424			     struct cdrom_multisession *ms_info)
425
426This function should implement the old corresponding *ioctl()*. For
427device *cdi->dev*, the start of the last session of the current disc
428should be returned in the pointer argument *ms_info*. Note that
429routines in `cdrom.c` have sanitized this argument: its requested
430format will **always** be of the type *CDROM_LBA* (linear block
431addressing mode), whatever the calling software requested. But
432sanitization goes even further: the low-level implementation may
433return the requested information in *CDROM_MSF* format if it wishes so
434(setting the *ms_info->addr_format* field appropriately, of
435course) and the routines in `cdrom.c` will make the transformation if
436necessary. The return value is 0 upon success.
437
438::
439
440	int get_mcn(struct cdrom_device_info *cdi,
441		    struct cdrom_mcn *mcn)
442
443Some discs carry a `Media Catalog Number` (MCN), also called
444`Universal Product Code` (UPC). This number should reflect the number
445that is generally found in the bar-code on the product. Unfortunately,
446the few discs that carry such a number on the disc don't even use the
447same format. The return argument to this function is a pointer to a
448pre-declared memory region of type *struct cdrom_mcn*. The MCN is
449expected as a 13-character string, terminated by a null-character.
450
451::
452
453	int reset(struct cdrom_device_info *cdi)
454
455This call should perform a hard-reset on the drive (although in
456circumstances that a hard-reset is necessary, a drive may very well not
457listen to commands anymore). Preferably, control is returned to the
458caller only after the drive has finished resetting. If the drive is no
459longer listening, it may be wise for the underlying low-level cdrom
460driver to time out.
461
462::
463
464	int audio_ioctl(struct cdrom_device_info *cdi,
465			unsigned int cmd, void *arg)
466
467Some of the CD-ROM-\ *ioctl()*\ 's defined in `cdrom.h` can be
468implemented by the routines described above, and hence the function
469*cdrom_ioctl* will use those. However, most *ioctl()*\ 's deal with
470audio-control. We have decided to leave these to be accessed through a
471single function, repeating the arguments *cmd* and *arg*. Note that
472the latter is of type *void*, rather than *unsigned long int*.
473The routine *cdrom_ioctl()* does do some useful things,
474though. It sanitizes the address format type to *CDROM_MSF* (Minutes,
475Seconds, Frames) for all audio calls. It also verifies the memory
476location of *arg*, and reserves stack-memory for the argument. This
477makes implementation of the *audio_ioctl()* much simpler than in the
478old driver scheme. For example, you may look up the function
479*cm206_audio_ioctl()* `cm206.c` that should be updated with
480this documentation.
481
482An unimplemented ioctl should return *-ENOSYS*, but a harmless request
483(e. g., *CDROMSTART*) may be ignored by returning 0 (success). Other
484errors should be according to the standards, whatever they are. When
485an error is returned by the low-level driver, the Uniform CD-ROM Driver
486tries whenever possible to return the error code to the calling program.
487(We may decide to sanitize the return value in *cdrom_ioctl()* though, in
488order to guarantee a uniform interface to the audio-player software.)
489
490::
491
492	int dev_ioctl(struct cdrom_device_info *cdi,
493		      unsigned int cmd, unsigned long arg)
494
495Some *ioctl()'s* seem to be specific to certain CD-ROM drives. That is,
496they are introduced to service some capabilities of certain drives. In
497fact, there are 6 different *ioctl()'s* for reading data, either in some
498particular kind of format, or audio data. Not many drives support
499reading audio tracks as data, I believe this is because of protection
500of copyrights of artists. Moreover, I think that if audio-tracks are
501supported, it should be done through the VFS and not via *ioctl()'s*. A
502problem here could be the fact that audio-frames are 2352 bytes long,
503so either the audio-file-system should ask for 75264 bytes at once
504(the least common multiple of 512 and 2352), or the drivers should
505bend their backs to cope with this incoherence (to which I would be
506opposed). Furthermore, it is very difficult for the hardware to find
507the exact frame boundaries, since there are no synchronization headers
508in audio frames. Once these issues are resolved, this code should be
509standardized in `cdrom.c`.
510
511Because there are so many *ioctl()'s* that seem to be introduced to
512satisfy certain drivers [#f2]_, any non-standard *ioctl()*\ s
513are routed through the call *dev_ioctl()*. In principle, `private`
514*ioctl()*\ 's should be numbered after the device's major number, and not
515the general CD-ROM *ioctl* number, `0x53`. Currently the
516non-supported *ioctl()'s* are:
517
518	CDROMREADMODE1, CDROMREADMODE2, CDROMREADAUDIO, CDROMREADRAW,
519	CDROMREADCOOKED, CDROMSEEK, CDROMPLAY-BLK and CDROM-READALL
520
521.. [#f2]
522
523   Is there software around that actually uses these? I'd be interested!
524
525.. _cdrom_capabilities:
526
527CD-ROM capabilities
528-------------------
529
530Instead of just implementing some *ioctl* calls, the interface in
531`cdrom.c` supplies the possibility to indicate the **capabilities**
532of a CD-ROM drive. This can be done by ORing any number of
533capability-constants that are defined in `cdrom.h` at the registration
534phase. Currently, the capabilities are any of::
535
536	CDC_CLOSE_TRAY		/* can close tray by software control */
537	CDC_OPEN_TRAY		/* can open tray */
538	CDC_LOCK		/* can lock and unlock the door */
539	CDC_SELECT_SPEED	/* can select speed, in units of * sim*150 ,kB/s */
540	CDC_SELECT_DISC		/* drive is juke-box */
541	CDC_MULTI_SESSION	/* can read sessions *> rm1* */
542	CDC_MCN			/* can read Media Catalog Number */
543	CDC_MEDIA_CHANGED	/* can report if disc has changed */
544	CDC_PLAY_AUDIO		/* can perform audio-functions (play, pause, etc) */
545	CDC_RESET		/* hard reset device */
546	CDC_IOCTLS		/* driver has non-standard ioctls */
547	CDC_DRIVE_STATUS	/* driver implements drive status */
548
549The capability flag is declared *const*, to prevent drivers from
550accidentally tampering with the contents. The capability flags actually
551inform `cdrom.c` of what the driver can do. If the drive found
552by the driver does not have the capability, is can be masked out by
553the *cdrom_device_info* variable *mask*. For instance, the SCSI CD-ROM
554driver has implemented the code for loading and ejecting CD-ROM's, and
555hence its corresponding flags in *capability* will be set. But a SCSI
556CD-ROM drive might be a caddy system, which can't load the tray, and
557hence for this drive the *cdrom_device_info* struct will have set
558the *CDC_CLOSE_TRAY* bit in *mask*.
559
560In the file `cdrom.c` you will encounter many constructions of the type::
561
562	if (cdo->capability & ~cdi->mask & CDC _<capability>) ...
563
564There is no *ioctl* to set the mask... The reason is that
565I think it is better to control the **behavior** rather than the
566**capabilities**.
567
568Options
569-------
570
571A final flag register controls the **behavior** of the CD-ROM
572drives, in order to satisfy different users' wishes, hopefully
573independently of the ideas of the respective author who happened to
574have made the drive's support available to the Linux community. The
575current behavior options are::
576
577	CDO_AUTO_CLOSE	/* try to close tray upon device open() */
578	CDO_AUTO_EJECT	/* try to open tray on last device close() */
579	CDO_USE_FFLAGS	/* use file_pointer->f_flags to indicate purpose for open() */
580	CDO_LOCK	/* try to lock door if device is opened */
581	CDO_CHECK_TYPE	/* ensure disc type is data if opened for data */
582
583The initial value of this register is
584`CDO_AUTO_CLOSE | CDO_USE_FFLAGS | CDO_LOCK`, reflecting my own view on user
585interface and software standards. Before you protest, there are two
586new *ioctl()'s* implemented in `cdrom.c`, that allow you to control the
587behavior by software. These are::
588
589	CDROM_SET_OPTIONS	/* set options specified in (int)arg */
590	CDROM_CLEAR_OPTIONS	/* clear options specified in (int)arg */
591
592One option needs some more explanation: *CDO_USE_FFLAGS*. In the next
593newsection we explain what the need for this option is.
594
595A software package `setcd`, available from the Debian distribution
596and `sunsite.unc.edu`, allows user level control of these flags.
597
598
599The need to know the purpose of opening the CD-ROM device
600=========================================================
601
602Traditionally, Unix devices can be used in two different `modes`,
603either by reading/writing to the device file, or by issuing
604controlling commands to the device, by the device's *ioctl()*
605call. The problem with CD-ROM drives, is that they can be used for
606two entirely different purposes. One is to mount removable
607file systems, CD-ROM's, the other is to play audio CD's. Audio commands
608are implemented entirely through *ioctl()\'s*, presumably because the
609first implementation (SUN?) has been such. In principle there is
610nothing wrong with this, but a good control of the `CD player` demands
611that the device can **always** be opened in order to give the
612*ioctl* commands, regardless of the state the drive is in.
613
614On the other hand, when used as a removable-media disc drive (what the
615original purpose of CD-ROM s is) we would like to make sure that the
616disc drive is ready for operation upon opening the device. In the old
617scheme, some CD-ROM drivers don't do any integrity checking, resulting
618in a number of i/o errors reported by the VFS to the kernel when an
619attempt for mounting a CD-ROM on an empty drive occurs. This is not a
620particularly elegant way to find out that there is no CD-ROM inserted;
621it more-or-less looks like the old IBM-PC trying to read an empty floppy
622drive for a couple of seconds, after which the system complains it
623can't read from it. Nowadays we can **sense** the existence of a
624removable medium in a drive, and we believe we should exploit that
625fact. An integrity check on opening of the device, that verifies the
626availability of a CD-ROM and its correct type (data), would be
627desirable.
628
629These two ways of using a CD-ROM drive, principally for data and
630secondarily for playing audio discs, have different demands for the
631behavior of the *open()* call. Audio use simply wants to open the
632device in order to get a file handle which is needed for issuing
633*ioctl* commands, while data use wants to open for correct and
634reliable data transfer. The only way user programs can indicate what
635their *purpose* of opening the device is, is through the *flags*
636parameter (see `open(2)`). For CD-ROM devices, these flags aren't
637implemented (some drivers implement checking for write-related flags,
638but this is not strictly necessary if the device file has correct
639permission flags). Most option flags simply don't make sense to
640CD-ROM devices: *O_CREAT*, *O_NOCTTY*, *O_TRUNC*, *O_APPEND*, and
641*O_SYNC* have no meaning to a CD-ROM.
642
643We therefore propose to use the flag *O_NONBLOCK* to indicate
644that the device is opened just for issuing *ioctl*
645commands. Strictly, the meaning of *O_NONBLOCK* is that opening and
646subsequent calls to the device don't cause the calling process to
647wait. We could interpret this as don't wait until someone has
648inserted some valid data-CD-ROM. Thus, our proposal of the
649implementation for the *open()* call for CD-ROM s is:
650
651- If no other flags are set than *O_RDONLY*, the device is opened
652  for data transfer, and the return value will be 0 only upon successful
653  initialization of the transfer. The call may even induce some actions
654  on the CD-ROM, such as closing the tray.
655- If the option flag *O_NONBLOCK* is set, opening will always be
656  successful, unless the whole device doesn't exist. The drive will take
657  no actions whatsoever.
658
659And what about standards?
660-------------------------
661
662You might hesitate to accept this proposal as it comes from the
663Linux community, and not from some standardizing institute. What
664about SUN, SGI, HP and all those other Unix and hardware vendors?
665Well, these companies are in the lucky position that they generally
666control both the hardware and software of their supported products,
667and are large enough to set their own standard. They do not have to
668deal with a dozen or more different, competing hardware
669configurations\ [#f3]_.
670
671.. [#f3]
672
673   Incidentally, I think that SUN's approach to mounting CD-ROM s is very
674   good in origin: under Solaris a volume-daemon automatically mounts a
675   newly inserted CD-ROM under `/cdrom/*<volume-name>*`.
676
677   In my opinion they should have pushed this
678   further and have **every** CD-ROM on the local area network be
679   mounted at the similar location, i. e., no matter in which particular
680   machine you insert a CD-ROM, it will always appear at the same
681   position in the directory tree, on every system. When I wanted to
682   implement such a user-program for Linux, I came across the
683   differences in behavior of the various drivers, and the need for an
684   *ioctl* informing about media changes.
685
686We believe that using *O_NONBLOCK* to indicate that a device is being opened
687for *ioctl* commands only can be easily introduced in the Linux
688community. All the CD-player authors will have to be informed, we can
689even send in our own patches to the programs. The use of *O_NONBLOCK*
690has most likely no influence on the behavior of the CD-players on
691other operating systems than Linux. Finally, a user can always revert
692to old behavior by a call to
693*ioctl(file_descriptor, CDROM_CLEAR_OPTIONS, CDO_USE_FFLAGS)*.
694
695The preferred strategy of *open()*
696----------------------------------
697
698The routines in `cdrom.c` are designed in such a way that run-time
699configuration of the behavior of CD-ROM devices (of **any** type)
700can be carried out, by the *CDROM_SET/CLEAR_OPTIONS* *ioctls*. Thus, various
701modes of operation can be set:
702
703`CDO_AUTO_CLOSE | CDO_USE_FFLAGS | CDO_LOCK`
704   This is the default setting. (With *CDO_CHECK_TYPE* it will be better, in
705   the future.) If the device is not yet opened by any other process, and if
706   the device is being opened for data (*O_NONBLOCK* is not set) and the
707   tray is found to be open, an attempt to close the tray is made. Then,
708   it is verified that a disc is in the drive and, if *CDO_CHECK_TYPE* is
709   set, that it contains tracks of type `data mode 1`. Only if all tests
710   are passed is the return value zero. The door is locked to prevent file
711   system corruption. If the drive is opened for audio (*O_NONBLOCK* is
712   set), no actions are taken and a value of 0 will be returned.
713
714`CDO_AUTO_CLOSE | CDO_AUTO_EJECT | CDO_LOCK`
715   This mimics the behavior of the current sbpcd-driver. The option flags are
716   ignored, the tray is closed on the first open, if necessary. Similarly,
717   the tray is opened on the last release, i. e., if a CD-ROM is unmounted,
718   it is automatically ejected, such that the user can replace it.
719
720We hope that these option can convince everybody (both driver
721maintainers and user program developers) to adopt the new CD-ROM
722driver scheme and option flag interpretation.
723
724Description of routines in `cdrom.c`
725====================================
726
727Only a few routines in `cdrom.c` are exported to the drivers. In this
728new section we will discuss these, as well as the functions that `take
729over` the CD-ROM interface to the kernel. The header file belonging
730to `cdrom.c` is called `cdrom.h`. Formerly, some of the contents of this
731file were placed in the file `ucdrom.h`, but this file has now been
732merged back into `cdrom.h`.
733
734::
735
736	struct file_operations cdrom_fops
737
738The contents of this structure were described in cdrom_api_.
739A pointer to this structure is assigned to the *fops* field
740of the *struct gendisk*.
741
742::
743
744	int register_cdrom(struct cdrom_device_info *cdi)
745
746This function is used in about the same way one registers *cdrom_fops*
747with the kernel, the device operations and information structures,
748as described in cdrom_api_, should be registered with the
749Uniform CD-ROM Driver::
750
751	register_cdrom(&<device>_info);
752
753
754This function returns zero upon success, and non-zero upon
755failure. The structure *<device>_info* should have a pointer to the
756driver's *<device>_dops*, as in::
757
758	struct cdrom_device_info <device>_info = {
759		<device>_dops;
760		...
761	}
762
763Note that a driver must have one static structure, *<device>_dops*, while
764it may have as many structures *<device>_info* as there are minor devices
765active. *Register_cdrom()* builds a linked list from these.
766
767
768::
769
770	void unregister_cdrom(struct cdrom_device_info *cdi)
771
772Unregistering device *cdi* with minor number *MINOR(cdi->dev)* removes
773the minor device from the list. If it was the last registered minor for
774the low-level driver, this disconnects the registered device-operation
775routines from the CD-ROM interface. This function returns zero upon
776success, and non-zero upon failure.
777
778::
779
780	int cdrom_open(struct inode * ip, struct file * fp)
781
782This function is not called directly by the low-level drivers, it is
783listed in the standard *cdrom_fops*. If the VFS opens a file, this
784function becomes active. A strategy is implemented in this routine,
785taking care of all capabilities and options that are set in the
786*cdrom_device_ops* connected to the device. Then, the program flow is
787transferred to the device_dependent *open()* call.
788
789::
790
791	void cdrom_release(struct inode *ip, struct file *fp)
792
793This function implements the reverse-logic of *cdrom_open()*, and then
794calls the device-dependent *release()* routine. When the use-count has
795reached 0, the allocated buffers are flushed by calls to *sync_dev(dev)*
796and *invalidate_buffers(dev)*.
797
798
799.. _cdrom_ioctl:
800
801::
802
803	int cdrom_ioctl(struct inode *ip, struct file *fp,
804			unsigned int cmd, unsigned long arg)
805
806This function handles all the standard *ioctl* requests for CD-ROM
807devices in a uniform way. The different calls fall into three
808categories: *ioctl()'s* that can be directly implemented by device
809operations, ones that are routed through the call *audio_ioctl()*, and
810the remaining ones, that are presumable device-dependent. Generally, a
811negative return value indicates an error.
812
813Directly implemented *ioctl()'s*
814--------------------------------
815
816The following `old` CD-ROM *ioctl()*\ 's are implemented by directly
817calling device-operations in *cdrom_device_ops*, if implemented and
818not masked:
819
820`CDROMMULTISESSION`
821	Requests the last session on a CD-ROM.
822`CDROMEJECT`
823	Open tray.
824`CDROMCLOSETRAY`
825	Close tray.
826`CDROMEJECT_SW`
827	If *arg\not=0*, set behavior to auto-close (close
828	tray on first open) and auto-eject (eject on last release), otherwise
829	set behavior to non-moving on *open()* and *release()* calls.
830`CDROM_GET_MCN`
831	Get the Media Catalog Number from a CD.
832
833*Ioctl*s routed through *audio_ioctl()*
834---------------------------------------
835
836The following set of *ioctl()'s* are all implemented through a call to
837the *cdrom_fops* function *audio_ioctl()*. Memory checks and
838allocation are performed in *cdrom_ioctl()*, and also sanitization of
839address format (*CDROM_LBA*/*CDROM_MSF*) is done.
840
841`CDROMSUBCHNL`
842	Get sub-channel data in argument *arg* of type
843	`struct cdrom_subchnl *`.
844`CDROMREADTOCHDR`
845	Read Table of Contents header, in *arg* of type
846	`struct cdrom_tochdr *`.
847`CDROMREADTOCENTRY`
848	Read a Table of Contents entry in *arg* and specified by *arg*
849	of type `struct cdrom_tocentry *`.
850`CDROMPLAYMSF`
851	Play audio fragment specified in Minute, Second, Frame format,
852	delimited by *arg* of type `struct cdrom_msf *`.
853`CDROMPLAYTRKIND`
854	Play audio fragment in track-index format delimited by *arg*
855	of type `struct cdrom_ti *`.
856`CDROMVOLCTRL`
857	Set volume specified by *arg* of type `struct cdrom_volctrl *`.
858`CDROMVOLREAD`
859	Read volume into by *arg* of type `struct cdrom_volctrl *`.
860`CDROMSTART`
861	Spin up disc.
862`CDROMSTOP`
863	Stop playback of audio fragment.
864`CDROMPAUSE`
865	Pause playback of audio fragment.
866`CDROMRESUME`
867	Resume playing.
868
869New *ioctl()'s* in `cdrom.c`
870----------------------------
871
872The following *ioctl()'s* have been introduced to allow user programs to
873control the behavior of individual CD-ROM devices. New *ioctl*
874commands can be identified by the underscores in their names.
875
876`CDROM_SET_OPTIONS`
877	Set options specified by *arg*. Returns the option flag register
878	after modification. Use *arg = \rm0* for reading the current flags.
879`CDROM_CLEAR_OPTIONS`
880	Clear options specified by *arg*. Returns the option flag register
881	after modification.
882`CDROM_SELECT_SPEED`
883	Select head-rate speed of disc specified as by *arg* in units
884	of standard cdrom speed (176\,kB/sec raw data or
885	150kB/sec file system data). The value 0 means `auto-select`,
886	i. e., play audio discs at real time and data discs at maximum speed.
887	The value *arg* is checked against the maximum head rate of the
888	drive found in the *cdrom_dops*.
889`CDROM_SELECT_DISC`
890	Select disc numbered *arg* from a juke-box.
891
892	First disc is numbered 0. The number *arg* is checked against the
893	maximum number of discs in the juke-box found in the *cdrom_dops*.
894`CDROM_MEDIA_CHANGED`
895	Returns 1 if a disc has been changed since the last call.
896	For juke-boxes, an extra argument *arg*
897	specifies the slot for which the information is given. The special
898	value *CDSL_CURRENT* requests that information about the currently
899	selected slot be returned.
900`CDROM_TIMED_MEDIA_CHANGE`
901	Checks whether the disc has been changed since a user supplied time
902	and returns the time of the last disc change.
903
904	*arg* is a pointer to a *cdrom_timed_media_change_info* struct.
905	*arg->last_media_change* may be set by calling code to signal
906	the timestamp of the last known media change (by the caller).
907	Upon successful return, this ioctl call will set
908	*arg->last_media_change* to the latest media change timestamp (in ms)
909	known by the kernel/driver and set *arg->has_changed* to 1 if
910	that timestamp is more recent than the timestamp set by the caller.
911`CDROM_DRIVE_STATUS`
912	Returns the status of the drive by a call to
913	*drive_status()*. Return values are defined in cdrom_drive_status_.
914	Note that this call doesn't return information on the
915	current playing activity of the drive; this can be polled through
916	an *ioctl* call to *CDROMSUBCHNL*. For juke-boxes, an extra argument
917	*arg* specifies the slot for which (possibly limited) information is
918	given. The special value *CDSL_CURRENT* requests that information
919	about the currently selected slot be returned.
920`CDROM_DISC_STATUS`
921	Returns the type of the disc currently in the drive.
922	It should be viewed as a complement to *CDROM_DRIVE_STATUS*.
923	This *ioctl* can provide *some* information about the current
924	disc that is inserted in the drive. This functionality used to be
925	implemented in the low level drivers, but is now carried out
926	entirely in Uniform CD-ROM Driver.
927
928	The history of development of the CD's use as a carrier medium for
929	various digital information has lead to many different disc types.
930	This *ioctl* is useful only in the case that CDs have \emph {only
931	one} type of data on them. While this is often the case, it is
932	also very common for CDs to have some tracks with data, and some
933	tracks with audio. Because this is an existing interface, rather
934	than fixing this interface by changing the assumptions it was made
935	under, thereby breaking all user applications that use this
936	function, the Uniform CD-ROM Driver implements this *ioctl* as
937	follows: If the CD in question has audio tracks on it, and it has
938	absolutely no CD-I, XA, or data tracks on it, it will be reported
939	as *CDS_AUDIO*. If it has both audio and data tracks, it will
940	return *CDS_MIXED*. If there are no audio tracks on the disc, and
941	if the CD in question has any CD-I tracks on it, it will be
942	reported as *CDS_XA_2_2*. Failing that, if the CD in question
943	has any XA tracks on it, it will be reported as *CDS_XA_2_1*.
944	Finally, if the CD in question has any data tracks on it,
945	it will be reported as a data CD (*CDS_DATA_1*).
946
947	This *ioctl* can return::
948
949		CDS_NO_INFO	/* no information available */
950		CDS_NO_DISC	/* no disc is inserted, or tray is opened */
951		CDS_AUDIO	/* Audio disc (2352 audio bytes/frame) */
952		CDS_DATA_1	/* data disc, mode 1 (2048 user bytes/frame) */
953		CDS_XA_2_1	/* mixed data (XA), mode 2, form 1 (2048 user bytes) */
954		CDS_XA_2_2	/* mixed data (XA), mode 2, form 1 (2324 user bytes) */
955		CDS_MIXED	/* mixed audio/data disc */
956
957	For some information concerning frame layout of the various disc
958	types, see a recent version of `cdrom.h`.
959
960`CDROM_CHANGER_NSLOTS`
961	Returns the number of slots in a juke-box.
962`CDROMRESET`
963	Reset the drive.
964`CDROM_GET_CAPABILITY`
965	Returns the *capability* flags for the drive. Refer to section
966	cdrom_capabilities_ for more information on these flags.
967`CDROM_LOCKDOOR`
968	 Locks the door of the drive. `arg == 0` unlocks the door,
969	 any other value locks it.
970`CDROM_DEBUG`
971	 Turns on debugging info. Only root is allowed to do this.
972	 Same semantics as CDROM_LOCKDOOR.
973
974
975Device dependent *ioctl()'s*
976----------------------------
977
978Finally, all other *ioctl()'s* are passed to the function *dev_ioctl()*,
979if implemented. No memory allocation or verification is carried out.
980
981How to update your driver
982=========================
983
984- Make a backup of your current driver.
985- Get hold of the files `cdrom.c` and `cdrom.h`, they should be in
986  the directory tree that came with this documentation.
987- Make sure you include `cdrom.h`.
988- Change the 3rd argument of *register_blkdev* from `&<your-drive>_fops`
989  to `&cdrom_fops`.
990- Just after that line, add the following to register with the Uniform
991  CD-ROM Driver::
992
993	register_cdrom(&<your-drive>_info);*
994
995  Similarly, add a call to *unregister_cdrom()* at the appropriate place.
996- Copy an example of the device-operations *struct* to your
997  source, e. g., from `cm206.c` *cm206_dops*, and change all
998  entries to names corresponding to your driver, or names you just
999  happen to like. If your driver doesn't support a certain function,
1000  make the entry *NULL*. At the entry *capability* you should list all
1001  capabilities your driver currently supports. If your driver
1002  has a capability that is not listed, please send me a message.
1003- Copy the *cdrom_device_info* declaration from the same example
1004  driver, and modify the entries according to your needs. If your
1005  driver dynamically determines the capabilities of the hardware, this
1006  structure should also be declared dynamically.
1007- Implement all functions in your `<device>_dops` structure,
1008  according to prototypes listed in  `cdrom.h`, and specifications given
1009  in cdrom_api_. Most likely you have already implemented
1010  the code in a large part, and you will almost certainly need to adapt the
1011  prototype and return values.
1012- Rename your `<device>_ioctl()` function to *audio_ioctl* and
1013  change the prototype a little. Remove entries listed in the first
1014  part in cdrom_ioctl_, if your code was OK, these are
1015  just calls to the routines you adapted in the previous step.
1016- You may remove all remaining memory checking code in the
1017  *audio_ioctl()* function that deals with audio commands (these are
1018  listed in the second part of cdrom_ioctl_. There is no
1019  need for memory allocation either, so most *case*s in the *switch*
1020  statement look similar to::
1021
1022	case CDROMREADTOCENTRY:
1023		get_toc_entry\bigl((struct cdrom_tocentry *) arg);
1024
1025- All remaining *ioctl* cases must be moved to a separate
1026  function, *<device>_ioctl*, the device-dependent *ioctl()'s*. Note that
1027  memory checking and allocation must be kept in this code!
1028- Change the prototypes of *<device>_open()* and
1029  *<device>_release()*, and remove any strategic code (i. e., tray
1030  movement, door locking, etc.).
1031- Try to recompile the drivers. We advise you to use modules, both
1032  for `cdrom.o` and your driver, as debugging is much easier this
1033  way.
1034
1035Thanks
1036======
1037
1038Thanks to all the people involved. First, Erik Andersen, who has
1039taken over the torch in maintaining `cdrom.c` and integrating much
1040CD-ROM-related code in the 2.1-kernel. Thanks to Scott Snyder and
1041Gerd Knorr, who were the first to implement this interface for SCSI
1042and IDE-CD drivers and added many ideas for extension of the data
1043structures relative to kernel~2.0. Further thanks to Heiko Eißfeldt,
1044Thomas Quinot, Jon Tombs, Ken Pizzini, Eberhard Mönkeberg and Andrew Kroll,
1045the Linux CD-ROM device driver developers who were kind
1046enough to give suggestions and criticisms during the writing. Finally
1047of course, I want to thank Linus Torvalds for making this possible in
1048the first place.
1049