| blob | 8f85b0e41046231d5a0a6351bb06e377efb1a617 |
30 \maketitle
34 \linux\ is probably the Unix-like operating system that supports
35 the widest variety of hardware devices. The reasons for this are
36 presumably
38 \item
39 The large list of hardware devices available for the many platforms
41 \item
42 The open design of the operating system, such that anybody can write a
43 driver for \linux.
44 \item
45 There is plenty of source code around as examples of how to write a driver.
47 The openness of \linux, and the many different types of available
48 hardware has allowed \linux\ to support many different hardware devices.
49 Unfortunately, the very openness that has allowed \linux\ to support
50 all these different devices has also allowed the behavior of each
51 device driver to differ significantly from one device to another.
52 This divergence of behavior has been very significant for \cdrom\
53 devices; the way a particular drive reacts to a `standard' $ioctl()$
54 call varies greatly from one device driver to another. To avoid making
56 drivers generally created new device drivers by understanding, copying,
57 and then changing an existing one. Unfortunately, this practice did not
60 This document describes an effort to establish Uniform behavior across
62 defines the various $ioctl$s, and how the low-level \cdrom\ device
64 development kernels) several low-level \cdrom\ device drivers, including
65 both IDE/ATAPI and SCSI, now use this Uniform interface.
68 and the computer was not specified in the standards. As a result, many
69 different \cdrom\ interfaces were developed. Some of them had their
70 own proprietary design (Sony, Mitsumi, Panasonic, Philips), other
71 manufacturers adopted an existing electrical interface and changed
72 the functionality (CreativeLabs/SoundBlaster, Teac, Funai) or simply
73 adapted their drives to one or more of the already existing electrical
74 interfaces (Aztech, Sanyo, Funai, Vertos, Longshine, Optics Storage and
75 most of the `NoName' manufacturers). In cases where a new drive really
76 brought its own interface or used its own command set and flow control
77 scheme, either a separate driver had to be written, or an existing
78 driver had to be enhanced. History has delivered us \cdrom\ support for
79 many of these different interfaces. Nowadays, almost all new \cdrom\
80 drives are either IDE/ATAPI or SCSI, and it is very unlikely that any
81 manufacturer will create a new interface. Even finding drives for the
82 old proprietary interfaces is getting difficult.
84 When (in the 1.3.70's) I looked at the existing software interface,
85 which was expressed through \cdromh, it appeared to be a rather wild
88 latest kernel that I was indirectly involved in.} It seemed that many
89 features of the software interface had been added to accommodate the
91 importantly, it appeared that the behavior of the `standard' commands
92 was different for most of the different drivers: \eg, some drivers
93 close the tray if an $open()$ call occurs when the tray is open, while
94 others do not. Some drivers lock the door upon opening the device, to
95 prevent an incoherent file system, but others don't, to allow software
96 ejection. Undoubtedly, the capabilities of the different drives vary,
97 but even when two drives have the same capability their drivers'
98 behavior was usually different.
101 drivers behave more uniformly. I began by contacting the developers of
105 driver is intended to be an additional software layer that sits on top
106 of the low-level device drivers for each \cdrom\ drive. By adding this
107 additional layer, it is possible to have all the different \cdrom\
109 hardware will allow).
112 have not yet taken steps to support this effort. The goal of \UCD\ is
113 simply to give people writing application programs for \cdrom\ drives
115 \cdrom\ devices. In addition, this also provides a consistent interface
116 between the low-level device driver code and the \linux\ kernel. Care
117 is taken that 100\,\% compatibility exists with the data structures and
118 programmer's interface defined in \cdromh. This guide was written to
120 defined in \cdromc.
122 Personally, I think that the most important hardware interfaces are
123 the IDE/ATAPI drives and, of course, the SCSI drives, but as prices
124 of hardware drop continuously, it is also likely that people may have
125 more than one \cdrom\ drive, possibly of mixed types. It is important
126 that these drives behave in the same way. In December 1994, one of the
127 cheapest \cdrom\ drives was a Philips cm206, a double-speed proprietary
128 drive. In the months that I was busy writing a \linux\ driver for it,
129 proprietary drives became obsolete and IDE/ATAPI drives became the
130 standard. At the time of the last update to this document (November
137 At the time this document was conceived, all drivers directly
138 implemented the \cdrom\ $ioctl()$ calls through their own routines. This
139 led to the danger of different drivers forgetting to do important things
140 like checking that the user was giving the driver valid data. More
141 importantly, this led to the divergence of behavior, which has already
142 been discussed.
145 drive behavior, and to provide a common set of services to the various
147 software-level, that separates the $ioctl()$ and $open()$ implementation
148 from the actual hardware implementation. Note that this effort has
149 made few changes which will affect a user's application programs. The
150 greatest change involved moving the contents of the various low-level
151 \cdrom\ drivers' header files to the kernel's cdrom directory. This was
152 done to help ensure that the user is only presented with only one cdrom
153 interface, the interface defined in \cdromh.
156 block-devices such as floppy or hard disc drives), to define a set
158 These operations are different from the classical block-device file
159 operations, $<block-device>_fops$.
161 The routines for the \UCD\ interface level are implemented in the file
163 device by registering the following general $struct\ file_operations$:
164 $$
167 &NULL, & lseek \cr
168 &block_read, & read---general block-dev read \cr
169 &block_write, & write---general block-dev write \cr
170 &NULL, & readdir \cr
171 &NULL, & select \cr
172 &cdrom_ioctl, & ioctl \cr
173 &NULL, & mmap \cr
174 &cdrom_open, & open \cr
175 &cdrom_release, & release \cr
176 &NULL, & fsync \cr
177 &NULL, & fasync \cr
178 &cdrom_media_changed, & media change \cr
179 &NULL & revalidate \cr
181 }
182 $$
184 Every active \cdrom\ device shares this $struct$. The routines
185 declared above are all implemented in \cdromc, since this file is the
186 place where the behavior of all \cdrom-devices is defined and
187 standardized. The actual interface to the various types of \cdrom\
188 hardware is still performed by various low-level \cdrom-device
190 that are common to all \cdrom\ (and really, all removable-media
191 devices).
193 Registration of a low-level \cdrom\ device driver is now done through
194 the general routines in \cdromc, not through the Virtual File System
195 (VFS) any more. The interface implemented in \cdromc\ is carried out
196 through two general structures that contain information about the
197 capabilities of the driver, and the specific drives on which the
198 driver operates. The structures are:
201 This structure contains information about the low-level driver for a
202 \cdrom\ device. This structure is conceptually connected to the major
203 number of the device (although some drivers may have different
204 major numbers, as is the case for the IDE driver).
206 This structure contains information about a particular \cdrom\ drive,
207 such as its device name, speed, etc. This structure is conceptually
208 connected to the minor number of the device.
212 low-level device driver though a call to:
213 $$register_cdrom(struct\ cdrom_device_info * <device>_info)
214 $$
215 The device information structure, $<device>_info$, contains all the
216 information needed for the kernel to interface with the low-level
217 \cdrom\ device driver. One of the most important entries in this
218 structure is a pointer to the $cdrom_device_ops$ structure of the
219 low-level driver.
221 The device operations structure, $cdrom_device_ops$, contains a list
222 of pointers to the functions which are implemented in the low-level
224 through the functions in this structure. It is impossible to know all
225 the capabilities of future \cdrom\ drives, so it is expected that this
226 list may need to be expanded from time to time as new technologies are
227 developed. For example, CD-R and CD-R/W drives are beginning to become
228 popular, and support will soon need to be added for them. For now, the
229 current $struct$ is:
230 $$
234 &int& (* open)(struct\ cdrom_device_info *, int)\cr
236 &int& (* drive_status)(struct\ cdrom_device_info *, int);\cr
237 &int& (* media_changed)(struct\ cdrom_device_info *, int);\cr
238 &int& (* tray_move)(struct\ cdrom_device_info *, int);\cr
239 &int& (* lock_door)(struct\ cdrom_device_info *, int);\cr
240 &int& (* select_speed)(struct\ cdrom_device_info *, int);\cr
241 &int& (* select_disc)(struct\ cdrom_device_info *, int);\cr
242 &int& (* get_last_session) (struct\ cdrom_device_info *,
243 struct\ cdrom_multisession *{});\cr
244 &int& (* get_mcn)(struct\ cdrom_device_info *, struct\ cdrom_mcn *{});\cr
246 &int& (* audio_ioctl)(struct\ cdrom_device_info *, unsigned\ int,
247 void *{});\cr
248 &int& (* dev_ioctl)(struct\ cdrom_device_info *, unsigned\ int,
249 unsigned\ long);\cr
250 \noalign{\medskip}
251 &const\ int& capability;& capability flags \cr
252 \};\cr
253 }
254 $$
255 When a low-level device driver implements one of these capabilities,
256 it should add a function pointer to this $struct$. When a particular
257 function is not implemented, however, this $struct$ should contain a
258 NULL instead. The $capability$ flags specify the capabilities of the
259 \cdrom\ hardware and/or low-level \cdrom\ driver when a \cdrom\ drive
260 is registered with the \UCD.
262 Note that most functions have fewer parameters than their
263 $blkdev_fops$ counterparts. This is because very little of the
264 information in the structures $inode$ and $file$ is used. For most
265 drivers, the main parameter is the $struct$ $cdrom_device_info$, from
266 which the major and minor number can be extracted. (Most low-level
267 \cdrom\ drivers don't even look at the major and minor number though,
268 since many of them only support one device.) This will be available
269 through $dev$ in $cdrom_device_info$ described below.
271 The drive-specific, minor-like information that is registered with
272 \cdromc, currently contains the following fields:
273 $$
274 \halign{$#$\ \hfil&$#$\ \hfil&\hbox to 10em{$#$\hss}&
275 $/*$ \rm# $*/$\hfil\cr
276 struct& cdrom_device_info\ \{ \hidewidth\cr
277 & struct\ cdrom_device_ops *& ops;& device operations for this major\cr
278 & struct\ cdrom_device_info *& next;& next device_info for this major\cr
279 & void *& handle;& driver-dependent data\cr
280 \noalign{\medskip}
281 & kdev_t& dev;& device number (incorporates minor)\cr
282 & int& mask;& mask of capability: disables them \cr
283 & int& speed;& maximum speed for reading data \cr
284 & int& capacity;& number of discs in a jukebox \cr
285 \noalign{\medskip}
286 &int& options : 30;& options flags \cr
287 &unsigned& mc_flags : 2;& media-change buffer flags \cr
288 & int& use_count;& number of times device is opened\cr
289 & char& name[20];& name of the device type\cr
290 \}\cr
291 }$$
292 Using this $struct$, a linked list of the registered minor devices is
293 built, using the $next$ field. The device number, the device operations
294 struct and specifications of properties of the drive are stored in this
295 structure.
297 The $mask$ flags can be used to mask out some of the capabilities listed
298 in $ops\to capability$, if a specific drive doesn't support a feature
299 of the driver. The value $speed$ specifies the maximum head-rate of the
300 drive, measured in units of normal audio speed (176\,kB/sec raw data or
301 150\,kB/sec file system data). The value $n_discs$ should reflect the
302 number of discs the drive can hold simultaneously, if it is designed
303 as a juke-box, or otherwise~1. The parameters are declared $const$
304 because they describe properties of the drive, which don't change after
305 registration.
307 A few registers contain variables local to the \cdrom\ drive. The
308 flags $options$ are used to specify how the general \cdrom\ routines
309 should behave. These various flags registers should provide enough
310 flexibility to adapt to the different users' wishes (and {\em not\/} the
311 `arbitrary' wishes of the author of the low-level device driver, as is
312 the case in the old scheme). The register $mc_flags$ is used to buffer
313 the information from $media_changed()$ to two separate queues. Other
314 data that is specific to a minor drive, can be accessed through $handle$,
315 which can point to a data structure specific to the low-level driver.
316 The fields $use_count$, $next$, $options$ and $mc_flags$ need not be
317 initialized.
319 The intermediate software layer that \cdromc\ forms will perform some
320 additional bookkeeping. The use count of the device (the number of
321 processes that have the device opened) is registered in $use_count$. The
322 function $cdrom_ioctl()$ will verify the appropriate user-memory regions
323 for read and write, and in case a location on the CD is transferred,
324 it will `sanitize' the format by making requests to the low-level
325 drivers in a standard format, and translating all formats between the
326 user-software and low level drivers. This relieves much of the drivers'
327 memory checking and format checking and translation. Also, the necessary
328 structures will be declared on the program stack.
330 The implementation of the functions should be as defined in the
331 following sections. Two functions {\em must\/} be implemented, namely
332 $open()$ and $release()$. Other functions may be omitted, their
333 corresponding capability flags will be cleared upon registration.
334 Generally, a function returns zero on success and negative on error. A
335 function call should return only after the command has completed, but of
336 course waiting for the device should not use processor time.
338 \subsection{$Int\ open(struct\ cdrom_device_info * cdi, int\ purpose)$}
340 $Open()$ should try to open the device for a specific $purpose$, which
341 can be either:
342 \begin{itemize}
343 \item[0] Open for reading data, as done by {\tt {mount()}} (2), or the
344 user commands {\tt {dd}} or {\tt {cat}}.
345 \item[1] Open for $ioctl$ commands, as done by audio-CD playing
346 programs.
347 \end{itemize}
348 Notice that any strategic code (closing tray upon $open()$, etc.)\ is
349 done by the calling routine in \cdromc, so the low-level routine
350 should only be concerned with proper initialization, such as spinning
351 up the disc, etc. % and device-use count
354 \subsection{$Void\ release(struct\ cdrom_device_info * cdi)$}
357 Device-specific actions should be taken such as spinning down the device.
358 However, strategic actions such as ejection of the tray, or unlocking
359 the door, should be left over to the general routine $cdrom_release()$.
360 This is the only function returning type $void$.
362 \subsection{$Int\ drive_status(struct\ cdrom_device_info * cdi, int\ slot_nr)$}
363 \label{drive status}
365 The function $drive_status$, if implemented, should provide
366 information on the status of the drive (not the status of the disc,
367 which may or may not be in the drive). If the drive is not a changer,
368 $slot_nr$ should be ignored. In \cdromh\ the possibilities are listed:
369 $$
370 \halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
371 CDS_NO_INFO& no information available\cr
372 CDS_NO_DISC& no disc is inserted, tray is closed\cr
373 CDS_TRAY_OPEN& tray is opened\cr
374 CDS_DRIVE_NOT_READY& something is wrong, tray is moving?\cr
375 CDS_DISC_OK& a disc is loaded and everything is fine\cr
376 }
377 $$
379 \subsection{$Int\ media_changed(struct\ cdrom_device_info * cdi, int\ disc_nr)$}
381 This function is very similar to the original function in $struct\
382 file_operations$. It returns 1 if the medium of the device $cdi\to
383 dev$ has changed since the last call, and 0 otherwise. The parameter
384 $disc_nr$ identifies a specific slot in a juke-box, it should be
385 ignored for single-disc drives. Note that by `re-routing' this
386 function through $cdrom_media_changed()$, we can implement separate
387 queues for the VFS and a new $ioctl()$ function that can report device
388 changes to software (\eg, an auto-mounting daemon).
390 \subsection{$Int\ tray_move(struct\ cdrom_device_info * cdi, int\ position)$}
392 This function, if implemented, should control the tray movement. (No
393 other function should control this.) The parameter $position$ controls
394 the desired direction of movement:
395 \begin{itemize}
396 \item[0] Close tray
397 \item[1] Open tray
398 \end{itemize}
399 This function returns 0 upon success, and a non-zero value upon
400 error. Note that if the tray is already in the desired position, no
401 action need be taken, and the return value should be 0.
403 \subsection{$Int\ lock_door(struct\ cdrom_device_info * cdi, int\ lock)$}
405 This function (and no other code) controls locking of the door, if the
406 drive allows this. The value of $lock$ controls the desired locking
407 state:
408 \begin{itemize}
409 \item[0] Unlock door, manual opening is allowed
410 \item[1] Lock door, tray cannot be ejected manually
411 \end{itemize}
412 This function returns 0 upon success, and a non-zero value upon
413 error. Note that if the door is already in the requested state, no
414 action need be taken, and the return value should be 0.
416 \subsection{$Int\ select_speed(struct\ cdrom_device_info * cdi, int\ speed)$}
418 Some \cdrom\ drives are capable of changing their head-speed. There
419 are several reasons for changing the speed of a \cdrom\ drive. Badly
420 pressed \cdrom s may benefit from less-than-maximum head rate. Modern
421 \cdrom\ drives can obtain very high head rates (up to $24\times$ is
422 common). It has been reported that these drives can make reading
423 errors at these high speeds, reducing the speed can prevent data loss
424 in these circumstances. Finally, some of these drives can
425 make an annoyingly loud noise, which a lower speed may reduce. %Finally,
426 %although the audio-low-pass filters probably aren't designed for it,
427 %more than real-time playback of audio might be used for high-speed
428 %copying of audio tracks.
430 This function specifies the speed at which data is read or audio is
431 played back. The value of $speed$ specifies the head-speed of the
432 drive, measured in units of standard cdrom speed (176\,kB/sec raw data
433 or 150\,kB/sec file system data). So to request that a \cdrom\ drive
434 operate at 300\,kB/sec you would call the CDROM_SELECT_SPEED $ioctl$
435 with $speed=2$. The special value `0' means `auto-selection', \ie,
436 maximum data-rate or real-time audio rate. If the drive doesn't have
437 this `auto-selection' capability, the decision should be made on the
438 current disc loaded and the return value should be positive. A negative
439 return value indicates an error.
441 \subsection{$Int\ select_disc(struct\ cdrom_device_info * cdi, int\ number)$}
443 If the drive can store multiple discs (a juke-box) this function
444 will perform disc selection. It should return the number of the
445 selected disc on success, a negative value on error. Currently, only
446 the ide-cd driver supports this functionality.
448 \subsection{$Int\ get_last_session(struct\ cdrom_device_info * cdi, struct\
449 cdrom_multisession * ms_info)$}
451 This function should implement the old corresponding $ioctl()$. For
452 device $cdi\to dev$, the start of the last session of the current disc
453 should be returned in the pointer argument $ms_info$. Note that
454 routines in \cdromc\ have sanitized this argument: its requested
455 format will {\em always\/} be of the type $CDROM_LBA$ (linear block
456 addressing mode), whatever the calling software requested. But
457 sanitization goes even further: the low-level implementation may
458 return the requested information in $CDROM_MSF$ format if it wishes so
459 (setting the $ms_info\rightarrow addr_format$ field appropriately, of
460 course) and the routines in \cdromc\ will make the transformation if
461 necessary. The return value is 0 upon success.
463 \subsection{$Int\ get_mcn(struct\ cdrom_device_info * cdi, struct\
464 cdrom_mcn * mcn)$}
466 Some discs carry a `Media Catalog Number' (MCN), also called
467 `Universal Product Code' (UPC). This number should reflect the number
468 that is generally found in the bar-code on the product. Unfortunately,
469 the few discs that carry such a number on the disc don't even use the
470 same format. The return argument to this function is a pointer to a
471 pre-declared memory region of type $struct\ cdrom_mcn$. The MCN is
472 expected as a 13-character string, terminated by a null-character.
474 \subsection{$Int\ reset(struct\ cdrom_device_info * cdi)$}
476 This call should perform a hard-reset on the drive (although in
477 circumstances that a hard-reset is necessary, a drive may very well not
478 listen to commands anymore). Preferably, control is returned to the
479 caller only after the drive has finished resetting. If the drive is no
480 longer listening, it may be wise for the underlying low-level cdrom
481 driver to time out.
483 \subsection{$Int\ audio_ioctl(struct\ cdrom_device_info * cdi, unsigned\
484 int\ cmd, void * arg)$}
486 Some of the \cdrom-$ioctl$s defined in \cdromh\ can be
487 implemented by the routines described above, and hence the function
488 $cdrom_ioctl$ will use those. However, most $ioctl$s deal with
489 audio-control. We have decided to leave these to be accessed through a
490 single function, repeating the arguments $cmd$ and $arg$. Note that
491 the latter is of type $void*{}$, rather than $unsigned\ long\
492 int$. The routine $cdrom_ioctl()$ does do some useful things,
493 though. It sanitizes the address format type to $CDROM_MSF$ (Minutes,
494 Seconds, Frames) for all audio calls. It also verifies the memory
495 location of $arg$, and reserves stack-memory for the argument. This
496 makes implementation of the $audio_ioctl()$ much simpler than in the
497 old driver scheme. For example, you may look up the function
498 $cm206_audio_ioctl()$ in {\tt {cm206.c}} that should be updated with
499 this documentation.
501 An unimplemented ioctl should return $-ENOSYS$, but a harmless request
502 (\eg, $CDROMSTART$) may be ignored by returning 0 (success). Other
503 errors should be according to the standards, whatever they are. When
504 an error is returned by the low-level driver, the \UCD\ tries whenever
505 possible to return the error code to the calling program. (We may decide
506 to sanitize the return value in $cdrom_ioctl()$ though, in order to
507 guarantee a uniform interface to the audio-player software.)
509 \subsection{$Int\ dev_ioctl(struct\ cdrom_device_info * cdi, unsigned\ int\
510 cmd, unsigned\ long\ arg)$}
512 Some $ioctl$s seem to be specific to certain \cdrom\ drives. That is,
513 they are introduced to service some capabilities of certain drives. In
514 fact, there are 6 different $ioctl$s for reading data, either in some
515 particular kind of format, or audio data. Not many drives support
516 reading audio tracks as data, I believe this is because of protection
517 of copyrights of artists. Moreover, I think that if audio-tracks are
518 supported, it should be done through the VFS and not via $ioctl$s. A
519 problem here could be the fact that audio-frames are 2352 bytes long,
520 so either the audio-file-system should ask for 75264 bytes at once
521 (the least common multiple of 512 and 2352), or the drivers should
522 bend their backs to cope with this incoherence (to which I would be
523 opposed). Furthermore, it is very difficult for the hardware to find
524 the exact frame boundaries, since there are no synchronization headers
525 in audio frames. Once these issues are resolved, this code should be
526 standardized in \cdromc.
528 Because there are so many $ioctl$s that seem to be introduced to
529 satisfy certain drivers,\footnote{Is there software around that
530 actually uses these? I'd be interested!} any `non-standard' $ioctl$s
531 are routed through the call $dev_ioctl()$. In principle, `private'
532 $ioctl$s should be numbered after the device's major number, and not
533 the general \cdrom\ $ioctl$ number, {\tt {0x53}}. Currently the
534 non-supported $ioctl$s are: {\it CDROMREADMODE1, CDROMREADMODE2,
535 CDROMREADAUDIO, CDROMREADRAW, CDROMREADCOOKED, CDROMSEEK,
536 CDROMPLAY\-BLK and CDROM\-READALL}.
539 \subsection{\cdrom\ capabilities}
540 \label{capability}
542 Instead of just implementing some $ioctl$ calls, the interface in
543 \cdromc\ supplies the possibility to indicate the {\em capabilities\/}
544 of a \cdrom\ drive. This can be done by ORing any number of
545 capability-constants that are defined in \cdromh\ at the registration
546 phase. Currently, the capabilities are any of:
547 $$
548 \halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
549 CDC_CLOSE_TRAY& can close tray by software control\cr
550 CDC_OPEN_TRAY& can open tray\cr
551 CDC_LOCK& can lock and unlock the door\cr
552 CDC_SELECT_SPEED& can select speed, in units of $\sim$150\,kB/s\cr
553 CDC_SELECT_DISC& drive is juke-box\cr
554 CDC_MULTI_SESSION& can read sessions $>\rm1$\cr
555 CDC_MCN& can read Media Catalog Number\cr
556 CDC_MEDIA_CHANGED& can report if disc has changed\cr
557 CDC_PLAY_AUDIO& can perform audio-functions (play, pause, etc)\cr
558 CDC_RESET& hard reset device\cr
559 CDC_IOCTLS& driver has non-standard ioctls\cr
560 CDC_DRIVE_STATUS& driver implements drive status\cr
561 }
562 $$
563 The capability flag is declared $const$, to prevent drivers from
564 accidentally tampering with the contents. The capability fags actually
565 inform \cdromc\ of what the driver can do. If the drive found
566 by the driver does not have the capability, is can be masked out by
567 the $cdrom_device_info$ variable $mask$. For instance, the SCSI \cdrom\
568 driver has implemented the code for loading and ejecting \cdrom's, and
569 hence its corresponding flags in $capability$ will be set. But a SCSI
570 \cdrom\ drive might be a caddy system, which can't load the tray, and
571 hence for this drive the $cdrom_device_info$ struct will have set
572 the $CDC_CLOSE_TRAY$ bit in $mask$.
574 In the file \cdromc\ you will encounter many constructions of the type
575 $$\it
576 if\ (cdo\rightarrow capability \mathrel\& \mathord{\sim} cdi\rightarrow mask
577 \mathrel{\&} CDC_<capability>) \ldots
578 $$
579 There is no $ioctl$ to set the mask\dots The reason is that
580 I think it is better to control the {\em behavior\/} rather than the
581 {\em capabilities}.
583 \subsection{Options}
585 A final flag register controls the {\em behavior\/} of the \cdrom\
586 drives, in order to satisfy different users' wishes, hopefully
587 independently of the ideas of the respective author who happened to
588 have made the drive's support available to the \linux\ community. The
589 current behavior options are:
590 $$
591 \halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
592 CDO_AUTO_CLOSE& try to close tray upon device $open()$\cr
593 CDO_AUTO_EJECT& try to open tray on last device $close()$\cr
594 CDO_USE_FFLAGS& use $file_pointer\rightarrow f_flags$ to indicate
595 purpose for $open()$\cr
596 CDO_LOCK& try to lock door if device is opened\cr
597 CDO_CHECK_TYPE& ensure disc type is data if opened for data\cr
598 }
599 $$
601 The initial value of this register is $CDO_AUTO_CLOSE \mathrel|
602 CDO_USE_FFLAGS \mathrel| CDO_LOCK$, reflecting my own view on user
603 interface and software standards. Before you protest, there are two
604 new $ioctl$s implemented in \cdromc, that allow you to control the
605 behavior by software. These are:
606 $$
607 \halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
608 CDROM_SET_OPTIONS& set options specified in $(int)\ arg$\cr
609 CDROM_CLEAR_OPTIONS& clear options specified in $(int)\ arg$\cr
610 }
611 $$
612 One option needs some more explanation: $CDO_USE_FFLAGS$. In the next
613 newsection we explain what the need for this option is.
615 A software package {\tt setcd}, available from the Debian distribution
616 and {\tt sunsite.unc.edu}, allows user level control of these flags.
618 \newsection{The need to know the purpose of opening the \cdrom\ device}
620 Traditionally, Unix devices can be used in two different `modes',
621 either by reading/writing to the device file, or by issuing
622 controlling commands to the device, by the device's $ioctl()$
623 call. The problem with \cdrom\ drives, is that they can be used for
624 two entirely different purposes. One is to mount removable
625 file systems, \cdrom s, the other is to play audio CD's. Audio commands
626 are implemented entirely through $ioctl$s, presumably because the
627 first implementation (SUN?) has been such. In principle there is
628 nothing wrong with this, but a good control of the `CD player' demands
629 that the device can {\em always\/} be opened in order to give the
630 $ioctl$ commands, regardless of the state the drive is in.
632 On the other hand, when used as a removable-media disc drive (what the
633 original purpose of \cdrom s is) we would like to make sure that the
634 disc drive is ready for operation upon opening the device. In the old
635 scheme, some \cdrom\ drivers don't do any integrity checking, resulting
636 in a number of i/o errors reported by the VFS to the kernel when an
637 attempt for mounting a \cdrom\ on an empty drive occurs. This is not a
638 particularly elegant way to find out that there is no \cdrom\ inserted;
639 it more-or-less looks like the old IBM-PC trying to read an empty floppy
640 drive for a couple of seconds, after which the system complains it
641 can't read from it. Nowadays we can {\em sense\/} the existence of a
642 removable medium in a drive, and we believe we should exploit that
643 fact. An integrity check on opening of the device, that verifies the
644 availability of a \cdrom\ and its correct type (data), would be
645 desirable.
647 These two ways of using a \cdrom\ drive, principally for data and
648 secondarily for playing audio discs, have different demands for the
649 behavior of the $open()$ call. Audio use simply wants to open the
650 device in order to get a file handle which is needed for issuing
651 $ioctl$ commands, while data use wants to open for correct and
652 reliable data transfer. The only way user programs can indicate what
653 their {\em purpose\/} of opening the device is, is through the $flags$
654 parameter (see {\tt {open(2)}}). For \cdrom\ devices, these flags aren't
655 implemented (some drivers implement checking for write-related flags,
656 but this is not strictly necessary if the device file has correct
657 permission flags). Most option flags simply don't make sense to
658 \cdrom\ devices: $O_CREAT$, $O_NOCTTY$, $O_TRUNC$, $O_APPEND$, and
659 $O_SYNC$ have no meaning to a \cdrom.
661 We therefore propose to use the flag $O_NONBLOCK$ to indicate
662 that the device is opened just for issuing $ioctl$
663 commands. Strictly, the meaning of $O_NONBLOCK$ is that opening and
664 subsequent calls to the device don't cause the calling process to
665 wait. We could interpret this as ``don't wait until someone has
666 inserted some valid data-\cdrom.'' Thus, our proposal of the
667 implementation for the $open()$ call for \cdrom s is:
668 \begin{itemize}
669 \item If no other flags are set than $O_RDONLY$, the device is opened
670 for data transfer, and the return value will be 0 only upon successful
671 initialization of the transfer. The call may even induce some actions
672 on the \cdrom, such as closing the tray.
673 \item If the option flag $O_NONBLOCK$ is set, opening will always be
674 successful, unless the whole device doesn't exist. The drive will take
675 no actions whatsoever.
676 \end{itemize}
678 \subsection{And what about standards?}
680 You might hesitate to accept this proposal as it comes from the
681 \linux\ community, and not from some standardizing institute. What
682 about SUN, SGI, HP and all those other Unix and hardware vendors?
683 Well, these companies are in the lucky position that they generally
684 control both the hardware and software of their supported products,
685 and are large enough to set their own standard. They do not have to
686 deal with a dozen or more different, competing hardware
687 configurations.\footnote{Incidentally, I think that SUN's approach to
688 mounting \cdrom s is very good in origin: under Solaris a
689 volume-daemon automatically mounts a newly inserted \cdrom\ under {\tt
690 {/cdrom/$<volume-name>$/}}. In my opinion they should have pushed this
691 further and have {\em every\/} \cdrom\ on the local area network be
692 mounted at the similar location, \ie, no matter in which particular
693 machine you insert a \cdrom, it will always appear at the same
694 position in the directory tree, on every system. When I wanted to
695 implement such a user-program for \linux, I came across the
696 differences in behavior of the various drivers, and the need for an
697 $ioctl$ informing about media changes.}
699 We believe that using $O_NONBLOCK$ to indicate that a device is being opened
700 for $ioctl$ commands only can be easily introduced in the \linux\
701 community. All the CD-player authors will have to be informed, we can
702 even send in our own patches to the programs. The use of $O_NONBLOCK$
703 has most likely no influence on the behavior of the CD-players on
704 other operating systems than \linux. Finally, a user can always revert
705 to old behavior by a call to $ioctl(file_descriptor, CDROM_CLEAR_OPTIONS,
706 CDO_USE_FFLAGS)$.
708 \subsection{The preferred strategy of $open()$}
710 The routines in \cdromc\ are designed in such a way that run-time
711 configuration of the behavior of \cdrom\ devices (of {\em any\/} type)
712 can be carried out, by the $CDROM_SET/CLEAR_OPTIONS$ $ioctls$. Thus, various
713 modes of operation can be set:
714 \begin{description}
715 \item[$CDO_AUTO_CLOSE \mathrel| CDO_USE_FFLAGS \mathrel| CDO_LOCK$] This
716 is the default setting. (With $CDO_CHECK_TYPE$ it will be better, in the
717 future.) If the device is not yet opened by any other process, and if
718 the device is being opened for data ($O_NONBLOCK$ is not set) and the
719 tray is found to be open, an attempt to close the tray is made. Then,
720 it is verified that a disc is in the drive and, if $CDO_CHECK_TYPE$ is
721 set, that it contains tracks of type `data mode 1.' Only if all tests
722 are passed is the return value zero. The door is locked to prevent file
723 system corruption. If the drive is opened for audio ($O_NONBLOCK$ is
724 set), no actions are taken and a value of 0 will be returned.
725 \item[$CDO_AUTO_CLOSE \mathrel| CDO_AUTO_EJECT \mathrel| CDO_LOCK$] This
726 mimics the behavior of the current sbpcd-driver. The option flags are
727 ignored, the tray is closed on the first open, if necessary. Similarly,
728 the tray is opened on the last release, \ie, if a \cdrom\ is unmounted,
729 it is automatically ejected, such that the user can replace it.
730 \end{description}
731 We hope that these option can convince everybody (both driver
732 maintainers and user program developers) to adopt the new \cdrom\
733 driver scheme and option flag interpretation.
735 \newsection{Description of routines in \cdromc}
737 Only a few routines in \cdromc\ are exported to the drivers. In this
738 new section we will discuss these, as well as the functions that `take
739 over' the \cdrom\ interface to the kernel. The header file belonging
740 to \cdromc\ is called \cdromh. Formerly, some of the contents of this
741 file were placed in the file {\tt {ucdrom.h}}, but this file has now been
742 merged back into \cdromh.
744 \subsection{$Struct\ file_operations\ cdrom_fops$}
746 The contents of this structure were described in section~\ref{cdrom.c}.
747 A pointer to this structure is assigned to the $fops$ field
748 of the $struct gendisk$.
750 \subsection{$Int\ register_cdrom( struct\ cdrom_device_info\ * cdi)$}
752 This function is used in about the same way one registers $cdrom_fops$
753 with the kernel, the device operations and information structures,
754 as described in section~\ref{cdrom.c}, should be registered with the
755 \UCD:
756 $$
757 register_cdrom(\&<device>_info));
758 $$
759 This function returns zero upon success, and non-zero upon
760 failure. The structure $<device>_info$ should have a pointer to the
761 driver's $<device>_dops$, as in
762 $$
763 \vbox{\halign{&$#$\hfil\cr
764 struct\ &cdrom_device_info\ <device>_info = \{\cr
765 & <device>_dops;\cr
766 &\ldots\cr
767 \}\cr
768 }}$$
769 Note that a driver must have one static structure, $<device>_dops$, while
770 it may have as many structures $<device>_info$ as there are minor devices
771 active. $Register_cdrom()$ builds a linked list from these.
773 \subsection{$Void\ unregister_cdrom(struct\ cdrom_device_info * cdi)$}
775 Unregistering device $cdi$ with minor number $MINOR(cdi\to dev)$ removes
776 the minor device from the list. If it was the last registered minor for
777 the low-level driver, this disconnects the registered device-operation
778 routines from the \cdrom\ interface. This function returns zero upon
779 success, and non-zero upon failure.
781 \subsection{$Int\ cdrom_open(struct\ inode * ip, struct\ file * fp)$}
783 This function is not called directly by the low-level drivers, it is
784 listed in the standard $cdrom_fops$. If the VFS opens a file, this
785 function becomes active. A strategy is implemented in this routine,
786 taking care of all capabilities and options that are set in the
787 $cdrom_device_ops$ connected to the device. Then, the program flow is
788 transferred to the device_dependent $open()$ call.
790 \subsection{$Void\ cdrom_release(struct\ inode *ip, struct\ file
791 *fp)$}
793 This function implements the reverse-logic of $cdrom_open()$, and then
794 calls the device-dependent $release()$ routine. When the use-count has
795 reached 0, the allocated buffers are flushed by calls to $sync_dev(dev)$
796 and $invalidate_buffers(dev)$.
799 \subsection{$Int\ cdrom_ioctl(struct\ inode *ip, struct\ file *fp,
800 unsigned\ int\ cmd, unsigned\ long\ arg)$}
801 \label{cdrom-ioctl}
803 This function handles all the standard $ioctl$ requests for \cdrom\
804 devices in a uniform way. The different calls fall into three
805 categories: $ioctl$s that can be directly implemented by device
806 operations, ones that are routed through the call $audio_ioctl()$, and
807 the remaining ones, that are presumable device-dependent. Generally, a
808 negative return value indicates an error.
810 \subsubsection{Directly implemented $ioctl$s}
811 \label{ioctl-direct}
813 The following `old' \cdrom-$ioctl$s are implemented by directly
814 calling device-operations in $cdrom_device_ops$, if implemented and
815 not masked:
816 \begin{description}
817 \item[CDROMMULTISESSION] Requests the last session on a \cdrom.
818 \item[CDROMEJECT] Open tray.
819 \item[CDROMCLOSETRAY] Close tray.
820 \item[CDROMEJECT_SW] If $arg\not=0$, set behavior to auto-close (close
821 tray on first open) and auto-eject (eject on last release), otherwise
822 set behavior to non-moving on $open()$ and $release()$ calls.
823 \item[CDROM_GET_MCN] Get the Media Catalog Number from a CD.
824 \end{description}
826 \subsubsection{$Ioctl$s routed through $audio_ioctl()$}
827 \label{ioctl-audio}
829 The following set of $ioctl$s are all implemented through a call to
830 the $cdrom_fops$ function $audio_ioctl()$. Memory checks and
831 allocation are performed in $cdrom_ioctl()$, and also sanitization of
832 address format ($CDROM_LBA$/$CDROM_MSF$) is done.
833 \begin{description}
834 \item[CDROMSUBCHNL] Get sub-channel data in argument $arg$ of type $struct\
835 cdrom_subchnl *{}$.
836 \item[CDROMREADTOCHDR] Read Table of Contents header, in $arg$ of type
837 $struct\ cdrom_tochdr *{}$.
838 \item[CDROMREADTOCENTRY] Read a Table of Contents entry in $arg$ and
839 specified by $arg$ of type $struct\ cdrom_tocentry *{}$.
840 \item[CDROMPLAYMSF] Play audio fragment specified in Minute, Second,
841 Frame format, delimited by $arg$ of type $struct\ cdrom_msf *{}$.
842 \item[CDROMPLAYTRKIND] Play audio fragment in track-index format
843 delimited by $arg$ of type $struct\ \penalty-1000 cdrom_ti *{}$.
844 \item[CDROMVOLCTRL] Set volume specified by $arg$ of type $struct\
845 cdrom_volctrl *{}$.
846 \item[CDROMVOLREAD] Read volume into by $arg$ of type $struct\
847 cdrom_volctrl *{}$.
848 \item[CDROMSTART] Spin up disc.
849 \item[CDROMSTOP] Stop playback of audio fragment.
850 \item[CDROMPAUSE] Pause playback of audio fragment.
851 \item[CDROMRESUME] Resume playing.
852 \end{description}
854 \subsubsection{New $ioctl$s in \cdromc}
856 The following $ioctl$s have been introduced to allow user programs to
857 control the behavior of individual \cdrom\ devices. New $ioctl$
858 commands can be identified by the underscores in their names.
859 \begin{description}
860 \item[CDROM_SET_OPTIONS] Set options specified by $arg$. Returns the
861 option flag register after modification. Use $arg = \rm0$ for reading
862 the current flags.
863 \item[CDROM_CLEAR_OPTIONS] Clear options specified by $arg$. Returns
864 the option flag register after modification.
865 \item[CDROM_SELECT_SPEED] Select head-rate speed of disc specified as
866 by $arg$ in units of standard cdrom speed (176\,kB/sec raw data or
867 150\,kB/sec file system data). The value 0 means `auto-select', \ie,
868 play audio discs at real time and data discs at maximum speed. The value
869 $arg$ is checked against the maximum head rate of the drive found in the
870 $cdrom_dops$.
871 \item[CDROM_SELECT_DISC] Select disc numbered $arg$ from a juke-box.
872 First disc is numbered 0. The number $arg$ is checked against the
873 maximum number of discs in the juke-box found in the $cdrom_dops$.
874 \item[CDROM_MEDIA_CHANGED] Returns 1 if a disc has been changed since
875 the last call. Note that calls to $cdrom_media_changed$ by the VFS
876 are treated by an independent queue, so both mechanisms will detect
877 a media change once. For juke-boxes, an extra argument $arg$
878 specifies the slot for which the information is given. The special
879 value $CDSL_CURRENT$ requests that information about the currently
880 selected slot be returned.
881 \item[CDROM_DRIVE_STATUS] Returns the status of the drive by a call to
882 $drive_status()$. Return values are defined in section~\ref{drive
883 status}. Note that this call doesn't return information on the
884 current playing activity of the drive; this can be polled through an
885 $ioctl$ call to $CDROMSUBCHNL$. For juke-boxes, an extra argument
886 $arg$ specifies the slot for which (possibly limited) information is
887 given. The special value $CDSL_CURRENT$ requests that information
888 about the currently selected slot be returned.
889 \item[CDROM_DISC_STATUS] Returns the type of the disc currently in the
890 drive. It should be viewed as a complement to $CDROM_DRIVE_STATUS$.
891 This $ioctl$ can provide \emph {some} information about the current
892 disc that is inserted in the drive. This functionality used to be
893 implemented in the low level drivers, but is now carried out
894 entirely in \UCD.
896 The history of development of the CD's use as a carrier medium for
897 various digital information has lead to many different disc types.
898 This $ioctl$ is useful only in the case that CDs have \emph {only
899 one} type of data on them. While this is often the case, it is
900 also very common for CDs to have some tracks with data, and some
901 tracks with audio. Because this is an existing interface, rather
902 than fixing this interface by changing the assumptions it was made
903 under, thereby breaking all user applications that use this
904 function, the \UCD\ implements this $ioctl$ as follows: If the CD in
905 question has audio tracks on it, and it has absolutely no CD-I, XA,
906 or data tracks on it, it will be reported as $CDS_AUDIO$. If it has
907 both audio and data tracks, it will return $CDS_MIXED$. If there
908 are no audio tracks on the disc, and if the CD in question has any
909 CD-I tracks on it, it will be reported as $CDS_XA_2_2$. Failing
910 that, if the CD in question has any XA tracks on it, it will be
911 reported as $CDS_XA_2_1$. Finally, if the CD in question has any
912 data tracks on it, it will be reported as a data CD ($CDS_DATA_1$).
914 This $ioctl$ can return:
915 $$
916 \halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
917 CDS_NO_INFO& no information available\cr
918 CDS_NO_DISC& no disc is inserted, or tray is opened\cr
919 CDS_AUDIO& Audio disc (2352 audio bytes/frame)\cr
920 CDS_DATA_1& data disc, mode 1 (2048 user bytes/frame)\cr
921 CDS_XA_2_1& mixed data (XA), mode 2, form 1 (2048 user bytes)\cr
922 CDS_XA_2_2& mixed data (XA), mode 2, form 1 (2324 user bytes)\cr
923 CDS_MIXED& mixed audio/data disc\cr
924 }
925 $$
926 For some information concerning frame layout of the various disc
927 types, see a recent version of \cdromh.
929 \item[CDROM_CHANGER_NSLOTS] Returns the number of slots in a
930 juke-box.
931 \item[CDROMRESET] Reset the drive.
932 \item[CDROM_GET_CAPABILITY] Returns the $capability$ flags for the
933 drive. Refer to section \ref{capability} for more information on
934 these flags.
935 \item[CDROM_LOCKDOOR] Locks the door of the drive. $arg == \rm0$
936 unlocks the door, any other value locks it.
937 \item[CDROM_DEBUG] Turns on debugging info. Only root is allowed
938 to do this. Same semantics as CDROM_LOCKDOOR.
939 \end{description}
941 \subsubsection{Device dependent $ioctl$s}
943 Finally, all other $ioctl$s are passed to the function $dev_ioctl()$,
944 if implemented. No memory allocation or verification is carried out.
946 \newsection{How to update your driver}
948 \begin{enumerate}
949 \item Make a backup of your current driver.
950 \item Get hold of the files \cdromc\ and \cdromh, they should be in
951 the directory tree that came with this documentation.
952 \item Make sure you include \cdromh.
953 \item Change the 3rd argument of $register_blkdev$ from
954 $\&<your-drive>_fops$ to $\&cdrom_fops$.
955 \item Just after that line, add the following to register with the \UCD:
956 $$register_cdrom(\&<your-drive>_info);$$
957 Similarly, add a call to $unregister_cdrom()$ at the appropriate place.
958 \item Copy an example of the device-operations $struct$ to your
959 source, \eg, from {\tt {cm206.c}} $cm206_dops$, and change all
960 entries to names corresponding to your driver, or names you just
961 happen to like. If your driver doesn't support a certain function,
962 make the entry $NULL$. At the entry $capability$ you should list all
963 capabilities your driver currently supports. If your driver
964 has a capability that is not listed, please send me a message.
965 \item Copy the $cdrom_device_info$ declaration from the same example
966 driver, and modify the entries according to your needs. If your
967 driver dynamically determines the capabilities of the hardware, this
968 structure should also be declared dynamically.
969 \item Implement all functions in your $<device>_dops$ structure,
970 according to prototypes listed in \cdromh, and specifications given
971 in section~\ref{cdrom.c}. Most likely you have already implemented
972 the code in a large part, and you will almost certainly need to adapt the
973 prototype and return values.
974 \item Rename your $<device>_ioctl()$ function to $audio_ioctl$ and
975 change the prototype a little. Remove entries listed in the first
976 part in section~\ref{cdrom-ioctl}, if your code was OK, these are
977 just calls to the routines you adapted in the previous step.
978 \item You may remove all remaining memory checking code in the
979 $audio_ioctl()$ function that deals with audio commands (these are
980 listed in the second part of section~\ref{cdrom-ioctl}). There is no
981 need for memory allocation either, so most $case$s in the $switch$
982 statement look similar to:
983 $$
984 case\ CDROMREADTOCENTRY\colon get_toc_entry\bigl((struct\
985 cdrom_tocentry *{})\ arg\bigr);
986 $$
987 \item All remaining $ioctl$ cases must be moved to a separate
988 function, $<device>_ioctl$, the device-dependent $ioctl$s. Note that
989 memory checking and allocation must be kept in this code!
990 \item Change the prototypes of $<device>_open()$ and
991 $<device>_release()$, and remove any strategic code (\ie, tray
992 movement, door locking, etc.).
993 \item Try to recompile the drivers. We advise you to use modules, both
994 for {\tt {cdrom.o}} and your driver, as debugging is much easier this
995 way.
996 \end{enumerate}
998 \newsection{Thanks}
1000 Thanks to all the people involved. First, Erik Andersen, who has
1001 taken over the torch in maintaining \cdromc\ and integrating much
1002 \cdrom-related code in the 2.1-kernel. Thanks to Scott Snyder and
1003 Gerd Knorr, who were the first to implement this interface for SCSI
1004 and IDE-CD drivers and added many ideas for extension of the data
1005 structures relative to kernel~2.0. Further thanks to Heiko Ei{\sz}feldt,
1006 Thomas Quinot, Jon Tombs, Ken Pizzini, Eberhard M\"onkeberg and Andrew
1007 Kroll, the \linux\ \cdrom\ device driver developers who were kind
1008 enough to give suggestions and criticisms during the writing. Finally
1009 of course, I want to thank Linus Torvalds for making this possible in
1010 the first place.
1012 \vfill
1013 $ \version\ $
1014 \eject
1015 \end{document}