2\def\version{$Id: cdrom-standard.tex,v 1.9 1997/12/28 15:42:49 david Exp $}
7\topmargin=-\headheight \advance\topmargin by -\headsep
8\textwidth=15.99cm \textheight=24.62cm % normal A4, 1'' margin
10\def\linux{{\sc Linux}}
11\def\cdrom{{\sc cd-rom}}
12\def\UCD{{\sc Uniform cd-rom Driver}}
13\def\cdromc{{\tt {cdrom.c}}}
14\def\cdromh{{\tt {cdrom.h}}}
15\def\fo{\sl} % foreign words
16\def\ie{{\fo i.e.}}
17\def\eg{{\fo e.g.}}
19\everymath{\it} \everydisplay{\it}
20\catcode `\_=\active \def_{\_\penalty100 }
21\catcode`\<=\active \def<#1>{{\langle\hbox{\rm#1}\rangle}}
24\title{A \linux\ \cdrom\ standard}
25\author{David van Leeuwen\\{\normalsize\tt}
26\\{\footnotesize updated by Erik Andersen {\tt(}}
27\\{\footnotesize updated by Jens Axboe {\tt(}}}
28\date{12 March 1999}
34\linux\ is probably the Unix-like operating system that supports
35the widest variety of hardware devices. The reasons for this are
39  The large list of hardware devices available for the many platforms
40  that \linux\ now supports (\ie, i386-PCs, Sparc Suns, etc.)
42  The open design of the operating system, such that anybody can write a
43  driver for \linux.
45  There is plenty of source code around as examples of how to write a driver.
47The openness of \linux, and the many different types of available
48hardware has allowed \linux\ to support many different hardware devices.
49Unfortunately, the very openness that has allowed \linux\ to support
50all these different devices has also allowed the behavior of each
51device driver to differ significantly from one device to another.
52This divergence of behavior has been very significant for \cdrom\
53devices; the way a particular drive reacts to a `standard' $ioctl()$
54call varies greatly from one device driver to another. To avoid making
55their drivers totally inconsistent, the writers of \linux\ \cdrom\
56drivers generally created new device drivers by understanding, copying,
57and then changing an existing one. Unfortunately, this practice did not
58maintain uniform behavior across all the \linux\ \cdrom\ drivers.
60This document describes an effort to establish Uniform behavior across
61all the different \cdrom\ device drivers for \linux. This document also
62defines the various $ioctl$s, and how the low-level \cdrom\ device
63drivers should implement them. Currently (as of the \linux\ 2.1.$x$
64development kernels) several low-level \cdrom\ device drivers, including
65both IDE/ATAPI and SCSI, now use this Uniform interface.
67When the \cdrom\ was developed, the interface between the \cdrom\ drive
68and the computer was not specified in the standards. As a result, many
69different \cdrom\ interfaces were developed. Some of them had their
70own proprietary design (Sony, Mitsumi, Panasonic, Philips), other
71manufacturers adopted an existing electrical interface and changed
72the functionality (CreativeLabs/SoundBlaster, Teac, Funai) or simply
73adapted their drives to one or more of the already existing electrical
74interfaces (Aztech, Sanyo, Funai, Vertos, Longshine, Optics Storage and
75most of the `NoName' manufacturers). In cases where a new drive really
76brought its own interface or used its own command set and flow control
77scheme, either a separate driver had to be written, or an existing
78driver had to be enhanced. History has delivered us \cdrom\ support for
79many of these different interfaces. Nowadays, almost all new \cdrom\
80drives are either IDE/ATAPI or SCSI, and it is very unlikely that any
81manufacturer will create a new interface. Even finding drives for the
82old proprietary interfaces is getting difficult.
84When (in the 1.3.70's) I looked at the existing software interface,
85which was expressed through \cdromh, it appeared to be a rather wild
86set of commands and data formats.\footnote{I cannot recollect what
87kernel version I looked at, then, presumably 1.2.13 and 1.3.34---the
88latest kernel that I was indirectly involved in.} It seemed that many
89features of the software interface had been added to accommodate the
90capabilities of a particular drive, in an {\fo ad hoc\/} manner. More
91importantly, it appeared that the behavior of the `standard' commands
92was different for most of the different drivers: \eg, some drivers
93close the tray if an $open()$ call occurs when the tray is open, while
94others do not. Some drivers lock the door upon opening the device, to
95prevent an incoherent file system, but others don't, to allow software
96ejection. Undoubtedly, the capabilities of the different drives vary,
97but even when two drives have the same capability their drivers'
98behavior was usually different.
100I decided to start a discussion on how to make all the \linux\ \cdrom\
101drivers behave more uniformly. I began by contacting the developers of
102the many \cdrom\ drivers found in the \linux\ kernel. Their reactions
103encouraged me to write the \UCD\ which this document is intended to
104describe. The implementation of the \UCD\ is in the file \cdromc. This
105driver is intended to be an additional software layer that sits on top
106of the low-level device drivers for each \cdrom\ drive. By adding this
107additional layer, it is possible to have all the different \cdrom\
108devices behave {\em exactly\/} the same (insofar as the underlying
109hardware will allow).
111The goal of the \UCD\ is {\em not\/} to alienate driver developers who
112have not yet taken steps to support this effort. The goal of \UCD\ is
113simply to give people writing application programs for \cdrom\ drives
114{\em one\/} \linux\ \cdrom\ interface with consistent behavior for all
115\cdrom\ devices. In addition, this also provides a consistent interface
116between the low-level device driver code and the \linux\ kernel. Care
117is taken that 100\,\% compatibility exists with the data structures and
118programmer's interface defined in \cdromh. This guide was written to
119help \cdrom\ driver developers adapt their code to use the \UCD\ code
120defined in \cdromc.
122Personally, I think that the most important hardware interfaces are
123the IDE/ATAPI drives and, of course, the SCSI drives, but as prices
124of hardware drop continuously, it is also likely that people may have
125more than one \cdrom\ drive, possibly of mixed types. It is important
126that these drives behave in the same way. In December 1994, one of the
127cheapest \cdrom\ drives was a Philips cm206, a double-speed proprietary
128drive. In the months that I was busy writing a \linux\ driver for it,
129proprietary drives became obsolete and IDE/ATAPI drives became the
130standard. At the time of the last update to this document (November
1311997) it is becoming difficult to even {\em find} anything less than a
13216 speed \cdrom\ drive, and 24 speed drives are common.
134\newsection{Standardizing through another software level}
137At the time this document was conceived, all drivers directly
138implemented the \cdrom\ $ioctl()$ calls through their own routines. This
139led to the danger of different drivers forgetting to do important things
140like checking that the user was giving the driver valid data. More
141importantly, this led to the divergence of behavior, which has already
142been discussed.
144For this reason, the \UCD\ was created to enforce consistent \cdrom\
145drive behavior, and to provide a common set of services to the various
146low-level \cdrom\ device drivers. The \UCD\ now provides another
147software-level, that separates the $ioctl()$ and $open()$ implementation
148from the actual hardware implementation. Note that this effort has
149made few changes which will affect a user's application programs. The
150greatest change involved moving the contents of the various low-level
151\cdrom\ drivers' header files to the kernel's cdrom directory. This was
152done to help ensure that the user is only presented with only one cdrom
153interface, the interface defined in \cdromh.
155\cdrom\ drives are specific enough (\ie, different from other
156block-devices such as floppy or hard disc drives), to define a set
157of common {\em \cdrom\ device operations}, $<cdrom-device>_dops$.
158These operations are different from the classical block-device file
159operations, $<block-device>_fops$.
161The routines for the \UCD\ interface level are implemented in the file
162\cdromc. In this file, the \UCD\ interfaces with the kernel as a block
163device by registering the following general $struct\ file_operations$:
165\halign{$#$\ \hfil&$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
166struct& file_operations\ cdrom_fops = \{\hidewidth\cr
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
184Every active \cdrom\ device shares this $struct$. The routines
185declared above are all implemented in \cdromc, since this file is the
186place where the behavior of all \cdrom-devices is defined and
187standardized. The actual interface to the various types of \cdrom\
188hardware is still performed by various low-level \cdrom-device
189drivers. These routines simply implement certain {\em capabilities\/}
190that are common to all \cdrom\ (and really, all removable-media
193Registration of a low-level \cdrom\ device driver is now done through
194the general routines in \cdromc, not through the Virtual File System
195(VFS) any more. The interface implemented in \cdromc\ is carried out
196through two general structures that contain information about the
197capabilities of the driver, and the specific drives on which the
198driver 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.
211Registering a particular \cdrom\ drive with the \UCD\ is done by the
212low-level device driver though a call to:
213$$register_cdrom(struct\ cdrom_device_info * <device>_info)
215The device information structure, $<device>_info$, contains all the
216information needed for the kernel to interface with the low-level
217\cdrom\ device driver. One of the most important entries in this
218structure is a pointer to the $cdrom_device_ops$ structure of the
219low-level driver.
221The device operations structure, $cdrom_device_ops$, contains a list
222of pointers to the functions which are implemented in the low-level
223device driver. When \cdromc\ accesses a \cdrom\ device, it does it
224through the functions in this structure. It is impossible to know all
225the capabilities of future \cdrom\ drives, so it is expected that this
226list may need to be expanded from time to time as new technologies are
227developed. For example, CD-R and CD-R/W drives are beginning to become
228popular, and support will soon need to be added for them. For now, the
229current $struct$ is:
231\halign{$#$\ \hfil&$#$\ \hfil&\hbox to 10em{$#$\hss}&
232  $/*$ \rm# $*/$\hfil\cr
233struct& cdrom_device_ops\ \{ \hidewidth\cr
234  &int& (* open)(struct\ cdrom_device_info *, int)\cr
235  &void& (* release)(struct\ cdrom_device_info *);\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
245  &int& (* reset)(struct\ cdrom_device_info *);\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
251  &const\ int& capability;& capability flags \cr
252  &int& n_minors;& number of active minor devices \cr
256When a low-level device driver implements one of these capabilities,
257it should add a function pointer to this $struct$. When a particular
258function is not implemented, however, this $struct$ should contain a
259NULL instead. The $capability$ flags specify the capabilities of the
260\cdrom\ hardware and/or low-level \cdrom\ driver when a \cdrom\ drive
261is registered with the \UCD. The value $n_minors$ should be a positive
262value indicating the number of minor devices that are supported by
263the low-level device driver, normally~1. Although these two variables
264are `informative' rather than `operational,' they are included in
265$cdrom_device_ops$ because they describe the capability of the {\em
266driver\/} rather than the {\em drive}. Nomenclature has always been
267difficult in computer programming.
269Note that most functions have fewer parameters than their
270$blkdev_fops$ counterparts. This is because very little of the
271information in the structures $inode$ and $file$ is used. For most
272drivers, the main parameter is the $struct$ $cdrom_device_info$, from
273which the major and minor number can be extracted. (Most low-level
274\cdrom\ drivers don't even look at the major and minor number though,
275since many of them only support one device.) This will be available
276through $dev$ in $cdrom_device_info$ described below.
278The drive-specific, minor-like information that is registered with
279\cdromc, currently contains the following fields:
281\halign{$#$\ \hfil&$#$\ \hfil&\hbox to 10em{$#$\hss}&
282  $/*$ \rm# $*/$\hfil\cr
283struct& cdrom_device_info\ \{ \hidewidth\cr
284  & struct\ cdrom_device_ops *& ops;& device operations for this major\cr
285  & struct\ cdrom_device_info *& next;& next device_info for this major\cr
286  & void *& handle;& driver-dependent data\cr
288  & kdev_t& dev;& device number (incorporates minor)\cr
289  & int& mask;& mask of capability: disables them \cr
290  & int& speed;& maximum speed for reading data \cr
291  & int& capacity;& number of discs in a jukebox \cr
293  &int& options : 30;& options flags \cr
294  &unsigned& mc_flags : 2;& media-change buffer flags \cr
295  & int& use_count;& number of times device is opened\cr
296  & char& name[20];& name of the device type\cr
299Using this $struct$, a linked list of the registered minor devices is
300built, using the $next$ field. The device number, the device operations
301struct and specifications of properties of the drive are stored in this
304The $mask$ flags can be used to mask out some of the capabilities listed
305in $ops\to capability$, if a specific drive doesn't support a feature
306of the driver. The value $speed$ specifies the maximum head-rate of the
307drive, measured in units of normal audio speed (176\,kB/sec raw data or
308150\,kB/sec file system data). The value $n_discs$ should reflect the
309number of discs the drive can hold simultaneously, if it is designed
310as a juke-box, or otherwise~1. The parameters are declared $const$
311because they describe properties of the drive, which don't change after
314A few registers contain variables local to the \cdrom\ drive. The
315flags $options$ are used to specify how the general \cdrom\ routines
316should behave. These various flags registers should provide enough
317flexibility to adapt to the different users' wishes (and {\em not\/} the
318`arbitrary' wishes of the author of the low-level device driver, as is
319the case in the old scheme). The register $mc_flags$ is used to buffer
320the information from $media_changed()$ to two separate queues. Other
321data that is specific to a minor drive, can be accessed through $handle$,
322which can point to a data structure specific to the low-level driver.
323The fields $use_count$, $next$, $options$ and $mc_flags$ need not be
326The intermediate software layer that \cdromc\ forms will perform some
327additional bookkeeping. The use count of the device (the number of
328processes that have the device opened) is registered in $use_count$. The
329function $cdrom_ioctl()$ will verify the appropriate user-memory regions
330for read and write, and in case a location on the CD is transferred,
331it will `sanitize' the format by making requests to the low-level
332drivers in a standard format, and translating all formats between the
333user-software and low level drivers. This relieves much of the drivers'
334memory checking and format checking and translation. Also, the necessary
335structures will be declared on the program stack.
337The implementation of the functions should be as defined in the
338following sections. Two functions {\em must\/} be implemented, namely
339$open()$ and $release()$. Other functions may be omitted, their
340corresponding capability flags will be cleared upon registration.
341Generally, a function returns zero on success and negative on error. A
342function call should return only after the command has completed, but of
343course waiting for the device should not use processor time.
345\subsection{$Int\ open(struct\ cdrom_device_info * cdi, int\ purpose)$}
347$Open()$ should try to open the device for a specific $purpose$, which
348can be either:
350\item[0] Open for reading data, as done by {\tt {mount()}} (2), or the
351user commands {\tt {dd}} or {\tt {cat}}.
352\item[1] Open for $ioctl$ commands, as done by audio-CD playing
355Notice that any strategic code (closing tray upon $open()$, etc.)\ is
356done by the calling routine in \cdromc, so the low-level routine
357should only be concerned with proper initialization, such as spinning
358up the disc, etc. % and device-use count
361\subsection{$Void\ release(struct\ cdrom_device_info * cdi)$}
364Device-specific actions should be taken such as spinning down the device.
365However, strategic actions such as ejection of the tray, or unlocking
366the door, should be left over to the general routine $cdrom_release()$.
367This is the only function returning type $void$.
369\subsection{$Int\ drive_status(struct\ cdrom_device_info * cdi, int\ slot_nr)$}
370\label{drive status}
372The function $drive_status$, if implemented, should provide
373information on the status of the drive (not the status of the disc,
374which may or may not be in the drive). If the drive is not a changer,
375$slot_nr$ should be ignored. In \cdromh\ the possibilities are listed:
377\halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
378CDS_NO_INFO& no information available\cr
379CDS_NO_DISC& no disc is inserted, tray is closed\cr
380CDS_TRAY_OPEN& tray is opened\cr
381CDS_DRIVE_NOT_READY& something is wrong, tray is moving?\cr
382CDS_DISC_OK& a disc is loaded and everything is fine\cr
386\subsection{$Int\ media_changed(struct\ cdrom_device_info * cdi, int\ disc_nr)$}
388This function is very similar to the original function in $struct\
389file_operations$. It returns 1 if the medium of the device $cdi\to
390dev$ has changed since the last call, and 0 otherwise. The parameter
391$disc_nr$ identifies a specific slot in a juke-box, it should be
392ignored for single-disc drives. Note that by `re-routing' this
393function through $cdrom_media_changed()$, we can implement separate
394queues for the VFS and a new $ioctl()$ function that can report device
395changes to software (\eg, an auto-mounting daemon).
397\subsection{$Int\ tray_move(struct\ cdrom_device_info * cdi, int\ position)$}
399This function, if implemented, should control the tray movement. (No
400other function should control this.) The parameter $position$ controls
401the desired direction of movement:
403\item[0] Close tray
404\item[1] Open tray
406This function returns 0 upon success, and a non-zero value upon
407error. Note that if the tray is already in the desired position, no
408action need be taken, and the return value should be 0.
410\subsection{$Int\ lock_door(struct\ cdrom_device_info * cdi, int\ lock)$}
412This function (and no other code) controls locking of the door, if the
413drive allows this. The value of $lock$ controls the desired locking
416\item[0] Unlock door, manual opening is allowed
417\item[1] Lock door, tray cannot be ejected manually
419This function returns 0 upon success, and a non-zero value upon
420error. Note that if the door is already in the requested state, no
421action need be taken, and the return value should be 0.
423\subsection{$Int\ select_speed(struct\ cdrom_device_info * cdi, int\ speed)$}
425Some \cdrom\ drives are capable of changing their head-speed. There
426are several reasons for changing the speed of a \cdrom\ drive. Badly
427pressed \cdrom s may benefit from less-than-maximum head rate. Modern
428\cdrom\ drives can obtain very high head rates (up to $24\times$ is
429common). It has been reported that these drives can make reading
430errors at these high speeds, reducing the speed can prevent data loss
431in these circumstances. Finally, some of these drives can
432make an annoyingly loud noise, which a lower speed may reduce. %Finally,
433%although the audio-low-pass filters probably aren't designed for it,
434%more than real-time playback of audio might be used for high-speed
435%copying of audio tracks.
437This function specifies the speed at which data is read or audio is
438played back. The value of $speed$ specifies the head-speed of the
439drive, measured in units of standard cdrom speed (176\,kB/sec raw data
440or 150\,kB/sec file system data). So to request that a \cdrom\ drive
441operate at 300\,kB/sec you would call the CDROM_SELECT_SPEED $ioctl$
442with $speed=2$. The special value `0' means `auto-selection', \ie,
443maximum data-rate or real-time audio rate. If the drive doesn't have
444this `auto-selection' capability, the decision should be made on the
445current disc loaded and the return value should be positive. A negative
446return value indicates an error.
448\subsection{$Int\ select_disc(struct\ cdrom_device_info * cdi, int\ number)$}
450If the drive can store multiple discs (a juke-box) this function
451will perform disc selection. It should return the number of the
452selected disc on success, a negative value on error. Currently, only
453the ide-cd driver supports this functionality.
455\subsection{$Int\ get_last_session(struct\ cdrom_device_info * cdi, struct\
456  cdrom_multisession * ms_info)$}
458This function should implement the old corresponding $ioctl()$. For
459device $cdi\to dev$, the start of the last session of the current disc
460should be returned in the pointer argument $ms_info$. Note that
461routines in \cdromc\ have sanitized this argument: its requested
462format will {\em always\/} be of the type $CDROM_LBA$ (linear block
463addressing mode), whatever the calling software requested. But
464sanitization goes even further: the low-level implementation may
465return the requested information in $CDROM_MSF$ format if it wishes so
466(setting the $ms_info\rightarrow addr_format$ field appropriately, of
467course) and the routines in \cdromc\ will make the transformation if
468necessary. The return value is 0 upon success.
470\subsection{$Int\ get_mcn(struct\ cdrom_device_info * cdi, struct\
471  cdrom_mcn * mcn)$}
473Some discs carry a `Media Catalog Number' (MCN), also called
474`Universal Product Code' (UPC). This number should reflect the number
475that is generally found in the bar-code on the product. Unfortunately,
476the few discs that carry such a number on the disc don't even use the
477same format. The return argument to this function is a pointer to a
478pre-declared memory region of type $struct\ cdrom_mcn$. The MCN is
479expected as a 13-character string, terminated by a null-character.
481\subsection{$Int\ reset(struct\ cdrom_device_info * cdi)$}
483This call should perform a hard-reset on the drive (although in
484circumstances that a hard-reset is necessary, a drive may very well not
485listen to commands anymore). Preferably, control is returned to the
486caller only after the drive has finished resetting. If the drive is no
487longer listening, it may be wise for the underlying low-level cdrom
488driver to time out.
490\subsection{$Int\ audio_ioctl(struct\ cdrom_device_info * cdi, unsigned\
491  int\ cmd, void * arg)$}
493Some of the \cdrom-$ioctl$s defined in \cdromh\ can be
494implemented by the routines described above, and hence the function
495$cdrom_ioctl$ will use those. However, most $ioctl$s deal with
496audio-control. We have decided to leave these to be accessed through a
497single function, repeating the arguments $cmd$ and $arg$. Note that
498the latter is of type $void*{}$, rather than $unsigned\ long\
499int$. The routine $cdrom_ioctl()$ does do some useful things,
500though. It sanitizes the address format type to $CDROM_MSF$ (Minutes,
501Seconds, Frames) for all audio calls. It also verifies the memory
502location of $arg$, and reserves stack-memory for the argument. This
503makes implementation of the $audio_ioctl()$ much simpler than in the
504old driver scheme. For example, you may look up the function
505$cm206_audio_ioctl()$ in {\tt {cm206.c}} that should be updated with
506this documentation.
508An unimplemented ioctl should return $-ENOSYS$, but a harmless request
509(\eg, $CDROMSTART$) may be ignored by returning 0 (success). Other
510errors should be according to the standards, whatever they are. When
511an error is returned by the low-level driver, the \UCD\ tries whenever
512possible to return the error code to the calling program. (We may decide
513to sanitize the return value in $cdrom_ioctl()$ though, in order to
514guarantee a uniform interface to the audio-player software.)
516\subsection{$Int\ dev_ioctl(struct\ cdrom_device_info * cdi, unsigned\ int\
517  cmd, unsigned\ long\ arg)$}
519Some $ioctl$s seem to be specific to certain \cdrom\ drives. That is,
520they are introduced to service some capabilities of certain drives. In
521fact, there are 6 different $ioctl$s for reading data, either in some
522particular kind of format, or audio data. Not many drives support
523reading audio tracks as data, I believe this is because of protection
524of copyrights of artists. Moreover, I think that if audio-tracks are
525supported, it should be done through the VFS and not via $ioctl$s. A
526problem here could be the fact that audio-frames are 2352 bytes long,
527so either the audio-file-system should ask for 75264 bytes at once
528(the least common multiple of 512 and 2352), or the drivers should
529bend their backs to cope with this incoherence (to which I would be
530opposed). Furthermore, it is very difficult for the hardware to find
531the exact frame boundaries, since there are no synchronization headers
532in audio frames. Once these issues are resolved, this code should be
533standardized in \cdromc.
535Because there are so many $ioctl$s that seem to be introduced to
536satisfy certain drivers,\footnote{Is there software around that
537  actually uses these? I'd be interested!} any `non-standard' $ioctl$s
538are routed through the call $dev_ioctl()$. In principle, `private'
539$ioctl$s should be numbered after the device's major number, and not
540the general \cdrom\ $ioctl$ number, {\tt {0x53}}. Currently the
541non-supported $ioctl$s are: {\it CDROMREADMODE1, CDROMREADMODE2,
546\subsection{\cdrom\ capabilities}
549Instead of just implementing some $ioctl$ calls, the interface in
550\cdromc\ supplies the possibility to indicate the {\em capabilities\/}
551of a \cdrom\ drive. This can be done by ORing any number of
552capability-constants that are defined in \cdromh\ at the registration
553phase. Currently, the capabilities are any of:
555\halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
556CDC_CLOSE_TRAY& can close tray by software control\cr
557CDC_OPEN_TRAY& can open tray\cr
558CDC_LOCK& can lock and unlock the door\cr
559CDC_SELECT_SPEED& can select speed, in units of $\sim$150\,kB/s\cr
560CDC_SELECT_DISC& drive is juke-box\cr
561CDC_MULTI_SESSION& can read sessions $>\rm1$\cr
562CDC_MCN& can read Media Catalog Number\cr
563CDC_MEDIA_CHANGED& can report if disc has changed\cr
564CDC_PLAY_AUDIO& can perform audio-functions (play, pause, etc)\cr
565CDC_RESET& hard reset device\cr
566CDC_IOCTLS& driver has non-standard ioctls\cr
567CDC_DRIVE_STATUS& driver implements drive status\cr
570The capability flag is declared $const$, to prevent drivers from
571accidentally tampering with the contents. The capability fags actually
572inform \cdromc\ of what the driver can do. If the drive found
573by the driver does not have the capability, is can be masked out by
574the $cdrom_device_info$ variable $mask$. For instance, the SCSI \cdrom\
575driver has implemented the code for loading and ejecting \cdrom's, and
576hence its corresponding flags in $capability$ will be set. But a SCSI
577\cdrom\ drive might be a caddy system, which can't load the tray, and
578hence for this drive the $cdrom_device_info$ struct will have set
579the $CDC_CLOSE_TRAY$ bit in $mask$.
581In the file \cdromc\ you will encounter many constructions of the type
583if\ (cdo\rightarrow capability \mathrel\& \mathord{\sim} cdi\rightarrow mask
584   \mathrel{\&} CDC_<capability>) \ldots
586There is no $ioctl$ to set the mask\dots The reason is that
587I think it is better to control the {\em behavior\/} rather than the
588{\em capabilities}.
592A final flag register controls the {\em behavior\/} of the \cdrom\
593drives, in order to satisfy different users' wishes, hopefully
594independently of the ideas of the respective author who happened to
595have made the drive's support available to the \linux\ community. The
596current behavior options are:
598\halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
599CDO_AUTO_CLOSE& try to close tray upon device $open()$\cr
600CDO_AUTO_EJECT& try to open tray on last device $close()$\cr
601CDO_USE_FFLAGS& use $file_pointer\rightarrow f_flags$ to indicate
602 purpose for $open()$\cr
603CDO_LOCK& try to lock door if device is opened\cr
604CDO_CHECK_TYPE& ensure disc type is data if opened for data\cr
608The initial value of this register is $CDO_AUTO_CLOSE \mathrel|
609CDO_USE_FFLAGS \mathrel| CDO_LOCK$, reflecting my own view on user
610interface and software standards. Before you protest, there are two
611new $ioctl$s implemented in \cdromc, that allow you to control the
612behavior by software. These are:
614\halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
615CDROM_SET_OPTIONS& set options specified in $(int)\ arg$\cr
616CDROM_CLEAR_OPTIONS& clear options specified in $(int)\ arg$\cr
619One option needs some more explanation: $CDO_USE_FFLAGS$. In the next
620newsection we explain what the need for this option is.
622A software package {\tt setcd}, available from the Debian distribution
623and {\tt}, allows user level control of these flags.
625\newsection{The need to know the purpose of opening the \cdrom\ device}
627Traditionally, Unix devices can be used in two different `modes',
628either by reading/writing to the device file, or by issuing
629controlling commands to the device, by the device's $ioctl()$
630call. The problem with \cdrom\ drives, is that they can be used for
631two entirely different purposes. One is to mount removable
632file systems, \cdrom s, the other is to play audio CD's. Audio commands
633are implemented entirely through $ioctl$s, presumably because the
634first implementation (SUN?) has been such. In principle there is
635nothing wrong with this, but a good control of the `CD player' demands
636that the device can {\em always\/} be opened in order to give the
637$ioctl$ commands, regardless of the state the drive is in.
639On the other hand, when used as a removable-media disc drive (what the
640original purpose of \cdrom s is) we would like to make sure that the
641disc drive is ready for operation upon opening the device. In the old
642scheme, some \cdrom\ drivers don't do any integrity checking, resulting
643in a number of i/o errors reported by the VFS to the kernel when an
644attempt for mounting a \cdrom\ on an empty drive occurs. This is not a
645particularly elegant way to find out that there is no \cdrom\ inserted;
646it more-or-less looks like the old IBM-PC trying to read an empty floppy
647drive for a couple of seconds, after which the system complains it
648can't read from it. Nowadays we can {\em sense\/} the existence of a
649removable medium in a drive, and we believe we should exploit that
650fact. An integrity check on opening of the device, that verifies the
651availability of a \cdrom\ and its correct type (data), would be
654These two ways of using a \cdrom\ drive, principally for data and
655secondarily for playing audio discs, have different demands for the
656behavior of the $open()$ call. Audio use simply wants to open the
657device in order to get a file handle which is needed for issuing
658$ioctl$ commands, while data use wants to open for correct and
659reliable data transfer. The only way user programs can indicate what
660their {\em purpose\/} of opening the device is, is through the $flags$
661parameter (see {\tt {open(2)}}). For \cdrom\ devices, these flags aren't
662implemented (some drivers implement checking for write-related flags,
663but this is not strictly necessary if the device file has correct
664permission flags). Most option flags simply don't make sense to
665\cdrom\ devices: $O_CREAT$, $O_NOCTTY$, $O_TRUNC$, $O_APPEND$, and
666$O_SYNC$ have no meaning to a \cdrom.
668We therefore propose to use the flag $O_NONBLOCK$ to indicate
669that the device is opened just for issuing $ioctl$
670commands. Strictly, the meaning of $O_NONBLOCK$ is that opening and
671subsequent calls to the device don't cause the calling process to
672wait. We could interpret this as ``don't wait until someone has
673inserted some valid data-\cdrom.'' Thus, our proposal of the
674implementation for the $open()$ call for \cdrom s is:
676\item If no other flags are set than $O_RDONLY$, the device is opened
677for data transfer, and the return value will be 0 only upon successful
678initialization of the transfer. The call may even induce some actions
679on the \cdrom, such as closing the tray.
680\item If the option flag $O_NONBLOCK$ is set, opening will always be
681successful, unless the whole device doesn't exist. The drive will take
682no actions whatsoever.
685\subsection{And what about standards?}
687You might hesitate to accept this proposal as it comes from the
688\linux\ community, and not from some standardizing institute. What
689about SUN, SGI, HP and all those other Unix and hardware vendors?
690Well, these companies are in the lucky position that they generally
691control both the hardware and software of their supported products,
692and are large enough to set their own standard. They do not have to
693deal with a dozen or more different, competing hardware
694configurations.\footnote{Incidentally, I think that SUN's approach to
695mounting \cdrom s is very good in origin: under Solaris a
696volume-daemon automatically mounts a newly inserted \cdrom\ under {\tt
697{/cdrom/$<volume-name>$/}}. In my opinion they should have pushed this
698further and have {\em every\/} \cdrom\ on the local area network be
699mounted at the similar location, \ie, no matter in which particular
700machine you insert a \cdrom, it will always appear at the same
701position in the directory tree, on every system. When I wanted to
702implement such a user-program for \linux, I came across the
703differences in behavior of the various drivers, and the need for an
704$ioctl$ informing about media changes.}
706We believe that using $O_NONBLOCK$ to indicate that a device is being opened
707for $ioctl$ commands only can be easily introduced in the \linux\
708community. All the CD-player authors will have to be informed, we can
709even send in our own patches to the programs. The use of $O_NONBLOCK$
710has most likely no influence on the behavior of the CD-players on
711other operating systems than \linux. Finally, a user can always revert
712to old behavior by a call to $ioctl(file_descriptor, CDROM_CLEAR_OPTIONS,
715\subsection{The preferred strategy of $open()$}
717The routines in \cdromc\ are designed in such a way that run-time
718configuration of the behavior of \cdrom\ devices (of {\em any\/} type)
719can be carried out, by the $CDROM_SET/CLEAR_OPTIONS$ $ioctls$. Thus, various
720modes of operation can be set:
722\item[$CDO_AUTO_CLOSE \mathrel| CDO_USE_FFLAGS \mathrel| CDO_LOCK$] This
723is the default setting. (With $CDO_CHECK_TYPE$ it will be better, in the
724future.) If the device is not yet opened by any other process, and if
725the device is being opened for data ($O_NONBLOCK$ is not set) and the
726tray is found to be open, an attempt to close the tray is made. Then,
727it is verified that a disc is in the drive and, if $CDO_CHECK_TYPE$ is
728set, that it contains tracks of type `data mode 1.' Only if all tests
729are passed is the return value zero. The door is locked to prevent file
730system corruption. If the drive is opened for audio ($O_NONBLOCK$ is
731set), no actions are taken and a value of 0 will be returned.
732\item[$CDO_AUTO_CLOSE \mathrel| CDO_AUTO_EJECT \mathrel| CDO_LOCK$] This
733mimics the behavior of the current sbpcd-driver. The option flags are
734ignored, the tray is closed on the first open, if necessary. Similarly,
735the tray is opened on the last release, \ie, if a \cdrom\ is unmounted,
736it is automatically ejected, such that the user can replace it.
738We hope that these option can convince everybody (both driver
739maintainers and user program developers) to adopt the new \cdrom\
740driver scheme and option flag interpretation.
742\newsection{Description of routines in \cdromc}
744Only a few routines in \cdromc\ are exported to the drivers. In this
745new section we will discuss these, as well as the functions that `take
746over' the \cdrom\ interface to the kernel. The header file belonging
747to \cdromc\ is called \cdromh. Formerly, some of the contents of this
748file were placed in the file {\tt {ucdrom.h}}, but this file has now been
749merged back into \cdromh.
751\subsection{$Struct\ file_operations\ cdrom_fops$}
753The contents of this structure were described in section~\ref{cdrom.c}.
754A pointer to this structure is assigned to the $fops$ field
755of the $struct gendisk$.
757\subsection{$Int\ register_cdrom( struct\ cdrom_device_info\ * cdi)$}
759This function is used in about the same way one registers $cdrom_fops$
760with the kernel, the device operations and information structures,
761as described in section~\ref{cdrom.c}, should be registered with the
766This function returns zero upon success, and non-zero upon
767failure. The structure $<device>_info$ should have a pointer to the
768driver's $<device>_dops$, as in
771struct\ &cdrom_device_info\ <device>_info = \{\cr
772& <device>_dops;\cr
776Note that a driver must have one static structure, $<device>_dops$, while
777it may have as many structures $<device>_info$ as there are minor devices
778active. $Register_cdrom()$ builds a linked list from these.
780\subsection{$Void\ unregister_cdrom(struct\ cdrom_device_info * cdi)$}
782Unregistering device $cdi$ with minor number $MINOR(cdi\to dev)$ removes
783the minor device from the list. If it was the last registered minor for
784the low-level driver, this disconnects the registered device-operation
785routines from the \cdrom\ interface. This function returns zero upon
786success, and non-zero upon failure.
788\subsection{$Int\ cdrom_open(struct\ inode * ip, struct\ file * fp)$}
790This function is not called directly by the low-level drivers, it is
791listed in the standard $cdrom_fops$. If the VFS opens a file, this
792function becomes active. A strategy is implemented in this routine,
793taking care of all capabilities and options that are set in the
794$cdrom_device_ops$ connected to the device. Then, the program flow is
795transferred to the device_dependent $open()$ call.
797\subsection{$Void\ cdrom_release(struct\ inode *ip, struct\ file
800This function implements the reverse-logic of $cdrom_open()$, and then
801calls the device-dependent $release()$ routine. When the use-count has
802reached 0, the allocated buffers are flushed by calls to $sync_dev(dev)$
803and $invalidate_buffers(dev)$.
806\subsection{$Int\ cdrom_ioctl(struct\ inode *ip, struct\ file *fp,
807unsigned\ int\ cmd, unsigned\ long\ arg)$}
810This function handles all the standard $ioctl$ requests for \cdrom\
811devices in a uniform way. The different calls fall into three
812categories: $ioctl$s that can be directly implemented by device
813operations, ones that are routed through the call $audio_ioctl()$, and
814the remaining ones, that are presumable device-dependent. Generally, a
815negative return value indicates an error.
817\subsubsection{Directly implemented $ioctl$s}
820The following `old' \cdrom-$ioctl$s are implemented by directly
821calling device-operations in $cdrom_device_ops$, if implemented and
822not masked:
824\item[CDROMMULTISESSION] Requests the last session on a \cdrom.
825\item[CDROMEJECT] Open tray.
826\item[CDROMCLOSETRAY] Close tray.
827\item[CDROMEJECT_SW] If $arg\not=0$, set behavior to auto-close (close
828tray on first open) and auto-eject (eject on last release), otherwise
829set behavior to non-moving on $open()$ and $release()$ calls.
830\item[CDROM_GET_MCN] Get the Media Catalog Number from a CD.
833\subsubsection{$Ioctl$s routed through $audio_ioctl()$}
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.
841\item[CDROMSUBCHNL] Get sub-channel data in argument $arg$ of type $struct\
842cdrom_subchnl *{}$.
843\item[CDROMREADTOCHDR] Read Table of Contents header, in $arg$ of type
844$struct\ cdrom_tochdr *{}$.
845\item[CDROMREADTOCENTRY] Read a Table of Contents entry in $arg$ and
846specified by $arg$ of type $struct\ cdrom_tocentry *{}$.
847\item[CDROMPLAYMSF] Play audio fragment specified in Minute, Second,
848Frame format, delimited by $arg$ of type $struct\ cdrom_msf *{}$.
849\item[CDROMPLAYTRKIND] Play audio fragment in track-index format
850delimited by $arg$ of type $struct\ \penalty-1000 cdrom_ti *{}$.
851\item[CDROMVOLCTRL] Set volume specified by $arg$ of type $struct\
852cdrom_volctrl *{}$.
853\item[CDROMVOLREAD] Read volume into by $arg$ of type $struct\
854cdrom_volctrl *{}$.
855\item[CDROMSTART] Spin up disc.
856\item[CDROMSTOP] Stop playback of audio fragment.
857\item[CDROMPAUSE] Pause playback of audio fragment.
858\item[CDROMRESUME] Resume playing.
861\subsubsection{New $ioctl$s in \cdromc}
863The following $ioctl$s have been introduced to allow user programs to
864control the behavior of individual \cdrom\ devices. New $ioctl$
865commands can be identified by the underscores in their names.
867\item[CDROM_SET_OPTIONS] Set options specified by $arg$. Returns the
868option flag register after modification. Use $arg = \rm0$ for reading
869the current flags.
870\item[CDROM_CLEAR_OPTIONS] Clear options specified by $arg$. Returns
871  the option flag register after modification.
872\item[CDROM_SELECT_SPEED] Select head-rate speed of disc specified as
873  by $arg$ in units of standard cdrom speed (176\,kB/sec raw data or
874  150\,kB/sec file system data). The value 0 means `auto-select', \ie,
875  play audio discs at real time and data discs at maximum speed. The value
876  $arg$ is checked against the maximum head rate of the drive found in the
877  $cdrom_dops$.
878\item[CDROM_SELECT_DISC] Select disc numbered $arg$ from a juke-box.
879  First disc is numbered 0. The number $arg$ is checked against the
880  maximum number of discs in the juke-box found in the $cdrom_dops$.
881\item[CDROM_MEDIA_CHANGED] Returns 1 if a disc has been changed since
882  the last call. Note that calls to $cdrom_media_changed$ by the VFS
883  are treated by an independent queue, so both mechanisms will detect
884  a media change once. For juke-boxes, an extra argument $arg$
885  specifies the slot for which the information is given. The special
886  value $CDSL_CURRENT$ requests that information about the currently
887  selected slot be returned.
888\item[CDROM_DRIVE_STATUS] Returns the status of the drive by a call to
889  $drive_status()$. Return values are defined in section~\ref{drive
890   status}. Note that this call doesn't return information on the
891  current playing activity of the drive; this can be polled through an
892  $ioctl$ call to $CDROMSUBCHNL$. For juke-boxes, an extra argument
893  $arg$ specifies the slot for which (possibly limited) information is
894  given. The special value $CDSL_CURRENT$ requests that information
895  about the currently selected slot be returned.
896\item[CDROM_DISC_STATUS] Returns the type of the disc currently in the
897  drive. It should be viewed as a complement to $CDROM_DRIVE_STATUS$.
898  This $ioctl$ can provide \emph {some} information about the current
899  disc that is inserted in the drive. This functionality used to be
900  implemented in the low level drivers, but is now carried out
901  entirely in \UCD.
903  The history of development of the CD's use as a carrier medium for
904  various digital information has lead to many different disc types.
905  This $ioctl$ is useful only in the case that CDs have \emph {only
906    one} type of data on them. While this is often the case, it is
907  also very common for CDs to have some tracks with data, and some
908  tracks with audio. Because this is an existing interface, rather
909  than fixing this interface by changing the assumptions it was made
910  under, thereby breaking all user applications that use this
911  function, the \UCD\ implements this $ioctl$ as follows: If the CD in
912  question has audio tracks on it, and it has absolutely no CD-I, XA,
913  or data tracks on it, it will be reported as $CDS_AUDIO$. If it has
914  both audio and data tracks, it will return $CDS_MIXED$. If there
915  are no audio tracks on the disc, and if the CD in question has any
916  CD-I tracks on it, it will be reported as $CDS_XA_2_2$. Failing
917  that, if the CD in question has any XA tracks on it, it will be
918  reported as $CDS_XA_2_1$. Finally, if the CD in question has any
919  data tracks on it, it will be reported as a data CD ($CDS_DATA_1$).
921  This $ioctl$ can return:
922  $$
923  \halign{$#$\ \hfil&$/*$ \rm# $*/$\hfil\cr
924    CDS_NO_INFO& no information available\cr
925    CDS_NO_DISC& no disc is inserted, or tray is opened\cr
926    CDS_AUDIO& Audio disc (2352 audio bytes/frame)\cr
927    CDS_DATA_1& data disc, mode 1 (2048 user bytes/frame)\cr
928    CDS_XA_2_1& mixed data (XA), mode 2, form 1 (2048 user bytes)\cr
929    CDS_XA_2_2& mixed data (XA), mode 2, form 1 (2324 user bytes)\cr
930    CDS_MIXED& mixed audio/data disc\cr
931    }
932  $$
933  For some information concerning frame layout of the various disc
934  types, see a recent version of \cdromh.
936\item[CDROM_CHANGER_NSLOTS] Returns the number of slots in a
937  juke-box.
938\item[CDROMRESET] Reset the drive.
939\item[CDROM_GET_CAPABILITY] Returns the $capability$ flags for the
940  drive. Refer to section \ref{capability} for more information on
941  these flags.
942\item[CDROM_LOCKDOOR] Locks the door of the drive. $arg == \rm0$
943  unlocks the door, any other value locks it.
944\item[CDROM_DEBUG] Turns on debugging info. Only root is allowed
945  to do this. Same semantics as CDROM_LOCKDOOR.
948\subsubsection{Device dependent $ioctl$s}
950Finally, all other $ioctl$s are passed to the function $dev_ioctl()$,
951if implemented. No memory allocation or verification is carried out.
953\newsection{How to update your driver}
956\item Make a backup of your current driver.
957\item Get hold of the files \cdromc\ and \cdromh, they should be in
958  the directory tree that came with this documentation.
959\item Make sure you include \cdromh.
960\item Change the 3rd argument of $register_blkdev$ from
961$\&<your-drive>_fops$ to $\&cdrom_fops$.
962\item Just after that line, add the following to register with the \UCD:
963  $$register_cdrom(\&<your-drive>_info);$$
964  Similarly, add a call to $unregister_cdrom()$ at the appropriate place.
965\item Copy an example of the device-operations $struct$ to your
966  source, \eg, from {\tt {cm206.c}} $cm206_dops$, and change all
967  entries to names corresponding to your driver, or names you just
968  happen to like. If your driver doesn't support a certain function,
969  make the entry $NULL$. At the entry $capability$ you should list all
970  capabilities your driver currently supports. If your driver
971  has a capability that is not listed, please send me a message.
972\item Copy the $cdrom_device_info$ declaration from the same example
973  driver, and modify the entries according to your needs. If your
974  driver dynamically determines the capabilities of the hardware, this
975  structure should also be declared dynamically.
976\item Implement all functions in your $<device>_dops$ structure,
977  according to prototypes listed in \cdromh, and specifications given
978  in section~\ref{cdrom.c}. Most likely you have already implemented
979  the code in a large part, and you will almost certainly need to adapt the
980  prototype and return values.
981\item Rename your $<device>_ioctl()$ function to $audio_ioctl$ and
982  change the prototype a little. Remove entries listed in the first
983  part in section~\ref{cdrom-ioctl}, if your code was OK, these are
984  just calls to the routines you adapted in the previous step.
985\item You may remove all remaining memory checking code in the
986  $audio_ioctl()$ function that deals with audio commands (these are
987  listed in the second part of section~\ref{cdrom-ioctl}). There is no
988  need for memory allocation either, so most $case$s in the $switch$
989  statement look similar to:
990  $$
991  case\ CDROMREADTOCENTRY\colon get_toc_entry\bigl((struct\
992  cdrom_tocentry *{})\ arg\bigr);
993  $$
994\item All remaining $ioctl$ cases must be moved to a separate
995  function, $<device>_ioctl$, the device-dependent $ioctl$s. Note that
996  memory checking and allocation must be kept in this code!
997\item Change the prototypes of $<device>_open()$ and
998  $<device>_release()$, and remove any strategic code (\ie, tray
999  movement, door locking, etc.).
1000\item Try to recompile the drivers. We advise you to use modules, both
1001  for {\tt {cdrom.o}} and your driver, as debugging is much easier this
1002  way.
1007Thanks to all the people involved. First, Erik Andersen, who has
1008taken over the torch in maintaining \cdromc\ and integrating much
1009\cdrom-related code in the 2.1-kernel. Thanks to Scott Snyder and
1010Gerd Knorr, who were the first to implement this interface for SCSI
1011and IDE-CD drivers and added many ideas for extension of the data
1012structures relative to kernel~2.0. Further thanks to Heiko Ei{\sz}feldt,
1013Thomas Quinot, Jon Tombs, Ken Pizzini, Eberhard M\"onkeberg and Andrew
1014Kroll, the \linux\ \cdrom\ device driver developers who were kind
1015enough to give suggestions and criticisms during the writing. Finally
1016of course, I want to thank Linus Torvalds for making this possible in
1017the first place.
1020$ \version\ $

Archive Download this file