1 This file contains brief information about the SCSI tape driver.
2 The driver is currently maintained by Kai Mäkisara (email
3 Kai.Makisara@kolumbus.fi)
5 Last modified: Mon Mar 7 21:14:44 2005 by kai.makisara
10 The driver is generic, i.e., it does not contain any code tailored
11 to any specific tape drive. The tape parameters can be specified with
12 one of the following three methods:
14 1. Each user can specify the tape parameters he/she wants to use
15 directly with ioctls. This is administratively a very simple and
16 flexible method and applicable to single-user workstations. However,
17 in a multiuser environment the next user finds the tape parameters in
18 state the previous user left them.
20 2. The system manager (root) can define default values for some tape
21 parameters, like block size and density using the MTSETDRVBUFFER ioctl.
22 These parameters can be programmed to come into effect either when a
23 new tape is loaded into the drive or if writing begins at the
24 beginning of the tape. The second method is applicable if the tape
25 drive performs auto-detection of the tape format well (like some
26 QIC-drives). The result is that any tape can be read, writing can be
27 continued using existing format, and the default format is used if
28 the tape is rewritten from the beginning (or a new tape is written
29 for the first time). The first method is applicable if the drive
30 does not perform auto-detection well enough and there is a single
31 "sensible" mode for the device. An example is a DAT drive that is
32 used only in variable block mode (I don't know if this is sensible
35 The user can override the parameters defined by the system
36 manager. The changes persist until the defaults again come into
39 3. By default, up to four modes can be defined and selected using the minor
40 number (bits 5 and 6). The number of modes can be changed by changing
41 ST_NBR_MODE_BITS in st.h. Mode 0 corresponds to the defaults discussed
42 above. Additional modes are dormant until they are defined by the
43 system manager (root). When specification of a new mode is started,
44 the configuration of mode 0 is used to provide a starting point for
45 definition of the new mode.
47 Using the modes allows the system manager to give the users choices
48 over some of the buffering parameters not directly accessible to the
49 users (buffered and asynchronous writes). The modes also allow choices
50 between formats in multi-tape operations (the explicitly overridden
51 parameters are reset when a new tape is loaded).
53 If more than one mode is used, all modes should contain definitions
54 for the same set of parameters.
56 Many Unices contain internal tables that associate different modes to
57 supported devices. The Linux SCSI tape driver does not contain such
58 tables (and will not do that in future). Instead of that, a utility
59 program can be made that fetches the inquiry data sent by the device,
60 scans its database, and sets up the modes using the ioctls. Another
61 alternative is to make a small script that uses mt to set the defaults
62 tailored to the system.
64 The driver supports fixed and variable block size (within buffer
65 limits). Both the auto-rewind (minor equals device number) and
66 non-rewind devices (minor is 128 + device number) are implemented.
68 In variable block mode, the byte count in write() determines the size
69 of the physical block on tape. When reading, the drive reads the next
70 tape block and returns to the user the data if the read() byte count
71 is at least the block size. Otherwise, error ENOMEM is returned.
73 In fixed block mode, the data transfer between the drive and the
74 driver is in multiples of the block size. The write() byte count must
75 be a multiple of the block size. This is not required when reading but
76 may be advisable for portability.
78 Support is provided for changing the tape partition and partitioning
79 of the tape with one or two partitions. By default support for
80 partitioned tape is disabled for each driver and it can be enabled
81 with the ioctl MTSETDRVBUFFER.
83 By default the driver writes one filemark when the device is closed after
84 writing and the last operation has been a write. Two filemarks can be
85 optionally written. In both cases end of data is signified by
86 returning zero bytes for two consecutive reads.
88 If rewind, offline, bsf, or seek is done and previous tape operation was
89 write, a filemark is written before moving tape.
91 The compile options are defined in the file linux/drivers/scsi/st_options.h.
93 4. If the open option O_NONBLOCK is used, open succeeds even if the
94 drive is not ready. If O_NONBLOCK is not used, the driver waits for
95 the drive to become ready. If this does not happen in ST_BLOCK_SECONDS
96 seconds, open fails with the errno value EIO. With O_NONBLOCK the
97 device can be opened for writing even if there is a write protected
98 tape in the drive (commands trying to write something return error if
104 The tape driver currently supports 128 drives by default. This number
105 can be increased by editing st.h and recompiling the driver if
106 necessary. The upper limit is 2^17 drives if 4 modes for each drive
109 The minor numbers consist of the following bit fields:
111 dev_upper non-rew mode dev-lower
113 The non-rewind bit is always bit 7 (the uppermost bit in the lowermost
114 byte). The bits defining the mode are below the non-rewind bit. The
115 remaining bits define the tape device number. This numbering is
116 backward compatible with the numbering used when the minor number was
122 The driver creates the directory /sys/class/scsi_tape and populates it with
123 directories corresponding to the existing tape devices. There are autorewind
124 and non-rewind entries for each mode. The names are stxy and nstxy, where x
125 is the tape number and y a character corresponding to the mode (none, l, m,
126 a). For example, the directories for the first tape device are (assuming four
127 modes): st0 nst0 st0l nst0l st0m nst0m st0a nst0a.
129 Each directory contains the entries: default_blksize default_compression
130 default_density defined dev device driver. The file 'defined' contains 1
131 if the mode is defined and zero if not defined. The files 'default_*' contain
132 the defaults set by the user. The value -1 means the default is not set. The
133 file 'dev' contains the device numbers corresponding to this device. The links
134 'device' and 'driver' point to the SCSI device and driver entries.
136 A link named 'tape' is made from the SCSI device directory to the class
137 directory corresponding to the mode 0 auto-rewind device (e.g., st0).
140 BSD AND SYS V SEMANTICS
142 The user can choose between these two behaviours of the tape driver by
143 defining the value of the symbol ST_SYSV. The semantics differ when a
144 file being read is closed. The BSD semantics leaves the tape where it
145 currently is whereas the SYS V semantics moves the tape past the next
146 filemark unless the filemark has just been crossed.
148 The default is BSD semantics.
153 The driver tries to do transfers directly to/from user space. If this
154 is not possible, a driver buffer allocated at run-time is used. If
155 direct i/o is not possible for the whole transfer, the driver buffer
156 is used (i.e., bounce buffers for individual pages are not
157 used). Direct i/o can be impossible because of several reasons, e.g.:
158 - one or more pages are at addresses not reachable by the HBA
159 - the number of pages in the transfer exceeds the number of
160 scatter/gather segments permitted by the HBA
161 - one or more pages can't be locked into memory (should not happen in
162 any reasonable situation)
164 The size of the driver buffers is always at least one tape block. In fixed
165 block mode, the minimum buffer size is defined (in 1024 byte units) by
166 ST_FIXED_BUFFER_BLOCKS. With small block size this allows buffering of
167 several blocks and using one SCSI read or write to transfer all of the
168 blocks. Buffering of data across write calls in fixed block mode is
169 allowed if ST_BUFFER_WRITES is non-zero and direct i/o is not used.
170 Buffer allocation uses chunks of memory having sizes 2^n * (page
171 size). Because of this the actual buffer size may be larger than the
172 minimum allowable buffer size.
174 NOTE that if direct i/o is used, the small writes are not buffered. This may
175 cause a surprise when moving from 2.4. There small writes (e.g., tar without
176 -b option) may have had good throughput but this is not true any more with
177 2.6. Direct i/o can be turned off to solve this problem but a better solution
178 is to use bigger write() byte counts (e.g., tar -b 64).
180 Asynchronous writing. Writing the buffer contents to the tape is
181 started and the write call returns immediately. The status is checked
182 at the next tape operation. Asynchronous writes are not done with
183 direct i/o and not in fixed block mode.
185 Buffered writes and asynchronous writes may in some rare cases cause
186 problems in multivolume operations if there is not enough space on the
187 tape after the early-warning mark to flush the driver buffer.
189 Read ahead for fixed block mode (ST_READ_AHEAD). Filling the buffer is
190 attempted even if the user does not want to get all of the data at
191 this read command. Should be disabled for those drives that don't like
192 a filemark to truncate a read request or that don't like backspacing.
194 Scatter/gather buffers (buffers that consist of chunks non-contiguous
195 in the physical memory) are used if contiguous buffers can't be
196 allocated. To support all SCSI adapters (including those not
197 supporting scatter/gather), buffer allocation is using the following
198 three kinds of chunks:
199 1. The initial segment that is used for all SCSI adapters including
200 those not supporting scatter/gather. The size of this buffer will be
201 (PAGE_SIZE << ST_FIRST_ORDER) bytes if the system can give a chunk of
202 this size (and it is not larger than the buffer size specified by
203 ST_BUFFER_BLOCKS). If this size is not available, the driver halves
204 the size and tries again until the size of one page. The default
205 settings in st_options.h make the driver to try to allocate all of the
207 2. The scatter/gather segments to fill the specified buffer size are
208 allocated so that as many segments as possible are used but the number
209 of segments does not exceed ST_FIRST_SG.
210 3. The remaining segments between ST_MAX_SG (or the module parameter
211 max_sg_segs) and the number of segments used in phases 1 and 2
212 are used to extend the buffer at run-time if this is necessary. The
213 number of scatter/gather segments allowed for the SCSI adapter is not
214 exceeded if it is smaller than the maximum number of scatter/gather
215 segments specified. If the maximum number allowed for the SCSI adapter
216 is smaller than the number of segments used in phases 1 and 2,
217 extending the buffer will always fail.
220 EOM BEHAVIOUR WHEN WRITING
222 When the end of medium early warning is encountered, the current write
223 is finished and the number of bytes is returned. The next write
224 returns -1 and errno is set to ENOSPC. To enable writing a trailer,
225 the next write is allowed to proceed and, if successful, the number of
226 bytes is returned. After this, -1 and the number of bytes are
227 alternately returned until the physical end of medium (or some other
228 error) is encountered.
233 The buffer size, write threshold, and the maximum number of allocated buffers
234 are configurable when the driver is loaded as a module. The keywords are:
236 buffer_kbs=xxx the buffer size for fixed block mode is set
238 write_threshold_kbs=xxx the write threshold in kilobytes set to xxx
239 max_sg_segs=xxx the maximum number of scatter/gather
241 try_direct_io=x try direct transfer between user buffer and
242 tape drive if this is non-zero
244 Note that if the buffer size is changed but the write threshold is not
245 set, the write threshold is set to the new buffer size - 2 kB.
248 BOOT TIME CONFIGURATION
250 If the driver is compiled into the kernel, the same parameters can be
251 also set using, e.g., the LILO command line. The preferred syntax is
252 to use the same keyword used when loading as module but prepended
253 with 'st.'. For instance, to set the maximum number of scatter/gather
254 segments, the parameter 'st.max_sg_segs=xx' should be used (xx is the
255 number of scatter/gather segments).
257 For compatibility, the old syntax from early 2.5 and 2.4 kernel
258 versions is supported. The same keywords can be used as when loading
259 the driver as module. If several parameters are set, the keyword-value
260 pairs are separated with a comma (no spaces allowed). A colon can be
261 used instead of the equal mark. The definition is prepended by the
262 string st=. Here is an example:
264 st=buffer_kbs:64,write_threshold_kbs:60
266 The following syntax used by the old kernel versions is also supported:
271 aa is the buffer size for fixed block mode in 1024 byte units
272 bb is the write threshold in 1024 byte units
273 dd is the maximum number of scatter/gather segments
278 The tape is positioned and the drive parameters are set with ioctls
279 defined in mtio.h The tape control program 'mt' uses these ioctls. Try
280 to find an mt that supports all of the Linux SCSI tape ioctls and
281 opens the device for writing if the tape contents will be modified
282 (look for a package mt-st* from the Linux ftp sites; the GNU mt does
283 not open for writing for, e.g., erase).
285 The supported ioctls are:
287 The following use the structure mtop:
289 MTFSF Space forward over count filemarks. Tape positioned after filemark.
290 MTFSFM As above but tape positioned before filemark.
291 MTBSF Space backward over count filemarks. Tape positioned before
293 MTBSFM As above but ape positioned after filemark.
294 MTFSR Space forward over count records.
295 MTBSR Space backward over count records.
296 MTFSS Space forward over count setmarks.
297 MTBSS Space backward over count setmarks.
298 MTWEOF Write count filemarks.
299 MTWSM Write count setmarks.
301 MTOFFL Set device off line (often rewind plus eject).
302 MTNOP Do nothing except flush the buffers.
303 MTRETEN Re-tension tape.
304 MTEOM Space to end of recorded data.
305 MTERASE Erase tape. If the argument is zero, the short erase command
306 is used. The long erase command is used with all other values
308 MTSEEK Seek to tape block count. Uses Tandberg-compatible seek (QFA)
309 for SCSI-1 drives and SCSI-2 seek for SCSI-2 drives. The file and
310 block numbers in the status are not valid after a seek.
311 MTSETBLK Set the drive block size. Setting to zero sets the drive into
312 variable block mode (if applicable).
313 MTSETDENSITY Sets the drive density code to arg. See drive
314 documentation for available codes.
315 MTLOCK and MTUNLOCK Explicitly lock/unlock the tape drive door.
316 MTLOAD and MTUNLOAD Explicitly load and unload the tape. If the
317 command argument x is between MT_ST_HPLOADER_OFFSET + 1 and
318 MT_ST_HPLOADER_OFFSET + 6, the number x is used sent to the
319 drive with the command and it selects the tape slot to use of
321 MTCOMPRESSION Sets compressing or uncompressing drive mode using the
322 SCSI mode page 15. Note that some drives other methods for
323 control of compression. Some drives (like the Exabytes) use
324 density codes for compression control. Some drives use another
325 mode page but this page has not been implemented in the
326 driver. Some drives without compression capability will accept
327 any compression mode without error.
328 MTSETPART Moves the tape to the partition given by the argument at the
329 next tape operation. The block at which the tape is positioned
330 is the block where the tape was previously positioned in the
331 new active partition unless the next tape operation is
332 MTSEEK. In this case the tape is moved directly to the block
333 specified by MTSEEK. MTSETPART is inactive unless
334 MT_ST_CAN_PARTITIONS set.
335 MTMKPART Formats the tape with one partition (argument zero) or two
336 partitions (the argument gives in megabytes the size of
337 partition 1 that is physically the first partition of the
338 tape). The drive has to support partitions with size specified
339 by the initiator. Inactive unless MT_ST_CAN_PARTITIONS set.
341 Is used for several purposes. The command is obtained from count
342 with mask MT_SET_OPTIONS, the low order bits are used as argument.
343 This command is only allowed for the superuser (root). The
346 The drive buffer option is set to the argument. Zero means
349 Sets the buffering options. The bits are the new states
350 (enabled/disabled) the following options (in the
351 parenthesis is specified whether the option is global or
352 can be specified differently for each mode):
353 MT_ST_BUFFER_WRITES write buffering (mode)
354 MT_ST_ASYNC_WRITES asynchronous writes (mode)
355 MT_ST_READ_AHEAD read ahead (mode)
356 MT_ST_TWO_FM writing of two filemarks (global)
357 MT_ST_FAST_EOM using the SCSI spacing to EOD (global)
358 MT_ST_AUTO_LOCK automatic locking of the drive door (global)
359 MT_ST_DEF_WRITES the defaults are meant only for writes (mode)
360 MT_ST_CAN_BSR backspacing over more than one records can
361 be used for repositioning the tape (global)
362 MT_ST_NO_BLKLIMS the driver does not ask the block limits
363 from the drive (block size can be changed only to
365 MT_ST_CAN_PARTITIONS enables support for partitioned
367 MT_ST_SCSI2LOGICAL the logical block number is used in
368 the MTSEEK and MTIOCPOS for SCSI-2 drives instead of
369 the device dependent address. It is recommended to set
370 this flag unless there are tapes using the device
371 dependent (from the old times) (global)
372 MT_ST_SYSV sets the SYSV semantics (mode)
373 MT_ST_NOWAIT enables immediate mode (i.e., don't wait for
374 the command to finish) for some commands (e.g., rewind)
375 MT_ST_DEBUGGING debugging (global; debugging must be
376 compiled into the driver)
379 Sets or clears the option bits.
380 MT_ST_WRITE_THRESHOLD
381 Sets the write threshold for this device to kilobytes
382 specified by the lowest bits.
384 Defines the default block size set automatically. Value
385 0xffffff means that the default is not used any more.
388 Used to set or clear the density (8 bits), and drive buffer
389 state (3 bits). If the value is MT_ST_CLEAR_DEFAULT
390 (0xfffff) the default will not be used any more. Otherwise
391 the lowermost bits of the value contain the new value of
393 MT_ST_DEF_COMPRESSION
394 The compression default will not be used if the value of
395 the lowermost byte is 0xff. Otherwise the lowermost bit
396 contains the new default. If the bits 8-15 are set to a
397 non-zero number, and this number is not 0xff, the number is
398 used as the compression algorithm. The value
399 MT_ST_CLEAR_DEFAULT can be used to clear the compression
402 Set the normal timeout in seconds for this device. The
403 default is 900 seconds (15 minutes). The timeout should be
404 long enough for the retries done by the device while
406 MT_ST_SET_LONG_TIMEOUT
407 Set the long timeout that is used for operations that are
408 known to take a long time. The default is 14000 seconds
409 (3.9 hours). For erase this value is further multiplied by
412 Set the cleaning request interpretation parameters using
413 the lowest 24 bits of the argument. The driver can set the
414 generic status bit GMT_CLN if a cleaning request bit pattern
415 is found from the extended sense data. Many drives set one or
416 more bits in the extended sense data when the drive needs
417 cleaning. The bits are device-dependent. The driver is
418 given the number of the sense data byte (the lowest eight
419 bits of the argument; must be >= 18 (values 1 - 17
420 reserved) and <= the maximum requested sense data sixe),
421 a mask to select the relevant bits (the bits 9-16), and the
422 bit pattern (bits 17-23). If the bit pattern is zero, one
423 or more bits under the mask indicate cleaning request. If
424 the pattern is non-zero, the pattern must match the masked
427 (The cleaning bit is set if the additional sense code and
428 qualifier 00h 17h are seen regardless of the setting of
431 The following ioctl uses the structure mtpos:
432 MTIOCPOS Reads the current position from the drive. Uses
433 Tandberg-compatible QFA for SCSI-1 drives and the SCSI-2
434 command for the SCSI-2 drives.
436 The following ioctl uses the structure mtget to return the status:
437 MTIOCGET Returns some status information.
438 The file number and block number within file are returned. The
439 block is -1 when it can't be determined (e.g., after MTBSF).
440 The drive type is either MTISSCSI1 or MTISSCSI2.
441 The number of recovered errors since the previous status call
442 is stored in the lower word of the field mt_erreg.
443 The current block size and the density code are stored in the field
444 mt_dsreg (shifts for the subfields are MT_ST_BLKSIZE_SHIFT and
445 MT_ST_DENSITY_SHIFT).
446 The GMT_xxx status bits reflect the drive status. GMT_DR_OPEN
447 is set if there is no tape in the drive. GMT_EOD means either
448 end of recorded data or end of tape. GMT_EOT means end of tape.
451 MISCELLANEOUS COMPILE OPTIONS
453 The recovered write errors are considered fatal if ST_RECOVERED_WRITE_FATAL
456 The maximum number of tape devices is determined by the define
457 ST_MAX_TAPES. If more tapes are detected at driver initialization, the
458 maximum is adjusted accordingly.
460 Immediate return from tape positioning SCSI commands can be enabled by
461 defining ST_NOWAIT. If this is defined, the user should take care that
462 the next tape operation is not started before the previous one has
463 finished. The drives and SCSI adapters should handle this condition
464 gracefully, but some drive/adapter combinations are known to hang the
465 SCSI bus in this case.
467 The MTEOM command is by default implemented as spacing over 32767
468 filemarks. With this method the file number in the status is
469 correct. The user can request using direct spacing to EOD by setting
470 ST_FAST_EOM 1 (or using the MT_ST_OPTIONS ioctl). In this case the file
471 number will be invalid.
473 When using read ahead or buffered writes the position within the file
474 may not be correct after the file is closed (correct position may
475 require backspacing over more than one record). The correct position
476 within file can be obtained if ST_IN_FILE_POS is defined at compile
477 time or the MT_ST_CAN_BSR bit is set for the drive with an ioctl.
478 (The driver always backs over a filemark crossed by read ahead if the
479 user does not request data that far.)
484 To enable debugging messages, edit st.c and #define DEBUG 1. As seen
485 above, debugging can be switched off with an ioctl if debugging is
486 compiled into the driver. The debugging output is not voluminous.
488 If the tape seems to hang, I would be very interested to hear where
489 the driver is waiting. With the command 'ps -l' you can see the state
490 of the process using the tape. If the state is D, the process is
491 waiting for something. The field WCHAN tells where the driver is
492 waiting. If you have the current System.map in the correct place (in
493 /boot for the procps I use) or have updated /etc/psdatabase (for kmem
494 ps), ps writes the function name in the WCHAN field. If not, you have
495 to look up the function from System.map.
497 Note also that the timeouts are very long compared to most other
498 drivers. This means that the Linux driver may appear hung although the
499 real reason is that the tape firmware has got confused.
projects / linux-2.6 / blob