Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1.. SPDX-License-Identifier: GPL-2.0
2
3====================
4The SCSI Tape Driver
5====================
6
7This file contains brief information about the SCSI tape driver.
8The driver is currently maintained by Kai Mäkisara (email
9Kai.Makisara@kolumbus.fi)
10
11Last modified: Tue Feb 9 21:54:16 2016 by kai.makisara
12
13
14Basics
15======
16
17The driver is generic, i.e., it does not contain any code tailored
18to any specific tape drive. The tape parameters can be specified with
19one of the following three methods:
20
211. Each user can specify the tape parameters he/she wants to use
22directly with ioctls. This is administratively a very simple and
23flexible method and applicable to single-user workstations. However,
24in a multiuser environment the next user finds the tape parameters in
25state the previous user left them.
26
272. The system manager (root) can define default values for some tape
28parameters, like block size and density using the MTSETDRVBUFFER ioctl.
29These parameters can be programmed to come into effect either when a
30new tape is loaded into the drive or if writing begins at the
31beginning of the tape. The second method is applicable if the tape
32drive performs auto-detection of the tape format well (like some
33QIC-drives). The result is that any tape can be read, writing can be
34continued using existing format, and the default format is used if
35the tape is rewritten from the beginning (or a new tape is written
36for the first time). The first method is applicable if the drive
37does not perform auto-detection well enough and there is a single
38"sensible" mode for the device. An example is a DAT drive that is
39used only in variable block mode (I don't know if this is sensible
40or not :-).
41
42The user can override the parameters defined by the system
43manager. The changes persist until the defaults again come into
44effect.
45
463. By default, up to four modes can be defined and selected using the minor
47number (bits 5 and 6). The number of modes can be changed by changing
48ST_NBR_MODE_BITS in st.h. Mode 0 corresponds to the defaults discussed
49above. Additional modes are dormant until they are defined by the
50system manager (root). When specification of a new mode is started,
51the configuration of mode 0 is used to provide a starting point for
52definition of the new mode.
53
54Using the modes allows the system manager to give the users choices
55over some of the buffering parameters not directly accessible to the
56users (buffered and asynchronous writes). The modes also allow choices
57between formats in multi-tape operations (the explicitly overridden
58parameters are reset when a new tape is loaded).
59
60If more than one mode is used, all modes should contain definitions
61for the same set of parameters.
62
63Many Unices contain internal tables that associate different modes to
64supported devices. The Linux SCSI tape driver does not contain such
65tables (and will not do that in future). Instead of that, a utility
66program can be made that fetches the inquiry data sent by the device,
67scans its database, and sets up the modes using the ioctls. Another
68alternative is to make a small script that uses mt to set the defaults
69tailored to the system.
70
71The driver supports fixed and variable block size (within buffer
72limits). Both the auto-rewind (minor equals device number) and
73non-rewind devices (minor is 128 + device number) are implemented.
74
75In variable block mode, the byte count in write() determines the size
76of the physical block on tape. When reading, the drive reads the next
77tape block and returns to the user the data if the read() byte count
78is at least the block size. Otherwise, error ENOMEM is returned.
79
80In fixed block mode, the data transfer between the drive and the
81driver is in multiples of the block size. The write() byte count must
82be a multiple of the block size. This is not required when reading but
83may be advisable for portability.
84
85Support is provided for changing the tape partition and partitioning
86of the tape with one or two partitions. By default support for
87partitioned tape is disabled for each driver and it can be enabled
88with the ioctl MTSETDRVBUFFER.
89
90By default the driver writes one filemark when the device is closed after
91writing and the last operation has been a write. Two filemarks can be
92optionally written. In both cases end of data is signified by
93returning zero bytes for two consecutive reads.
94
95Writing filemarks without the immediate bit set in the SCSI command block acts
96as a synchronization point, i.e., all remaining data form the drive buffers is
97written to tape before the command returns. This makes sure that write errors
98are caught at that point, but this takes time. In some applications, several
99consecutive files must be written fast. The MTWEOFI operation can be used to
100write the filemarks without flushing the drive buffer. Writing filemark at
101close() is always flushing the drive buffers. However, if the previous
102operation is MTWEOFI, close() does not write a filemark. This can be used if
103the program wants to close/open the tape device between files and wants to
104skip waiting.
105
106If rewind, offline, bsf, or seek is done and previous tape operation was
107write, a filemark is written before moving tape.
108
109The compile options are defined in the file linux/drivers/scsi/st_options.h.
110
1114. If the open option O_NONBLOCK is used, open succeeds even if the
112drive is not ready. If O_NONBLOCK is not used, the driver waits for
113the drive to become ready. If this does not happen in ST_BLOCK_SECONDS
114seconds, open fails with the errno value EIO. With O_NONBLOCK the
115device can be opened for writing even if there is a write protected
116tape in the drive (commands trying to write something return error if
117attempted).
118
119
120Minor Numbers
121=============
122
123The tape driver currently supports up to 2^17 drives if 4 modes for
124each drive are used.
125
126The minor numbers consist of the following bit fields::
127
128 dev_upper non-rew mode dev-lower
129 20 - 8 7 6 5 4 0
130
131The non-rewind bit is always bit 7 (the uppermost bit in the lowermost
132byte). The bits defining the mode are below the non-rewind bit. The
133remaining bits define the tape device number. This numbering is
134backward compatible with the numbering used when the minor number was
135only 8 bits wide.
136
137
138Sysfs Support
139=============
140
141The driver creates the directory /sys/class/scsi_tape and populates it with
142directories corresponding to the existing tape devices. There are autorewind
143and non-rewind entries for each mode. The names are stxy and nstxy, where x
144is the tape number and y a character corresponding to the mode (none, l, m,
145a). For example, the directories for the first tape device are (assuming four
146modes): st0 nst0 st0l nst0l st0m nst0m st0a nst0a.
147
148Each directory contains the entries: default_blksize default_compression
149default_density defined dev device driver. The file 'defined' contains 1
150if the mode is defined and zero if not defined. The files 'default_*' contain
151the defaults set by the user. The value -1 means the default is not set. The
152file 'dev' contains the device numbers corresponding to this device. The links
153'device' and 'driver' point to the SCSI device and driver entries.
154
155Each directory also contains the entry 'options' which shows the currently
156enabled driver and mode options. The value in the file is a bit mask where the
157bit definitions are the same as those used with MTSETDRVBUFFER in setting the
158options.
159
160Each directory contains the entry 'position_lost_in_reset'. If this value is
161one, reading and writing to the device is blocked after device reset. Most
162devices rewind the tape after reset and the writes/read don't access the
163tape position the user expects.
164
165A link named 'tape' is made from the SCSI device directory to the class
166directory corresponding to the mode 0 auto-rewind device (e.g., st0).
167
168
169Sysfs and Statistics for Tape Devices
170=====================================
171
172The st driver maintains statistics for tape drives inside the sysfs filesystem.
173The following method can be used to locate the statistics that are
174available (assuming that sysfs is mounted at /sys):
175
1761. Use opendir(3) on the directory /sys/class/scsi_tape
1772. Use readdir(3) to read the directory contents
1783. Use regcomp(3)/regexec(3) to match directory entries to the extended
179 regular expression "^st[0-9]+$"
1804. Access the statistics from the /sys/class/scsi_tape/<match>/stats
181 directory (where <match> is a directory entry from /sys/class/scsi_tape
182 that matched the extended regular expression)
183
184The reason for using this approach is that all the character devices
185pointing to the same tape drive use the same statistics. That means
186that st0 would have the same statistics as nst0.
187
188The directory contains the following statistics files:
189
1901. in_flight
191 - The number of I/Os currently outstanding to this device.
1922. io_ns
193 - The amount of time spent waiting (in nanoseconds) for all I/O
194 to complete (including read and write). This includes tape movement
195 commands such as seeking between file or set marks and implicit tape
196 movement such as when rewind on close tape devices are used.
1973. other_cnt
198 - The number of I/Os issued to the tape drive other than read or
199 write commands. The time taken to complete these commands uses the
200 following calculation io_ms-read_ms-write_ms.
2014. read_byte_cnt
202 - The number of bytes read from the tape drive.
2035. read_cnt
204 - The number of read requests issued to the tape drive.
2056. read_ns
206 - The amount of time (in nanoseconds) spent waiting for read
207 requests to complete.
2087. write_byte_cnt
209 - The number of bytes written to the tape drive.
2108. write_cnt
211 - The number of write requests issued to the tape drive.
2129. write_ns
213 - The amount of time (in nanoseconds) spent waiting for write
214 requests to complete.
21510. resid_cnt
216 - The number of times during a read or write we found
217 the residual amount to be non-zero. This should mean that a program
218 is issuing a read larger thean the block size on tape. For write
219 not all data made it to tape.
220
221.. Note::
222
223 The in_flight value is incremented when an I/O starts the I/O
224 itself is not added to the statistics until it completes.
225
226The total of read_cnt, write_cnt, and other_cnt may not total to the same
227value as iodone_cnt at the device level. The tape statistics only count
228I/O issued via the st module.
229
230When read the statistics may not be temporally consistent while I/O is in
231progress. The individual values are read and written to atomically however
232when reading them back via sysfs they may be in the process of being
233updated when starting an I/O or when it is completed.
234
235The value shown in in_flight is incremented before any statstics are
236updated and decremented when an I/O completes after updating statistics.
237The value of in_flight is 0 when there are no I/Os outstanding that are
238issued by the st driver. Tape statistics do not take into account any
239I/O performed via the sg device.
240
241BSD and Sys V Semantics
242=======================
243
244The user can choose between these two behaviours of the tape driver by
245defining the value of the symbol ST_SYSV. The semantics differ when a
246file being read is closed. The BSD semantics leaves the tape where it
247currently is whereas the SYS V semantics moves the tape past the next
248filemark unless the filemark has just been crossed.
249
250The default is BSD semantics.
251
252
253Buffering
254=========
255
256The driver tries to do transfers directly to/from user space. If this
257is not possible, a driver buffer allocated at run-time is used. If
258direct i/o is not possible for the whole transfer, the driver buffer
259is used (i.e., bounce buffers for individual pages are not
260used). Direct i/o can be impossible because of several reasons, e.g.:
261
262- one or more pages are at addresses not reachable by the HBA
263- the number of pages in the transfer exceeds the number of
264 scatter/gather segments permitted by the HBA
265- one or more pages can't be locked into memory (should not happen in
266 any reasonable situation)
267
268The size of the driver buffers is always at least one tape block. In fixed
269block mode, the minimum buffer size is defined (in 1024 byte units) by
270ST_FIXED_BUFFER_BLOCKS. With small block size this allows buffering of
271several blocks and using one SCSI read or write to transfer all of the
272blocks. Buffering of data across write calls in fixed block mode is
273allowed if ST_BUFFER_WRITES is non-zero and direct i/o is not used.
274Buffer allocation uses chunks of memory having sizes 2^n * (page
275size). Because of this the actual buffer size may be larger than the
276minimum allowable buffer size.
277
278NOTE that if direct i/o is used, the small writes are not buffered. This may
279cause a surprise when moving from 2.4. There small writes (e.g., tar without
280-b option) may have had good throughput but this is not true any more with
2812.6. Direct i/o can be turned off to solve this problem but a better solution
282is to use bigger write() byte counts (e.g., tar -b 64).
283
284Asynchronous writing. Writing the buffer contents to the tape is
285started and the write call returns immediately. The status is checked
286at the next tape operation. Asynchronous writes are not done with
287direct i/o and not in fixed block mode.
288
289Buffered writes and asynchronous writes may in some rare cases cause
290problems in multivolume operations if there is not enough space on the
291tape after the early-warning mark to flush the driver buffer.
292
293Read ahead for fixed block mode (ST_READ_AHEAD). Filling the buffer is
294attempted even if the user does not want to get all of the data at
295this read command. Should be disabled for those drives that don't like
296a filemark to truncate a read request or that don't like backspacing.
297
298Scatter/gather buffers (buffers that consist of chunks non-contiguous
299in the physical memory) are used if contiguous buffers can't be
300allocated. To support all SCSI adapters (including those not
301supporting scatter/gather), buffer allocation is using the following
302three kinds of chunks:
303
3041. The initial segment that is used for all SCSI adapters including
305 those not supporting scatter/gather. The size of this buffer will be
306 (PAGE_SIZE << ST_FIRST_ORDER) bytes if the system can give a chunk of
307 this size (and it is not larger than the buffer size specified by
308 ST_BUFFER_BLOCKS). If this size is not available, the driver halves
309 the size and tries again until the size of one page. The default
310 settings in st_options.h make the driver to try to allocate all of the
311 buffer as one chunk.
3122. The scatter/gather segments to fill the specified buffer size are
313 allocated so that as many segments as possible are used but the number
314 of segments does not exceed ST_FIRST_SG.
3153. The remaining segments between ST_MAX_SG (or the module parameter
316 max_sg_segs) and the number of segments used in phases 1 and 2
317 are used to extend the buffer at run-time if this is necessary. The
318 number of scatter/gather segments allowed for the SCSI adapter is not
319 exceeded if it is smaller than the maximum number of scatter/gather
320 segments specified. If the maximum number allowed for the SCSI adapter
321 is smaller than the number of segments used in phases 1 and 2,
322 extending the buffer will always fail.
323
324
325EOM Behaviour When Writing
326==========================
327
328When the end of medium early warning is encountered, the current write
329is finished and the number of bytes is returned. The next write
330returns -1 and errno is set to ENOSPC. To enable writing a trailer,
331the next write is allowed to proceed and, if successful, the number of
332bytes is returned. After this, -1 and the number of bytes are
333alternately returned until the physical end of medium (or some other
334error) is encountered.
335
336Module Parameters
337=================
338
339The buffer size, write threshold, and the maximum number of allocated buffers
340are configurable when the driver is loaded as a module. The keywords are:
341
342========================== ===========================================
343buffer_kbs=xxx the buffer size for fixed block mode is set
344 to xxx kilobytes
345write_threshold_kbs=xxx the write threshold in kilobytes set to xxx
346max_sg_segs=xxx the maximum number of scatter/gather
347 segments
348try_direct_io=x try direct transfer between user buffer and
349 tape drive if this is non-zero
350========================== ===========================================
351
352Note that if the buffer size is changed but the write threshold is not
353set, the write threshold is set to the new buffer size - 2 kB.
354
355
356Boot Time Configuration
357=======================
358
359If the driver is compiled into the kernel, the same parameters can be
360also set using, e.g., the LILO command line. The preferred syntax is
361to use the same keyword used when loading as module but prepended
362with 'st.'. For instance, to set the maximum number of scatter/gather
363segments, the parameter 'st.max_sg_segs=xx' should be used (xx is the
364number of scatter/gather segments).
365
366For compatibility, the old syntax from early 2.5 and 2.4 kernel
367versions is supported. The same keywords can be used as when loading
368the driver as module. If several parameters are set, the keyword-value
369pairs are separated with a comma (no spaces allowed). A colon can be
370used instead of the equal mark. The definition is prepended by the
371string st=. Here is an example::
372
373 st=buffer_kbs:64,write_threshold_kbs:60
374
375The following syntax used by the old kernel versions is also supported::
376
377 st=aa[,bb[,dd]]
378
379where:
380
381 - aa is the buffer size for fixed block mode in 1024 byte units
382 - bb is the write threshold in 1024 byte units
383 - dd is the maximum number of scatter/gather segments
384
385
386IOCTLs
387======
388
389The tape is positioned and the drive parameters are set with ioctls
390defined in mtio.h The tape control program 'mt' uses these ioctls. Try
391to find an mt that supports all of the Linux SCSI tape ioctls and
392opens the device for writing if the tape contents will be modified
393(look for a package mt-st* from the Linux ftp sites; the GNU mt does
394not open for writing for, e.g., erase).
395
396The supported ioctls are:
397
398The following use the structure mtop:
399
400MTFSF
401 Space forward over count filemarks. Tape positioned after filemark.
402MTFSFM
403 As above but tape positioned before filemark.
404MTBSF
405 Space backward over count filemarks. Tape positioned before
406 filemark.
407MTBSFM
408 As above but ape positioned after filemark.
409MTFSR
410 Space forward over count records.
411MTBSR
412 Space backward over count records.
413MTFSS
414 Space forward over count setmarks.
415MTBSS
416 Space backward over count setmarks.
417MTWEOF
418 Write count filemarks.
419MTWEOFI
420 Write count filemarks with immediate bit set (i.e., does not
421 wait until data is on tape)
422MTWSM
423 Write count setmarks.
424MTREW
425 Rewind tape.
426MTOFFL
427 Set device off line (often rewind plus eject).
428MTNOP
429 Do nothing except flush the buffers.
430MTRETEN
431 Re-tension tape.
432MTEOM
433 Space to end of recorded data.
434MTERASE
435 Erase tape. If the argument is zero, the short erase command
436 is used. The long erase command is used with all other values
437 of the argument.
438MTSEEK
439 Seek to tape block count. Uses Tandberg-compatible seek (QFA)
440 for SCSI-1 drives and SCSI-2 seek for SCSI-2 drives. The file and
441 block numbers in the status are not valid after a seek.
442MTSETBLK
443 Set the drive block size. Setting to zero sets the drive into
444 variable block mode (if applicable).
445MTSETDENSITY
446 Sets the drive density code to arg. See drive
447 documentation for available codes.
448MTLOCK and MTUNLOCK
449 Explicitly lock/unlock the tape drive door.
450MTLOAD and MTUNLOAD
451 Explicitly load and unload the tape. If the
452 command argument x is between MT_ST_HPLOADER_OFFSET + 1 and
453 MT_ST_HPLOADER_OFFSET + 6, the number x is used sent to the
454 drive with the command and it selects the tape slot to use of
455 HP C1553A changer.
456MTCOMPRESSION
457 Sets compressing or uncompressing drive mode using the
458 SCSI mode page 15. Note that some drives other methods for
459 control of compression. Some drives (like the Exabytes) use
460 density codes for compression control. Some drives use another
461 mode page but this page has not been implemented in the
462 driver. Some drives without compression capability will accept
463 any compression mode without error.
464MTSETPART
465 Moves the tape to the partition given by the argument at the
466 next tape operation. The block at which the tape is positioned
467 is the block where the tape was previously positioned in the
468 new active partition unless the next tape operation is
469 MTSEEK. In this case the tape is moved directly to the block
470 specified by MTSEEK. MTSETPART is inactive unless
471 MT_ST_CAN_PARTITIONS set.
472MTMKPART
473 Formats the tape with one partition (argument zero) or two
474 partitions (argument non-zero). If the argument is positive,
475 it specifies the size of partition 1 in megabytes. For DDS
476 drives and several early drives this is the physically first
477 partition of the tape. If the argument is negative, its absolute
478 value specifies the size of partition 0 in megabytes. This is
479 the physically first partition of many later drives, like the
480 LTO drives from LTO-5 upwards. The drive has to support partitions
481 with size specified by the initiator. Inactive unless
482 MT_ST_CAN_PARTITIONS set.
483MTSETDRVBUFFER
484 Is used for several purposes. The command is obtained from count
485 with mask MT_SET_OPTIONS, the low order bits are used as argument.
486 This command is only allowed for the superuser (root). The
487 subcommands are:
488
489 * 0
490 The drive buffer option is set to the argument. Zero means
491 no buffering.
492 * MT_ST_BOOLEANS
493 Sets the buffering options. The bits are the new states
494 (enabled/disabled) the following options (in the
495 parenthesis is specified whether the option is global or
496 can be specified differently for each mode):
497
498 MT_ST_BUFFER_WRITES
499 write buffering (mode)
500 MT_ST_ASYNC_WRITES
501 asynchronous writes (mode)
502 MT_ST_READ_AHEAD
503 read ahead (mode)
504 MT_ST_TWO_FM
505 writing of two filemarks (global)
506 MT_ST_FAST_EOM
507 using the SCSI spacing to EOD (global)
508 MT_ST_AUTO_LOCK
509 automatic locking of the drive door (global)
510 MT_ST_DEF_WRITES
511 the defaults are meant only for writes (mode)
512 MT_ST_CAN_BSR
513 backspacing over more than one records can
514 be used for repositioning the tape (global)
515 MT_ST_NO_BLKLIMS
516 the driver does not ask the block limits
517 from the drive (block size can be changed only to
518 variable) (global)
519 MT_ST_CAN_PARTITIONS
520 enables support for partitioned
521 tapes (global)
522 MT_ST_SCSI2LOGICAL
523 the logical block number is used in
524 the MTSEEK and MTIOCPOS for SCSI-2 drives instead of
525 the device dependent address. It is recommended to set
526 this flag unless there are tapes using the device
527 dependent (from the old times) (global)
528 MT_ST_SYSV
529 sets the SYSV semantics (mode)
530 MT_ST_NOWAIT
531 enables immediate mode (i.e., don't wait for
532 the command to finish) for some commands (e.g., rewind)
533 MT_ST_NOWAIT_EOF
534 enables immediate filemark mode (i.e. when
535 writing a filemark, don't wait for it to complete). Please
536 see the BASICS note about MTWEOFI with respect to the
537 possible dangers of writing immediate filemarks.
538 MT_ST_SILI
539 enables setting the SILI bit in SCSI commands when
540 reading in variable block mode to enhance performance when
541 reading blocks shorter than the byte count; set this only
542 if you are sure that the drive supports SILI and the HBA
543 correctly returns transfer residuals
544 MT_ST_DEBUGGING
545 debugging (global; debugging must be
546 compiled into the driver)
547
548 * MT_ST_SETBOOLEANS, MT_ST_CLEARBOOLEANS
549 Sets or clears the option bits.
550 * MT_ST_WRITE_THRESHOLD
551 Sets the write threshold for this device to kilobytes
552 specified by the lowest bits.
553 * MT_ST_DEF_BLKSIZE
554 Defines the default block size set automatically. Value
555 0xffffff means that the default is not used any more.
556 * MT_ST_DEF_DENSITY, MT_ST_DEF_DRVBUFFER
557 Used to set or clear the density (8 bits), and drive buffer
558 state (3 bits). If the value is MT_ST_CLEAR_DEFAULT
559 (0xfffff) the default will not be used any more. Otherwise
560 the lowermost bits of the value contain the new value of
561 the parameter.
562 * MT_ST_DEF_COMPRESSION
563 The compression default will not be used if the value of
564 the lowermost byte is 0xff. Otherwise the lowermost bit
565 contains the new default. If the bits 8-15 are set to a
566 non-zero number, and this number is not 0xff, the number is
567 used as the compression algorithm. The value
568 MT_ST_CLEAR_DEFAULT can be used to clear the compression
569 default.
570 * MT_ST_SET_TIMEOUT
571 Set the normal timeout in seconds for this device. The
572 default is 900 seconds (15 minutes). The timeout should be
573 long enough for the retries done by the device while
574 reading/writing.
575 * MT_ST_SET_LONG_TIMEOUT
576 Set the long timeout that is used for operations that are
577 known to take a long time. The default is 14000 seconds
578 (3.9 hours). For erase this value is further multiplied by
579 eight.
580 * MT_ST_SET_CLN
581 Set the cleaning request interpretation parameters using
582 the lowest 24 bits of the argument. The driver can set the
583 generic status bit GMT_CLN if a cleaning request bit pattern
584 is found from the extended sense data. Many drives set one or
585 more bits in the extended sense data when the drive needs
586 cleaning. The bits are device-dependent. The driver is
587 given the number of the sense data byte (the lowest eight
588 bits of the argument; must be >= 18 (values 1 - 17
589 reserved) and <= the maximum requested sense data sixe),
590 a mask to select the relevant bits (the bits 9-16), and the
591 bit pattern (bits 17-23). If the bit pattern is zero, one
592 or more bits under the mask indicate cleaning request. If
593 the pattern is non-zero, the pattern must match the masked
594 sense data byte.
595
596 (The cleaning bit is set if the additional sense code and
597 qualifier 00h 17h are seen regardless of the setting of
598 MT_ST_SET_CLN.)
599
600The following ioctl uses the structure mtpos:
601
602MTIOCPOS
603 Reads the current position from the drive. Uses
604 Tandberg-compatible QFA for SCSI-1 drives and the SCSI-2
605 command for the SCSI-2 drives.
606
607The following ioctl uses the structure mtget to return the status:
608
609MTIOCGET
610 Returns some status information.
611 The file number and block number within file are returned. The
612 block is -1 when it can't be determined (e.g., after MTBSF).
613 The drive type is either MTISSCSI1 or MTISSCSI2.
614 The number of recovered errors since the previous status call
615 is stored in the lower word of the field mt_erreg.
616 The current block size and the density code are stored in the field
617 mt_dsreg (shifts for the subfields are MT_ST_BLKSIZE_SHIFT and
618 MT_ST_DENSITY_SHIFT).
619 The GMT_xxx status bits reflect the drive status. GMT_DR_OPEN
620 is set if there is no tape in the drive. GMT_EOD means either
621 end of recorded data or end of tape. GMT_EOT means end of tape.
622
623
624Miscellaneous Compile Options
625=============================
626
627The recovered write errors are considered fatal if ST_RECOVERED_WRITE_FATAL
628is defined.
629
630The maximum number of tape devices is determined by the define
631ST_MAX_TAPES. If more tapes are detected at driver initialization, the
632maximum is adjusted accordingly.
633
634Immediate return from tape positioning SCSI commands can be enabled by
635defining ST_NOWAIT. If this is defined, the user should take care that
636the next tape operation is not started before the previous one has
637finished. The drives and SCSI adapters should handle this condition
638gracefully, but some drive/adapter combinations are known to hang the
639SCSI bus in this case.
640
641The MTEOM command is by default implemented as spacing over 32767
642filemarks. With this method the file number in the status is
643correct. The user can request using direct spacing to EOD by setting
644ST_FAST_EOM 1 (or using the MT_ST_OPTIONS ioctl). In this case the file
645number will be invalid.
646
647When using read ahead or buffered writes the position within the file
648may not be correct after the file is closed (correct position may
649require backspacing over more than one record). The correct position
650within file can be obtained if ST_IN_FILE_POS is defined at compile
651time or the MT_ST_CAN_BSR bit is set for the drive with an ioctl.
652(The driver always backs over a filemark crossed by read ahead if the
653user does not request data that far.)
654
655
656Debugging Hints
657===============
658
659Debugging code is now compiled in by default but debugging is turned off
660with the kernel module parameter debug_flag defaulting to 0. Debugging
661can still be switched on and off with an ioctl. To enable debug at
662module load time add debug_flag=1 to the module load options, the
663debugging output is not voluminous. Debugging can also be enabled
664and disabled by writing a '0' (disable) or '1' (enable) to the sysfs
665file /sys/bus/scsi/drivers/st/debug_flag.
666
667If the tape seems to hang, I would be very interested to hear where
668the driver is waiting. With the command 'ps -l' you can see the state
669of the process using the tape. If the state is D, the process is
670waiting for something. The field WCHAN tells where the driver is
671waiting. If you have the current System.map in the correct place (in
672/boot for the procps I use) or have updated /etc/psdatabase (for kmem
673ps), ps writes the function name in the WCHAN field. If not, you have
674to look up the function from System.map.
675
676Note also that the timeouts are very long compared to most other
677drivers. This means that the Linux driver may appear hung although the
678real reason is that the tape firmware has got confused.