1SGM_DD(8) SG3_UTILS SGM_DD(8)
2
6 sgm_dd - copies data to and from files and devices. Specialized for
7 devices that understand the SCSI command set and does memory mapped
8 transfers from sg devices.
9
11 sgm_dd [bs=BS] [count=COUNT] [ibs=BS] [if=IFILE] [iflag=FLAGS] [obs=BS]
12 [of=OFILE] [oflag=FLAGS] [seek=SEEK] [skip=SKIP] [--help] [--version]
13 [bpt=BPT] [cdbsz=6|10|12|16] [dio=0|1] [sync=0|1] [time=0|1] [ver‐
15 bose=VERB]
16
18 Copy data to and from any files. Specialized for "files" that are Linux
19 SCSI generic (sg) devices and raw devices. Uses memory mapped transfers
20 on sg devices. Similar syntax and semantics to dd(1) but does not per‐
21 form any conversions.
22 Will only perform memory mapped transfers when IFILE or OFILE are SCSI
24 generic (sg) devices.
25 If both IFILE and OFILE are sg devices then memory mapped transfers are
27 performed on IFILE. If no other flags are specified then indirect IO is
28 performed on OFILE. If 'oflag=dio' is given then direct IO is attempted
29 on OFILE. If 'oflag=smmap' is given then shared mmap-ed IO (sharing the
30 mmap-ed reserve buffer associated with IFILE) is attempted. In both
31 latter cases if the faster IO option is not available, they fall back
32 to indirect IO and report this at the end of the copy.
33 The first group in the synopsis above are "standard" Unix dd(1) oper‐
35 ands. The second group are extra options added by this utility. Both
36 groups are defined below.
37
39 bpt=BPT
40 each IO transaction will be made using BPT blocks (or less if
41 near the end of the copy). Default is 128 for block sizes less
42 that 2048 bytes, otherwise the default is 32. So for bs=512 the
43 reads and writes will each convey 64 KiB of data by default
44 (less if near the end of the transfer or memory restrictions).
45 When cd/dvd drives are accessed, the block size is typically
46 2048 bytes and bpt defaults to 32 which again implies 64 KiB
47 transfers.
48 bs=BS where BS must be the block size of the physical device. Note
50 that this differs from dd(1) which permits BS to be an integral
51 multiple. Default is 512 which is usually correct for disks but
52 incorrect for cdroms (which normally have 2048 byte blocks). For
53 this utility the maximum size of each individual IO operation is
54 BS * BPT bytes.
55 cdbsz=6 | 10 | 12 | 16
57 size of SCSI READ and/or WRITE commands issued on sg device
58 names. Default is 10 byte SCSI command blocks (unless calcula‐
59 tions indicate that a 4 byte block number may be exceeded, in
60 which case it defaults to 16 byte SCSI commands).
61 count=COUNT
63 copy COUNT blocks from IFILE to OFILE. Default is the minimum
64 (of IFILE and OFILE) number of blocks that sg devices report
65 from SCSI READ CAPACITY commands or that block devices (or their
66 partitions) report. Normal files are not probed for their size.
67 If skip=SKIP or skip=SEEK are given and the count is derived
68 (i.e. not explicitly given) then the derived count is scaled
69 back so that the copy will not overrun the device. If the file
70 name is a block device partition and COUNT is not given then the
71 size of the partition rather than the size of the whole device
72 is used. If COUNT is not given and cannot be derived then an
73 error message is issued and no copy takes place.
74 dio=0 | 1
76 permits direct IO to be selected on the write-side (i.e. on
77 OFILE). Only allowed when the read-side (i.e. IFILE) is a sg
78 device. When 1 there may be a "zero copy" copy (i.e. mmap-ed
79 transfer on the read into the user space and direct IO from
80 there on the write, potentially two DMAs and no data copying
81 from the CPU). Default is 0. The same action as 'dio=1' is also
82 available with 'oflag=dio'.
83 ibs=BS if given must be the same as BS given to 'bs=' option.
85 if=IFILE
87 read from IFILE instead of stdin. If IFILE is '-' then stdin is
88 read. Starts reading at the beginning of IFILE unless SKIP is
89 given.
90 iflag=FLAGS
92 where FLAGS is a comma separated list of one or more flags out‐
93 lined below. These flags are associated with IFILE and are
94 ignored when IFILE is stdin.
95 obs=BS if given must be the same as BS given to 'bs=' option.
97 of=OFILE
99 write to OFILE instead of stdout. If OFILE is '-' then writes to
100 stdout. If OFILE is /dev/null then no actual writes are per‐
101 formed. If OFILE is '.' (period) then it is treated the same
102 way as /dev/null (this is a shorthand notation). If OFILE exists
103 then it is _not_ truncated; it is overwritten from the start of
104 OFILE unless 'oflag=append' or SEEK is given.
105 oflag=FLAGS
107 where FLAGS is a comma separated list of one or more flags out‐
108 lined below. These flags are associated with OFILE and are
109 ignored when OFILE is /dev/null, '.' (period), or stdout.
110 seek=SEEK
112 start writing SEEK bs-sized blocks from the start of OFILE.
113 Default is block 0 (i.e. start of file).
114 skip=SKIP
116 start reading SKIP bs-sized blocks from the start of IFILE.
117 Default is block 0 (i.e. start of file).
118 sync=0 | 1
120 when 1, does SYNCHRONIZE CACHE command on OFILE at the end of
121 the transfer. Only active when OFILE is a sg device file name.
122 time=0 | 1
124 when 1, times transfer and does throughput calculation, out‐
125 putting the results (to stderr) at completion. When 0 (default)
126 doesn't perform timing.
127 verbose=VERB
129 as VERB increases so does the amount of debug output sent to
130 stderr. Default value is zero which yields the minimum amount
131 of debug output. A value of 1 reports extra information that is
132 not repetitive. A value 2 reports cdbs and responses for SCSI
133 commands that are not repetitive (i.e. other that READ and
134 WRITE). Error processing is not considered repetitive. Values of
135 3 and 4 yield output for all SCSI commands (and Unix read() and
136 write() calls) so there can be a lot of output.
137 --help outputs usage message and exits.
139 --version
141 outputs version number information and exits.
142
144 Here is a list of flags and their meanings:
145 append causes the O_APPEND flag to be added to the open of OFILE. For
147 normal files this will lead to data appended to the end of any
148 existing data. Cannot be used together with the seek=SEEK
149 option as they conflict. The default action of this utility is
150 to overwrite any existing data from the beginning of the file
151 or, if SEEK is given, starting at block SEEK. Note that attempt‐
152 ing to 'append' to a device file (e.g. a disk) will usually be
153 ignored or may cause an error to be reported.
154 dio is only active with oflag (i.e. 'oflag=dio'). Its action is
156 described in the 'dio=1' option description above.
157 direct causes the O_DIRECT flag to be added to the open of IFILE and/or
159 OFILE. This flag requires some memory alignment on IO. Hence
160 user memory buffers are aligned to the page size. Has no effect
161 on sg, normal or raw files.
162 dpo set the DPO bit (disable page out) in SCSI READ and WRITE com‐
164 mands. Not supported for 6 byte cdb variants of READ and WRITE.
165 Indicates that data is unlikely to be required to stay in device
166 (e.g. disk) cache. May speed media copy and/or cause a media
167 copy to have less impact on other device users.
168 dsync causes the O_SYNC flag to be added to the open of IFILE and/or
170 OFILE. The "d" is prepended to lower confusion with the
171 'sync=0|1' option which has another action (i.e. a synchronisa‐
172 tion to media at the end of the transfer).
173 excl causes the O_EXCL flag to be added to the open of IFILE and/or
175 OFILE.
176 fua causes the FUA (force unit access) bit to be set in SCSI READ
178 and/or WRITE commands. This only has effect with sg devices. The
179 6 byte variants of the SCSI READ and WRITE commands do not sup‐
180 port the FUA bit. Only active for sg device file names.
181 null has no affect, just a placeholder.
183 smmap is only active for oflag. It sets shared mmap IO usage on OFILE
185 if it is a sg device node. The IFILE also needs to be a sg
186 device node (or there is no mmap-ed reserve buffer to share).
187
189 Here are some retired options that are still present:
190 fua=0 | 1 | 2 | 3
192 force unit access bit. When 3, fua is set on both IFILE and
193 OFILE; when 2, fua is set on IFILE; when 1, fua is set on OFILE;
194 when 0 (default), fua is cleared on both. See the 'fua' flag.
195
197 A raw device must be bound to a block device prior to using sgm_dd.
198 See raw(8) for more information about binding raw devices. To be safe,
199 the sg device mapping to SCSI block devices should be checked with 'cat
200 /proc/scsi/scsi' before use.
201 Raw device partition information can often be found with fdisk(8) [the
203 "-ul" argument is useful in this respect].
204 Various numeric arguments (e.g. SKIP) may include multiplicative suf‐
206 fixes or be given in hexadecimal. See the "NUMERIC ARGUMENTS" section
207 in the sg3_utils(8) man page.
208 The count, skip and seek parameters can take 64 bit values (i.e. very
210 big numbers). Other values are limited to what can fit in a signed 32
211 bit number.
212 Data usually gets to the user space in a 2 stage process: first the
214 SCSI adapter DMAs into kernel buffers and then the sg driver copies
215 this data into user memory (write operations reverse this sequence).
216 With memory mapped transfers a kernel buffer reserved by sg is memory
217 mapped (see the mmap(2) system call) into the user space. When this is
218 done the second (redundant) copy from kernel buffers to user space is
219 not needed. Hence the transfer is faster and requires less "grunt" from
220 the CPU.
221 All informative, warning and error output is sent to stderr so that
223 dd's output file can be stdout and remain unpolluted. If no options are
224 given, then the usage message is output and nothing else happens.
225 For sg devices this utility issues SCSI READ and WRITE (SBC) commands
227 which are appropriate for disks and reading from CD/DVD/BD drives.
228 Those commands are not formatted correctly for tape devices so sgm_dd
229 should not be used on tape devices.
230 This utility stops the copy if any error is encountered. For more
232 advanced "copy on error" logic see the sg_dd utility (and its 'coe'
233 flag).
234
236 See the examples given in the man page for sg_dd(8).
237
239 The signal handling has been borrowed from dd: SIGINT, SIGQUIT and SIG‐
240 PIPE output the number of remaining blocks to be transferred and the
241 records in + out counts; then they have their default action. SIGUSR1
242 causes the same information to be output yet the copy continues. All
243 output caused by signals is sent to stderr.
244
246 The exit status of sgm_dd is 0 when it is successful. Otherwise see the
247 sg3_utils(8) man page. Since this utility works at a higher level than
248 individual commands, and there are 'coe' and 'retries' flags, individ‐
249 ual SCSI command failures do not necessary cause the process to exit.
250
252 Written by Doug Gilbert and Peter Allworth.
253
255 Report bugs to <dgilbert at interlog dot com>.
256
258 Copyright © 2000-2009 Douglas Gilbert
259 This software is distributed under the GPL version 2. There is NO war‐
260 ranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PUR‐
261 POSE.
262
264 The simplest variant of this utility is called sg_dd. A POSIX threads
265 version of this utility called sgp_dd is in the sg3_utils package. The
266 lmbench package contains lmdd which is also interesting. raw(8), dd(1)
267sg3_utils-1.27 March 2009 SGM_DD(8)