Start development on BSD 4_1_snap
[unix-history] / .ref-BSD-4 / usr / man / man4 / ht.4
CommitLineData
60913c9a
BJ
1.TH HT 4
2.UC 4
3.SH NAME
4ht \- RH-11/TE-16 magtape interface
5.SH DESCRIPTION
6The files
7.I "mt0, ..., mt15"
8refer to the DEC RH/TM/TE-16 magtape.
9The files
10.I "mt0, ..., mt7"
11are 800 bpi, and
12.I "mt8, ..., mt15"
13are 1600bpi.
14The files
15.I "mt0, ..., mt3"
16and
17.I "mt8, ..., mt11"
18are rewound when closed; the others are not.
19When a file open for writing is closed, a double end-of-file is written.
20.PP
21A standard tape consists of a
22series of 1024 byte records terminated by an
23end-of-file.
24To the extent possible, the system makes
25it possible, if inefficient, to treat
26the tape like any other file.
27Seeks have their usual meaning and it is possible
28to read or write a byte at a time.
29Writing in very small units is inadvisable,
30however, because it tends to create monstrous record
31gaps.
32.PP
33The
34.I mt
35files discussed above are useful
36when it is desired to access the tape in a way
37compatible with ordinary files.
38When foreign tapes are to be dealt with, and especially
39when long records are to be read or written, the
40`raw' interface is appropriate.
41The associated files may be named
42.I "rmt0, ..., rmt15,"
43but the same minor-device considerations
44as for the regular files still apply.
45.PP
46Each
47.I read
48or
49.I write
50call reads or writes the next record on the tape.
51In the write case the record has the same length as the
52buffer given.
53During a read, the record size is passed
54back as the number of bytes read, provided it is no greater
55than the buffer size;
56if the record is long, an error is indicated.
57In raw tape I/O, the buffer must begin on a word boundary
58and the count must be even.
59Seeks are ignored.
60A zero count is returned when a tape mark is read;
61another read will fetch the first record of the
62next tape file.
63.SH FILES
64/dev/mt\(**,
65/dev/rmt\(**
66.SH "SEE ALSO"
67tp(1)
68.SH BUGS
69The magtape system is supposed to be able
70to take 64 drives.
71Such addressing has never been tried.
72These bugs will be fixed when
73we get more experience with this device.
74.PP
75The driver is limited to four transports.
76.PP
77If any non-data error is encountered, it refuses to do anything
78more until closed.
79In raw I/O, there should be a way
80to perform forward and backward record and file spacing and
81to write an EOF mark explicitly.