Support SLIP encapsulation as a tunnelling mechanism.

[tripe] / doc / tripe.8

diff --git a/doc/tripe.8 b/doc/tripe.8
index 373271a7440920e78e8292fefea0f0a0fa288265..bcfdcbb4a94829c059ecdefe7b2db03e38b9bf12 100644 (file)
--- a/doc/tripe.8
+++ b/doc/tripe.8
@@ -477,6 +477,34 @@ at half the underlying hash function's output length.
 A `mask-generation function', used in the key-exchange.  The default is
 .IB hash \-mgf
 and there's no good reason to change it.
+.SS "Using SLIP interfaces"
+Though not for the faint of heart, it is possible to get
+.B tripe
+to read and write network packets to a pair of file descriptors using
+SLIP encapsulation.  No fancy header compression of any kind is
+supported.  The intended use is that you allocate a pty pair, set the
+slave's line discipline to SLIP, and hand the
+.B tripe
+server the master end, though in fact any old file descriptors will do.
+.PP
+The SLIP descriptors must be set up and running before the daemon is
+started, and you must inform it about the SLIP descriptors it's meant to
+use in an environment variable
+.BR TRIPE_SLIPIF ,
+whose value must be a colon-separated list of items of the form
+.IR infd [ \c
+.BI , outfd \c
+.RB ] \c
+.BI = \c
+.IR ifname .
+If the
+.I outfd
+is omitted, the same file descriptor is used for input and output.
+.PP
+The output file descriptor should not block unless it really needs to:
+the
+.B tripe
+daemon assumes that it won't, and will get wait for it to accept output.
 .SS "About the name"
 The program's name is
 .BR tripe ,