doc: Add BTSnoop/Monitor protocol documentation
diff --git a/Makefile.am b/Makefile.am
index b34b633..701da6d 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -262,6 +262,8 @@
EXTRA_DIST += doc/pics-opp.txt doc/pixit-opp.txt \
doc/pts-opp.txt
+EXTRA_DIST += doc/btsnoop.txt
+
EXTRA_DIST += tools/magic.btsnoop
AM_CFLAGS += @DBUS_CFLAGS@ @GLIB_CFLAGS@
diff --git a/doc/btsnoop.txt b/doc/btsnoop.txt
new file mode 100644
index 0000000..ed56b5a
--- /dev/null
+++ b/doc/btsnoop.txt
@@ -0,0 +1,59 @@
+BTSnoop/Monitor protocol formats
+********************************
+
+TTY-based protocol
+==================
+
+This section covers the protocol that can be parsed by btmon when
+passing it the --tty parameter. The protocol is little endian, packet
+based, and has the following header for each packet:
+
+struct tty_hdr {
+ uint16_t data_len;
+ uint16_t opcode;
+ uint8_t flags;
+ uint8_t hdr_len;
+ uint8_t ext_hdr[0];
+} __attribute__ ((packed));
+
+The actual payload starts at ext_hdr + hdr_len and has the length of
+data_len - 4 - hdr_len. Each field of the header is defined as follows:
+
+data_len:
+ This is the total length of the entire packet, excuding the
+ data_len field itself.
+
+opcode:
+ The BTSnoop opcode
+
+flags:
+ Special flags for the packet. Currently no flags are defined.
+
+hdr_len:
+ Length of the extended header.
+
+ext_hdr:
+ This is a sequence of header extension fields formatted as:
+
+ struct {
+ uint8_t type;
+ uint8_t value[length];
+ }
+
+ The length of the value is dependent on the type. Currently the
+ following types are defined:
+
+ Type Length Meaning
+ ----------------------------------------------------------------
+ 1 Command drops 1 byte Dropped HCI command packets
+ 2 Event drops 1 byte Dropped HCI event packets
+ 3 ACL TX drops 1 byte Dropped ACL TX packets
+ 4 ACL RX drops 1 byte Dropped ACL RX packets
+ 5 SCO TX drops 1 byte Dropped SCO TX packets
+ 6 SCO RX drops 1 byte Dropped SCO RX packets
+ 7 Other drops 1 byte Dropped other packets
+ 8 32-bit timestamp 4 bytes Timestamp in 1/10th ms
+
+ The fields of the extended header must be sorted by increasing
+ type. This is essential so that unknown types can be ignored and
+ the parser can jump to processing the payload.
