1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
|
.\" $FreeBSD$
.\"
.Dd December 29, 1997
.Dt NATM 4
.Os
.Sh NAME
.Nm natm
.Nd Native Mode ATM protocol layer
.Sh DESCRIPTION
The BSD ATM software comes with a
.Em native mode ATM protocol layer
which provides socket level access to AAL0 and AAL5 virtual circuits.
To enable this protocol layer, add
.Dl options NATM
to your kernel configuration file and re-make the kernel (don't forget
to do
.Dq make clean
).
.Sh NATM API
The NATM layer uses a
.Dv struct sockaddr_natm
to specify a virtual circuit:
.Bd -literal -offset indent
struct sockaddr_natm {
u_int8_t snatm_len; /* length */
u_int8_t snatm_family; /* AF_NATM */
char snatm_if[IFNAMSIZ]; /* interface name */
u_int16_t snatm_vci; /* vci */
u_int8_t snatm_vpi; /* vpi */
};
.Ed
.Pp
To create an AAL5 connection to a virtual circuit with VPI 0, VCI 201
one would use the following:
.Bd -literal -offset indent
struct sockaddr_natm snatm;
int s, r;
s = socket(AF_NATM, SOCK_STREAM, PROTO_NATMAAL5);
/* note: PROTO_NATMAAL0 is AAL0 */
if (s < 0) { perror("socket"); exit(1); }
bzero(&snatm, sizeof(snatm));
snatm.snatm_len = sizeof(snatm);
snatm.snatm_family = AF_NATM;
sprintf(snatm.snatm_if, "en0");
snatm.snatm_vci = 201;
snatm.snatm_vpi = 0;
r = connect(s, (struct sockaddr *)&snatm, sizeof(snatm));
if (r < 0) { perror("connect"); exit(1); }
/* s now connected to ATM! */
.Ed
.Pp
The
.Fn socket
call simply creates an unconnected NATM socket. The
.Fn connect
call associates an unconnected NATM socket with a
virtual circuit and tells the driver to enable that virtual circuit
for receiving data. After the
.Fn connect
call one can
.Fn read
or
.Fn write
to the socket to perform ATM I/O.
.Sh Internal NATM operation
Internally, the NATM protocol layer keeps a list of all active virtual
circuits on the system in
.Dv natm_pcbs .
This includes circuits currently being used for IP to prevent NATM and
IP from clashing over virtual circuit usage.
.Pp
When a virtual circuit is enabled for receiving data, the NATM
protocol layer passes the address of the protocol control block down
to the driver as a receive
.Dq handle .
When inbound data arrives, the driver passes the data back with the
appropriate receive handle. The NATM layer uses this to avoid the
overhead of a protocol control block lookup. This allows us to take
advantage of the fact that ATM has already demultiplexed the data for
us.
.Sh Other NATM issues
We are currently involved with a video server project and are using
this driver as part of it. We have a device we build called an MMX.
You can connect a video camera to an MMX and have it send you a stream
of AAL0 cells with the video output in it. Of course this stream
is pretty rapid (in fact, it is massive!), and the normal AAL0
handling of the driver is unable to handle it (you end up with a cell
per small mbuf trying to make it to the application ... it turns out
the socket layer can't keep up with that sort of data stream). To
solve this problem we have implemented a
.Dq raw
mode which batches unprocessed AAL0 info from the card into larger
data chunks blocks. We can save this data to disk in real-time
without the socket layer freaking out. Unfortunately, the data has
RBD (receive buffer descriptors) and cells headers in it, and this has
to be filtered out after capture.
To enable
.Dq raw
mode one does the following ioctl:
.Bd -literal -offset indent
int size = 4000; /* bytes */
ret = ioctl(s, SIOCRAWATM, (caddr_t)&size);
.Ed
.Pp
This tells the driver to batch AAL0 data into 4000 bytes chunks,
rather than the usual 48 bytes chunks. Admittedly this is somewhat
gross, but our current application requires it. In the future we
hope that video sources send data in nice large AAL5 frames.
.Sh CAVEAT
The NATM protocol support is subject to change as
the ATM protocols develop. Users should not depend
on details of the current implementation, but rather
the services exported.
.Sh SEE ALSO
.Xr en 4
.Sh AUTHORS
.An Chuck Cranor
of Washington University implemented the NATM protocol layer
along with the EN ATM driver in 1996 for
.Nx .
|
