IPv6: Set loopback-policy for output traffic
[project/netifd.git] / DESIGN
1 Design of the the network interface daemon (netifd)
2 ----------------------------------------------------
3
4 The primary elements of netifd's state are devices and interfaces.
5
6 Devices
7 -------
8
9 A device represents either a physical Linux interface (e.g. eth0), or a
10 virtual link, bound to a static route, a socket or some other external trigger
11 (e.g. for VPN links or tunnels).
12
13 Anything that needs to be aware of device state registers a device_user, which
14 is bound to the device and contains a callback for receiving events on device
15 changes. As soon as the last device_user is removed, the device itself is freed
16 immediately.
17
18 Devices can also internally reference other devices, this is used to manage
19 specific kinds of devices, such as bridges or vlans, which do not require
20 special treatment from interfaces or other high level data structures, with
21 the exception of adding more member interfaces via hotplug, which is useful
22 for bridges or bonding interfaces.
23
24 The device up/down state is managed through refcounting. A device can be
25 brought up using claim_device(), and its reference released again with
26 release_device(). As soon as the refcount goes to zero, the device is
27 immediately brought down.
28 If the device cannot be brought up, claim_device() will return a non-zero
29 status and the caller will not have increased the refcount.
30
31 A registered device may not be available immediately, an interface or another
32 device could also be attached to it, waiting for it to appear in the system,
33 to support triggering interfaces via hotplug.
34
35 All device state transitions are announced via events to all the device_user
36 callbacks. The following event types are supported:
37
38 DEV_EVENT_ADD:
39 The device is now present in the system. When a device_user is being added
40 to a device and the device was already present, this event is generated
41 immediately.
42
43 DEV_EVENT_REMOVE:
44 The device is no longer available. Either it is currently being removed,
45 or it has already disappeared. All users should drop their references
46 immediately and clean up their state for this device.
47
48 DEV_EVENT_SETUP:
49 The device is about to be brought up. This allows device users to apply
50 some low level configuration parameters if necessary, however this event
51 may not be emitted in all cases. Externally managed devices added via
52 hotplug may be already up, and in that case this notification is skipped.
53
54 DEV_EVENT_UP:
55 The device has been successfully brought up.
56
57 DEV_EVENT_TEARDOWN:
58 The device is about to be brought down
59
60 DEV_EVENT_DOWN:
61 The device has been brought down
62
63 The following optional events are supported on some devices:
64
65 DEV_EVENT_LINK_UP: a link has been established on this device
66 DEV_EVENT_LINK_DOWN: the link has been lost
67
68
69
70 Interfaces
71 ----------
72
73 An interface represents a high level configuration applied to one or more
74 devices. An active interface must always be bound to one main device and
75 to a layer 3 device. By default, the layer 3 device points at the reference
76 to the main device, based on how simple protocols like static, dhcp, etc.
77 are set up. More complex protcol handlers such as ppp/pptp or VPN software
78 can remap the layer 3 interface to something else, and external modules
79 such as the firewall can take care of both interfaces if necessary.
80
81 An interface always has the following state information:
82
83 active:
84 The interface can be brought up (its main device is available)
85
86 autostart:
87 If the interface switches from inactive to active, netifd will attempt
88 to bring it up immediately. Manually setting an interface to up (regardless
89 of whether that was successful or not) will set this flag.
90
91 state:
92 IFS_SETUP:
93 The interface is currently being configured by the protocol handler
94 IFS_UP:
95 The interface is fully configured
96 IFS_TEARDOWN:
97 The interface is being deconfigured
98 IFS_DOWN:
99 The interface is down
100
101 An interface references only one protocol handler state, modular protocol
102 handlers such as PPP are expected to be written in a way that allows them
103 to be set up as slave to another protocol handler if necessary (useful for
104 PPPoE or PPTP).
105
106
107
108 Protocol handlers
109 -----------------
110
111 A protocol handler can be attached to anything that provides a callback
112 for state changes. For the simple case it is usually attached to an interface
113 directly.
114
115 The state of a protocol handler is tracked in a struct interface_proto_state,
116 and it depends on the state of the entity that's controlling it.
117
118 It responds to PROTO_CMD_SETUP and PROTO_SETUP_TEARDOWN commands, which
119 should not block at any point in time. Completion is signalled back to the
120 master by sending IFPEV_UP and IFPEV_DOWN events.
121
122 If the setup can be done fast without blocking for a noticeable amount of
123 time, the callback can do it and send back those events immediately.
124 If the setup can take longer, it should use uloop to schedule its actions
125 asynchronously and (if necessary) fork.
126
127 The protocol handler must be able to abort a setup request that's being
128 processed when it encounters a teardown command.
129
130 When a PROTO_SETUP_TEARDOWN call is issued and the 'force' parameter is
131 set, the protocol handler needs to clean up immediately as good as possible,
132 without waiting for its pending actions to complete. If it has spawned
133 any child processes, it needs to kill them and clean up their mess.
134
135 Simple protocol handlers can set the PROTO_FLAG_IMMEDIATE flag if they
136 can perform all required actions immediately without waiting and thus
137 do not need to schedule IFPEV_UP and IFPEV_DOWN transitions. This will
138 cause those events to be generated by core code instead.
139
140 ## TODO: Configuration management, ubus callbacks