Transcript LWIP TCP/IP Stack

LWIP TCP/IP Stack
김백규
What is LWIP?
An implementation of the TCP/IP protocol stack.
The focus of the lwIP stack is to reduce memory usage and
code size
suitable for use in small clients with very limited
resources such as embedded systems.
uses a tailor made API that does not require any data
copying.

Features of TCP/IP stack(Traditional version)
Designing in a layered
fashion leads to…

communication overhead
between layers
Network communication is
similar to IPC or file I/O


APP can’t aware of the
buffer mechanisms.
(e.g. reuse buffers with
frequently used data.)
<Layered model>
Features of TCP/IP stack(LWIP version)
Do not maintain a strict layer.
 This allows using a more relaxed scheme for
communication between layers.
(By means of shared memory)
- APP layer can use the buffer handling mechanisms
used by the lower layers.
APP can more efficiently reuse buffers.
Application process can use the same memory as the
networking code
 App can read and write directly to the internal buffers.
 Saving the expense of performing a copy
Process model of LWIP
All protocols reside in a single process thus are
separated from the OS kernel.

Allow to be portable across different OS.
APP may either reside in the LWIP process or be
in separate processes.


Communicate are done by function calls.
Or a more abstract API.
The operating system emulation layers
OS specific function calls and data structures are not used
directly in the code.
 The operating system emulation layer is used.
The OS emulation layer provides
 Timers, process synchronization, message passing
mechanisms, and so on.
Porting to a different OS
 Only need the operating system emulation layer.
Buffer and memory management
Packet buffers – pbufs
 A pbuf is LWIP’s internal representation of a packet,

And is designed for the special needs of the minimal
stack.
Types of pbufs


PBUF_RAM, PBUF_ROM, PBUF_POOL
A pbuf chain may consist of multiple types of pbufs.
PBUF_RAM pbuf
has the packet data stored in memory managed by the pbuf
subsystem.
used when an application sends data that is dynamically
generated.
PBUF_ROM pbuf
Used when an application sends data that is located in
memory managed by the application.
The main use is when the data is located in ROM
Header that are prepended to the data in a PBUF_ROM pbuf
are stored in a PBUF_RAM pbuf.
PBUF_POOL
Consist of fixed size pbufs allocated from a pool of fixed size
pbufs.
Mainly used by network device drivers since the operation of
allocating a single pbuf is fast and is therefore suitable for use
in an interrupt handler
Network interfaces
The network interfaces are
kept on a global linked list.
Reflect the kind of H/W
Ex) Bluetooth => bt
WLAN => wl
The function the device driver
should call when a packet has
been received.
The function in the device driver
that transmits a packet on the
physical network and it is called
by the IP layer when a packet is
to be sent.
Points to device driver specific
state for the network interface and
is set by the device driver.
IP processing(1/3)
Receiving packets

Network device driver calls ip_input() function.



Checking IP version, header length
Computing the header checksum
Checking destination address.
Sending packets

Handled by the function ip_output()




Find the appropriate network interface.
All IP header fields are filled.
IP header checksum is computed.
The source and destination address are passed.
IP processing(2/3)
Forwarding packets

The packet should be forwarded…


When none of the network interfaces has the
same IP address as an incoming packet’s
destination address.
This is done by the function ip_forward()


ttl field is decreased.
If ttl reaches zero, an ICMP error message is
sent.
IP processing(3/3)
ICMP processing
This is for ICMP ECHO message.
Just swapping the IP destination
and source address of the incoming
packet.
UDP processing(1/2)
The udp_pcb structure
The UDP PCBs are kept on a linked list
which is searched for a match when a UDP
datagram arrives.
Called when a
datagram is
received.
UDP processing(2/2)
UDP processing
TCP processing(1/2)
Function to call when a
listener has been
connected.
Next sequence
number
Receiver’s
window
Timer for
TIME-WAIT
state
Used when passing
received data to the
application layer.
TCP processing(2/2)
Application Program Interface
The BSD socket API


Require data that is to be sent to be copied from
application program to internal buffers in the
TCP/IP stack.
Since the two layers reside in different protection
domains.
The LWIP socket API



Utilizes knowledge of the internal structure of LWIP
to achieve effectiveness.
Does not require that data is copied.
Since the application program can manipulate the
internal buffers directly.
Data types
netconn
- Knowledge of the internal structure of the struct should not be used
in application programs.
- Instead, the API provides functions for modifying and extracting
necessary fields.
Network connection function(1/2)
netconn new()

struct netconn * netconn new(enum netconn type type)
netconn delete()

void netconn delete(struct netconn *conn)
netconn bind()

int netconn bind(struct netconn *conn, struct ip addr *addr,
unsigned short port)
netconn connect()

int netconn connect(struct netconn *conn, struct ip addr
*remote addr, unsigned short remote port)
Network connection function(2/2)
netconn listen()
 int netconn listen(struct netconn *conn)
netconn accept()
 struct netconn * netconn accept(struct netconn *conn)
netconn recv()
 struct netbuf * netconn recv(struct netconn *conn)
netconn write()
 int netconn write(struct netconn *conn, void *data, int len,
unsigned int flags)
Example #1
<This example shows how to open a TCP server on port 2000>
Int main()
{
struct netconn *conn, *newconn;
/* create a connection structure */
conn = netconn_new(NETCONN_TCP);
/* bind the connection to port 2000 on any local
IP address */
netconn_bind(conn, NULL, 2000);
/* tell the connection to listen for incoming
connection requests */
netconn_listen(conn);
/* block until we get an incoming connection */
newconn = netconn_accept(conn);
/* do something with the connection */
process_connection(newconn);
/* deallocate both connections */
netconn_delete(newconn);
netconn_delete(conn);
}
Example #2
<This is a small example that shows a suggested use of the netconn_recv() function.>
Void example_function(struct netconn *conn)
{
struct netbuf *buf;
/* receive data until the other host closes
the connection */
while((buf = netconn_recv(conn)) != NULL) {
do_something(buf);
}
/* the connection has now been closed by the
other end, so we close our end */
netconn_close(conn);
}
Example #3
<This example shows the basic usage of netconn_write().>
Int main()
{
struct netconn *conn;
char data[10];
char text[] = "Static text";
int i;
/* set up the connection conn */
/* [...] */
/* create some arbitrary data */
for(i = 0; i < 10; i++)
data[i] = i;
netconn_write(conn, data, 10, NETCONN_COPY);
netconn_write(conn, text, sizeof(text), NETCONN_NOCOPY);
28
16 NETWORK CONNECTION FUNCTIONS
/* the data can be modified */
for(i = 0; i < 10; i++)
data[i] = 10 - i;
/* take down the connection conn */
netconn_close(conn);
}