TCP/IP Porting

2.3Creating the IP port file

This section contains the following information:

•The ipport.h file on page 2-4

•Standard macros and definitions on page 2-4

•CPU architecture on page 2-5

•Pre-emption and protection on page 2-6

•Debugging aids on page 2-6

•Timers and multitasking on page 2-8

•Stack features and options on page 2-8

•Optional compilation switches on page 2-9.

2.3.1The ipport.h file

The ipport.h file contains the port-dependent definitions for the IP layer, and the architectural definitions for all the IP-related code. It also controls CPU architectures (big-endian or little-endian), compiler options, and optional features (DHCP, multiple interfaces, and IP routing support).

Caution

A mistake in this file (such as using big-endian in place of little-endian) can create severe problems, so it is important to set this file up correctly.

You must create a version of the IP port file before you compile the portable TCP/IP stack files, and must #include the ipport.h file in every C file of every module throughout the TCP/IP software.

2.3.2Standard macros and definitions

The ARM TCP/IP stack expects TRUE, FALSE, and NULL to be defined in ipport.h. Typically, the best way to do this is to include the standard C library file stdio.h in ipport.h. If stdio.h is impractical to use or not available on your system, the following example works in most environments:

#ifndef TRUE #define TRUE -1 #define FALSE 0 #endif

#ifndef NULL

#define NULL (void*)0 #endif

TCP/IP Porting

2.3.3CPU architecture

Four common macros from Berkeley UNIX are used for doing byte order conversions between different CPU architecture types:

•htons()

•htonl()

•ntohs()

•ntohl().

You can use these functions as either macros or functions. They accept 16-bit and 32-bit quantities as shown and convert them from network format (big-endian) to the format supported by the local system.

If your system is using the ARM processor in big-endian mode, these macros can return the variable passed, for example:

#define htonl(l) (l) #define htons(s) (s) #define ntohl(l) (l) #define ntohs(s) (s)

If your system is using the ARM processor in little-endian mode, the byte order must be swapped. For htonl() and ntohl(), use the lswap() function provided in the \armthumb directory (see Sample package directories on page 1-7). For htons() and ntohs(), use a byte-swapping macro, as shown below:

TCP/IP Porting

2.3.4Pre-emption and protection

You must define primitives in order to protect sections of code that must not be interrupted or pre-empted (see Implementing pre-emption and protection on page 2-16).

The critical section protection scheme is typically used on embedded systems that lack a multitasking capability:

Those systems are described in detail in ENTER_CRIT_SECTION() and

EXIT_CRIT_SECTION() on page 3-6.

The lock net resource macros are typically used on real-time kernels, such as VRTX and VxWorks:

void UNLOCK_NET_RESOURCE(); /* end re-entrance protection */

2.3.5Debugging aids

You must include the following macros in your functions to provide support while you are debugging your code:

•dtrap on page 2-6

•initmsg() and dprintf() on page 2-7

•NPDEBUG on page 2-7.

dtrap

This macro is called by the stack code when it detects a situation that should not be occurring. The intention is for the dtrap() macro to invoke whatever debugger may be in use by the programmer. In this way, it acts like an embedded breakpoint. The stack code can continue executing after a dtrap(), but the dtrap() typically indicates that something is wrong with the port.

Note

Products based on this code should not be shipped until all calls to dtrap() have been removed from the code. You can redefine dtrap() to a null macro to slightly reduce code size.

TCP/IP Porting

initmsg() and dprintf()

The initmsg() and dprintf() macros have the same function and syntax as printf(). They have different names so their output can be redirected to different locations, or so they can be individually disabled.

The initmsg() macro is called by various stack functions to print function status messages during initialization. These messages are for information and are not warnings.

The dprintf() macro is used throughout the stack code to print warning messages when something seems to be wrong.

In most ports, these can both be mapped to printf() while the product is under development.

Note

Mapping initmsg() and dprintf() to printf() works with the ADS, but performance of the stack is severely affected by the time taken to transfer debugging information across the RDI link to the ARM debugger.

A more efficient alternative is to use low-level functions to replace putchar() and printf() and to write output to a UART or other console device. The file \misclib\ttyio.c contains an example dprintf() function that can be used in conjunction with \integrator\uartio.c for this purpose.

#define initmsg printf /* Same parms as printf. Called at */ /* boot time */

#define dprintf printf /* Same parms as printf. Called */ /* during run time */

For some products, it may be desirable to define these to a null macro before releasing.

NPDEBUG

Defining the NPDEBUG macro causes the debug code to be compiled into the application. The debug code performs tasks, such as checking for valid parameters and sensible configurations during runtime. The debug code invokes dtrap() or dprintf() to inform the programmer of detected problems. To make use of this feature, make sure NPDEBUG is defined during development.

#define NPDEBUG 1 /* enable debug checks */