IMHO do not follow example from FreeRTOS but rather one from *
../src/netif/ethernetif.c*
It is much more simple to understand and for modification.
Your mention about OS functions also leads me to draw Your attention on
files *lwipopts.h* and *sys_arch.c*, *sys_arch.h*
where you have to modify settings for your environment - cpu, compiler, etc.
If you are not using OS you should set *NO_SYS* to *1*. If you do not have
file *lwipopts.h* try to
create copy based on some example included in lwip package. Just remember it
is used to
complete and/or redefine default settings from *opt.h* file.
There is one more setting from *lwipopt.h* (*opt.h*), which seems to me to
be most trickiest thing in
process of connecting network driver with stack. It is *ETH_PAD_SIZE* which
describe padding of bytes
transfered to and from network controller hardware. This define is also used
in some important
structures definition, so pay attention on it!
Marek
2008/9/3 jini-2003 <[EMAIL PROTECTED]>:
> Thx for your reply!I think I understand what you say generally.
> There are several functions in ethernetif.c based FREERTOS(ethernetif.c
come
> from an example project) as follows:
> static void low_level_init(struct netif *netif)
> {
> // struct ethernetif *ethernetif = netif->state;
> unsigned portBASE_TYPE uxPriority;
> /* maximum transfer unit */
> netif->mtu = netifMTU;
>
> /* broadcast capability */
> netif->flags = NETIF_FLAG_BROADCAST;
>
> /* Do whatever else is needed to initialize interface. */
> xNetIf = netif;
> /* Initialise the MACB. This routine contains code that polls status
> bits.
> If the Ethernet cable is not plugged in then this can take a
considerable
> time. To prevent this starving lower priority tasks of processing time
we
> lower our priority prior to the call, then raise it back again once the
> initialisation is complete. */
> uxPriority = uxTaskPriorityGet( NULL );
> vTaskPrioritySet( NULL, tskIDLE_PRIORITY );
> while( xMACBInit(&AVR32_MACB) == FALSE )
> {
> __asm__ __volatile__ ( "nop" );
> }
> vTaskPrioritySet( NULL, uxPriority );
> /* Create the task that handles the MACB. */
> // xTaskCreate( ethernetif_input, ( signed portCHAR * ) "ETH_INT",
> netifINTERFACE_TASK_STACK_SIZE, NULL, netifINTERFACE_TASK_PRIORITY, NULL
);
> sys_thread_new( ethernetif_input, NULL, netifINTERFACE_TASK_PRIORITY );
> }
>
> as you see,there are uxTaskPriorityGet,sys_thread_new and so on in this
> function and I really do not know how to deal it if an os not used.
> The similar problem also exist in other functions.
>
>
> 在2008-09-03，"Marek Matej" <[EMAIL PROTECTED]> 写道：
>>Hi Jini,
>>I am working on project with lwip 1.3.0 and system with no OS layer.
>>Therefore RAW API is ised.
>>In ethernetif.c, I have implemented :
>>--ethernetif_init(); which is used for init any of netif
>>--mac_init(); which is intended for initiating network hardware
>>--mac_input(); for reading received packets from network controller
>>--mac_output(); gues for what :)
>>--ethernetif_handlepackets(); running in main loop
>>As I understand, implementation of ethernetif.c is user specific and
>>there is only recommendation how to put it together with Your hardware
>>driver. In my case, is like this:
>>Adding netif using function netif_add() needs to pass in argument
>>function for initiating any network interface (ethernetif_init() ) and
>>function intended to start processing any informing data (ip_input()
>>in my case)
>>Next, ethernetif_handlepackets(struct netif*) is reading packets from
>>net. hw. and putting it to process function of IP stack. Thats nearly
>>all from lower side.
>>The higher side of implementation, is using c++ objects to emulate
>>sockets, with using only RAW API. (due to no OS layer).
>>
>>I try to explain how I deal with ethernetif.c, but I believe there is
>>more even sophisticated way to attach network driver with stack.
>>
>>Mark
>>
>>
>>2008/9/3 jini-2003
>
> ________________________________
> 200万入主万科中粮紫苑155平米8重空间大宅 周周送好礼
> _______________________________________________
> lwip-users mailing list
> [email protected]
> http://lists.nongnu.org/mailman/listinfo/lwip-users
>
_______________________________________________
lwip-users mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/lwip-users
