Re: [lwip-users] Timers in LWIP and socket protection

[address@hidden]

Hi,

1. There are other timers, but you might not need them (dhcp, arp, 
autoip, dns, ip_reassembly). Other than that, you are right.
2. Yes, that's correct.
3. The tcp timer is only running when a tcp pcb is registered. This gets 
triggered from the function tcp_timer_needed() in tcpip.c. This is a 
rather unclean part of the code, since the prototype resides in tcp.h 
and is only declared #if NO_SYS == 0.
4. That should be enough. You seem to already know that you mustn't use 
an lwip socket from 2 threads simultaneously... :-) (I should note that 
I have never tried it, but if you make sure there are not two (or more) 
threads accessing the socket's data structures (or mboxes) at a time, 
you should be fine.)

Simon

Thomas Taranowski schrieb:
> the tcpip_init function starts up all the required timers (which 
> happens in tcpip_thread startup).  You should not be starting them 
> 'manually'.  The timeout handlers usually have a call to sys_timeout 
> that set's itself as the timeout handler, which essentially sets up a 
> periodic timer.
>
> On Jan 14, 2008 12:18 AM, Muhamad Ikhwan Ismail 
> <address@hidden <mailto:address@hidden>> wrote:
>
>     Hi,
>
>     I got some qs on timer functions on lwip, read some thread on it
>     previously but it doesnt really answer my qs.
>     I ll try to generalize this so that any future user can use this
>     thread for the same problem.
>     Im using lwip 1.2 porting it on a SMX OS on PowerPC MPC852T. Just
>     wanna make sure I am right before i run some multithreading test.
>
>     1. The are 2 timers for the whole stack, which are the arptimer
>     and the tcp timer am I right ?
>
>     2. The arptimer is initialized in ethernetif_init  through this
>     piece of code :
>     sys_timeout(ARP_TMR_INTERVAL, arp_timer, NULL);
>     which then will be assign a struct timeout to the tcpip thread's
>     timeout (since I call it from the tcpip thread initialization)
>     linked list and will be checked and updated each time I wait on
>     semaphore or fetch a message. Hence I dont have to call the
>     arptimer().
>     Am I correct so far ?
>
>     3. My biggest problem is the tcp_timer. It is called by the
>     tcpip_tcp_timer by tcpip API. But no other function calls the
>     tcpip_tcp_timer
>     hence I guess I need to call it myself ? If I do have to call it
>     myself, I guess there are 2 ways. One is I call it before I wait
>     for semaphore after tcpip_thread finishes the job
>     it got with the message it fetched. Or call another thread to call
>     tcpip_tcp_timer which is literally bad cause the stack itself is
>     not multithreading safe, isnt it?
>      If  anyone has done something like this and has tips where I can
>     use the timer safely, I'd appreciate any tips.
>
>     4.I am writing a telnet server program which have 2 task, both
>     using the same socket for transmit and receive. This is possible
>     if i do a critical section protection each time
>     the socket does receive or send, preventing simultaneous access on
>     the socket, am I right ?
>
>     I am thankful for any answers you could provide me.
>
>     Greetings,
>     Ikhwan
>
>
>
>
>     ------------------------------------------------------------------------
>     Share life as it happens with the new Windows Live. Start sharing!
>     
> <http://www.windowslive.com/share.html?ocid=TXT_TAGHM_Wave2_sharelife_012008>
>
>     _______________________________________________
>     lwip-users mailing list
>     address@hidden <mailto:address@hidden>
>     http://lists.nongnu.org/mailman/listinfo/lwip-users
>
>
>
>
> --
> Thomas Taranowski
> Expert embedded software design/contracting
> address@hidden <mailto:address@hidden> (remove 
> the n0sp4m)
> ------------------------------------------------------------------------
>
> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/lwip-users
> ------------------------------------------------------------------------
>
> No virus found in this incoming message.
> Checked by AVG Free Edition. 
> Version: 7.5.516 / Virus Database: 269.19.0/1216 - Release Date: 09.01.2008 10:16
>