gmane.network.lwip.general

Socket Api send copy

Laurent — 2013-05-25T18:45:20

Hi,

I'm currently porting Lwip 1.4.1 to ALMOS (an open-source research operating
system currently under development aiming cc-NUMA Architecture, if you want
to learn about it you can find it at  https://www.almos.fr/trac/almos
<https://www.almos.fr/trac/almos>  ) 
I'm using Lwip-thread to handle TCP/IP Connection. I run into a problem when
using the socket API provided.

When i use lwip_send(), the copy of the buffer parameter is copied during
the lwip processing of the message. This leads to make the following code
being random : 

lwip_send(fd,buf,size,0);
memset(buf,0,size);

It can send a buffer full of zeros, a buffer half of zeros, or the buffer
with nice datas. I'm pretty sure the Socket API ensure a sequential
execution. Did anyone run into that problem and is there a fix for this?



--
View this message in context: http://lwip.100.n7.nabble.com/Socket-Api-send-copy-tp21449.html
Sent from the lwip-users mailing list archive at Nabble.com.

Add Subject; help request with freeing netconn buffers

William D. Anderson — 2013-05-24T14:50:34

Sorry I forgot to add a subject line.

From: lwip-users-bounces+wanderson=artel.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org [mailto:lwip-users-bounces+wanderson=artel.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org] On Behalf Of William D. Anderson
Sent: Thursday, May 23, 2013 10:51 PM
To: lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
Subject: [lwip-users] (no subject)

Hello,
  I have been beating my head against an issue related to losing netconn buffers and hoped someone on the list might shed some light as to what I should look out for.
 I have lwip up and running and it responds very reliably to external arp and ping traffic. When I originate pings  I can currently send only 8 ping requests using either Sockets or Raw api.  On the 9th request I can no longer allocate a socket or pbuf depending on the mode being used.  When I  use socket API for each socket request I call sys_thead_new.  After processing each ping I either call lwip_close(s) or raw_remove(ping_pcb) respectively;   I am using Xilinix pro

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

Sylvain Rochet — 2013-05-24T08:42:38

Hi,


On Fri, May 24, 2013 at 07:35:49AM +0200, Furiantes Furiantes wrote:

No, as said in doc/rawapi.txt :-)


Raw API is *NOT* thread safe: udp_*(), tcp_*(), netif_*(), ...

Netconn API is thread safe: netconn_*()
Socket API is thread safe: lwip_accept(), lwip_bind(), lwip_connect(), ...
NetifAPI is thread safe: netifapi_*()



Nah, this is wrong, you *have to* use netconn or socket API.


Also, and this is very important too, Netconn and Socket API are thread 
safe against lwIP thread but NOT against themselves, so a UDP or a TCP 
PCB must be handled by only one thread. If you want to share a PCB 
between threads you can of course add the necessary mutex around Netconn 
or Socket calls.


Sylvain
_______________________________________________
lwip-users mailing list
lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

Furiantes Furiantes — 2013-05-24T05:35:49

Is tcp_new() thread safe too ? or just netcon. My scenario is that:

- i will call a TCP staff only in seperated threads. Like do_download =
tcp_new() in one thread and than like listen to some port initialized in
seperated thread etc. etc.

Is that ok to do or not ? With global access can I access from TCP_1 thread
to TCP_2 thread like TCP_2 kill TCP_1 thread.

Thank you very much for your help.

Danijel


2013/5/23 Sylvain Rochet <gradator-XWGZPxRNpGHk1uMJSBkQmQ< at >public.gmane.org>

_______________________________________________
lwip-users mailing list
lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

(no subject)

William D. Anderson — 2013-05-24T02:50:43

Hello,
  I have been beating my head against an issue related to losing netconn buffers and hoped someone on the list might shed some light as to what I should look out for.
 I have lwip up and running and it responds very reliably to external arp and ping traffic. When I originate pings  I can currently send only 8 ping requests using either Sockets or Raw api.  On the 9th request I can no longer allocate a socket or pbuf depending on the mode being used.  When I  use socket API for each socket request I call sys_thead_new.  After processing each ping I either call lwip_close(s) or raw_remove(ping_pcb) respectively;   I am using Xilinix provide lwip140_v1_03_a with xilkernel.  Example debug output below shows one good request followed by failing request.  I hope I Am  correct to assume that I can open and close then reuse raw sockets and raw_pcbs ?   Any help greatly appreciated.

Thanks
Bill Anderson

pingloop c0a8000b cnt = 8
pinghost c0a8000b
Ping w socket Dst IP c0a8000b
sys_thread_new funcp 0xF0005C34

Re: ppp problem

yueyue papa — 2013-05-23T23:41:47

pppWrite[536910800]: len=0

Why len=0 ?

Whether the data is send out though PPP IO?




On Fri, May 24, 2013 at 7:30 AM, yueyue papa <yueyuepapa-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:

_______________________________________________
lwip-users mailing list
lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

Re: ppp problem

yueyue papa — 2013-05-23T23:30:54

The log is still not clear.  First step is simplify the process

1. Do not let ethenet interface working if PPP interface is not on.
 (Current your PPP is not on yet.)
2. Enable PPP SIO log.

Your log did not provide enough PPP debug information. It is hard to know
which kind of problem happen.   It is not easy to find PPP problem from
your log. IP package is received, but it is hard to know  from PPP inteface
or from net interface.




On Thu, May 23, 2013 at 11:17 PM, Ahmed Fayek <ahmad_fayek-/E1597aS9LQAvxtiuMwx3w< at >public.gmane.org> wrote:

_______________________________________________
lwip-users mailing list
lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

Re: lwip ppp using gprs modem

Nikolas Karakotas — 2013-05-23T22:07:51

It is the new PPP stack sylvain has updated. You need to grab the latest 
source code from git. It's not in 1.4.1

-----Original Message----- 
From: afayek
Sent: Thursday, May 23, 2013 11:51 PM
To: lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
Subject: Re: [lwip-users] lwip ppp using gprs modem

what is these functions you are using?:

ppp_new();
ppp_set_netif_statuscallback();
ppp_set_netif_linkcallback();

they are not in lwip stack.



--
View this message in context: 
http://lwip.100.n7.nabble.com/lwip-ppp-using-gprs-modem-tp21391p21425.html
Sent from the lwip-users mailing list archive at Nabble.com.

_______________________________________________
lwip-users mailing list
lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

Sylvain Rochet — 2013-05-23T16:59:36

Hi,

On Thu, May 23, 2013 at 12:56:07PM -0400, Pomeroy, Marty wrote:

Not true, netconn API *IS* thread-safe.

Sylvain
_______________________________________________
lwip-users mailing list
lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

Pomeroy, Marty — 2013-05-23T16:58:11

 
Set it up using tcpip_init().  For DHCP (started after the phy is configured), my startup is:
{
   ip_addr_t eth0ipaddr, eth0netmask, eth0gw;
   IP4_ADDR(&eth0gw, 0,0,0,0);  // Gateway is for routing.  Not used here.
   IP4_ADDR(&eth0ipaddr, 0,0,0,0);  // for DHCP, start with this IP 
   IP4_ADDR(&eth0netmask, 255,255,255,0);
   tcpip_init( NULL, NULL);
   netif_add( &g_NetIF, &eth0ipaddr, &eth0netmask, &eth0gw, NULL, ethernetif_init, tcpip_input);
   // Start netif using DHCP or stop netif if the phy link goes down
}
 
Marty

________________________________

From: lwip-users-bounces+martypomeroy=novabio.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org [mailto:lwip-users-bounces+martypomeroy=novabio.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org] On Behalf Of furiantes
Sent: Thursday, May 23, 2013 12:47 PM
To: Mailing list for lwIP users
Subject: Re: [lwip-users] LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1


Thank you for your reply! I will see that options too. As I see the lwip is already have a thread to serve T

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

Pomeroy, Marty — 2013-05-23T16:56:07

 
Do not call tcp_new() or netconn_new() from outside the lwip thread.  As you say, things will work most of the time...  Until you hit the race condition and something breaks.  But eventually you will hit the race condition.
 
I use DHCP and the socket API.  If you don't have memory limitations, that works.  The raw api will save some space, but you have to know more about what you are doing.  Yeah, lots of options.  Use existing examples wherever you can.
 
Marty


________________________________

From: lwip-users-bounces+martypomeroy=novabio.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org [mailto:lwip-users-bounces+martypomeroy=novabio.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org] On Behalf Of furiantes
Sent: Thursday, May 23, 2013 12:38 PM
To: Mailing list for lwIP users
Subject: Re: [lwip-users] LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1


I have some more questions.

1. Can i call tcp_new() in different FreeRTOS threads. So functions to set tcp recieve functions, sent handler, errors handler etc. is that ok

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

furiantes — 2013-05-23T16:47:07

Thank you for your reply! I will see that options too. As I see the lwip 
is already have a thread to serve TCP (main_lwip_thread)

This is actully main part of TCP for lwip, if i understand right.

A source from


  lwIP/api/tcpip.c


/**
  * The main lwIP thread. This thread has exclusive access to lwIP core functions
  * (unless access to them is not locked). Other threads communicate with this
  * thread using message boxes.
  *
  * It also starts all the timers to make sure they are running in the right
  * thread context.
  *
  * < at >param arg unused argument
  */
extern  struct  netif  *pxNetIfInUse;
sys_mbox_t  *toOutput;
extern  uint32_t  cableState;
static  void
tcpip_thread(void  *arg)
{
   struct  tcpip_msg  *msg;
   LWIP_UNUSED_ARG(arg);

   if  (tcpip_init_done  !=  NULL)  {
     tcpip_init_done(tcpip_init_done_arg);
   }
   toOutput  =  &mbox;
   LOCK_TCPIP_CORE();
   while  (1)  {                           /* MAIN Loop */
     UNLOCK_TCPIP_CORE();
     LWIP_TCPIP_THREAD_ALIVE();
     /* wait for

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

furiantes — 2013-05-23T16:37:51

Oh thank you very much! In about PHY i have a problem becouse he don't 
have interrupt for that,
so i need to do a polling. But as you suggested i will try to do with 
that as you sad it.

In my polling timer interrupt call ->

tcpip_callback_with_block( s_HandlePhyInterrupt, NULL, 0);

Where i will have logic:

if(cableState == 0)
{
    //link off
    iface_up = 0;
    netif_set_link_down(pxNetIfInUse);
}
else
{
   //link on
   netif_set_link_up(pxNetIfInUse);
}

I have some more questions.

1. Can i call tcp_new() in different FreeRTOS threads. So functions to 
set tcp recieve functions, sent handler, errors handler etc. is that ok?

2. This will be a dumb question but i would like to know.

I will have to make these connections:
- UDP broadcast listener
- do a polling to a server (30 second) intervals
- One file download connection

What is the best/better way to do it:

- Netcon API (which is sequential executing i know)
- TCP/IP as it is
- Socket API

Can i call netcon_new in various threads, so the thi

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

FreeRTOS Info — 2013-05-23T15:28:28



On 23/05/2013 16:20, Pomeroy, Marty wrote:

Normally in a FreeRTOS application you would want this to be event
driven.  There are some example FreeRTOS lwIP integrations you can use
as examples, but the general approach (although for different stack) can
be seen in the example code at the bottom of this page:

http://www.freertos.org/Pend-on-multiple-rtos-objects.html

In this code the cable in/out interrupt would be packaged as a
eNetworkDownEvent event.
Regards,
Richard.

+ http://www.FreeRTOS.org
Designed for microcontrollers. More than 103000 downloads in 2012.

+ http://www.FreeRTOS.org/plus
Trace, safety certification, FAT FS, TCP/IP, training, and more...

Re: ppp problem

Ahmed Fayek — 2013-05-23T15:17:35

I initialize the GPRS and send dialup ATD*99***1# 
it respond correctly that is handled in this function GSM_PPP_ConnectProcess()

part of the debug log is attached.
the application is to work as a GPRS router that forward IP packets from ethernet and GPRS.




________________________________
 From: yueyue papa <yueyuepapa-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>
To: Ahmed Fayek <ahmad_fayek-/E1597aS9LQAvxtiuMwx3w< at >public.gmane.org> 
Cc: Mailing list for lwIP users <lwip-users< at >nongnu.org> 
Sent: Thursday, May 23, 2013 4:37 PM
Subject: Re: [lwip-users] ppp problem
 


It is hard to know from your code. Provide the debug log with AT information.






On Thu, May 23, 2013 at 10:32 PM, Ahmed Fayek <ahmad_fayek-/E1597aS9LQAvxtiuMwx3w< at >public.gmane.org> wrote:

yes modem is initialized corectly and PPP channel opened.
       int setup = 0;
 after boot up, to
 536910832
<CR>pbuf_alloc(length=1504)
pbuf_alloc(length=1504) == 200016d4
pppMain: unit 536911872: Connecting
tcpip_thread: CALLBACK 20006808
pppStar

Re: LPC1768/9 LAN8720 FreeRTOS 7.4 LWIP 1.4.1

Pomeroy, Marty — 2013-05-23T15:20:20

If I'm seeing this right, you are polling for link status, and want to
figure out which is the better way to poll.

First, all tcpip thread stuff MUST be done on the tcpip thread, so DO
NOT call it from the timer interrupt.  Option 2 is eventually going to
cause errors.

But better still, most phy chips will generate an interrupt when a cable
is inserted or pulled.  In that interrupt, send a message to update
status.  My interrupt is basically:
   tcpip_callback_with_block( s_HandlePhyInterrupt, NULL, 0);
This passes a static function to be called from the tcpip thread.

That's my preference - use the interrupt, let the phy chip notify your
code when there is a change, have the interrupt pass a task-time
interrupt handler to the tcpip thread.

Marty
 

-----Original Message-----
From: lwip-users-bounces+martypomeroy=novabio.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
[mailto:lwip-users-bounces+martypomeroy=novabio.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org] On
Behalf Of furiantes
Sent: Wednesday, May 22, 2

Re: lwIP for eCos (LPC1768)

Pomeroy, Marty — 2013-05-23T14:48:50

This is late to the party, but I always have a second ethernet card on
my PC.  Configure it and your LPC1768 board with compatible static IP
addresses (192.168.1.nn with 255.255.255.0 for mask), and connect them
through a switch.

I don't know the particular example, but with a lot of these, you can
then type the board IP address in the PC browser, and see the web
content served.  At least you should be able to ping the board from the
PC.

Marty


-----Original Message-----
From: lwip-users-bounces+martypomeroy=novabio.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
[mailto:lwip-users-bounces+martypomeroy=novabio.com-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org] On
Behalf Of enricoG
Sent: Wednesday, May 15, 2013 11:01 AM
To: lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
Subject: [lwip-users] lwIP for eCos (LPC1768)

Hi all, i'm in trouble with lwIP. I'm trying the examples of lwIP for
eCos SO on my board (LPC1768 , a cortexM board). I found any examples of
lwIP in eCos repository

(you find it here: https://www

Re: ppp problem

yueyue papa — 2013-05-23T14:21:55

whether you init your modem?

e.g AT command for setup your module.   setup the APN,  whether the dial
process success?




On Thu, May 23, 2013 at 10:18 PM, Ahmed Fayek <ahmad_fayek-/E1597aS9LQAvxtiuMwx3w< at >public.gmane.org> wrote:

_______________________________________________
lwip-users mailing list
lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

Re: ppp problem

yueyue papa — 2013-05-23T14:39:47

Linux PPPD log,   you could see there is AT command and dial command before
PPPD working
#########
send (^MAT^M)s3c2440,s3c6410,ok6410,fl2440,arm,linux,winbond,nuvoton,w90p710,w90n745,开源,嵌入式,操作系统,嵌入式开发,嵌入式联盟,linux,ecos,uclinux,t-kernel,freeos,rtems,ucos,skyeye,
 _0
< at >+ {! `) {7 B
expect (OK)
^M^M; z$ l4 X; w0 q$ v& c
OK嵌入式开发联盟. e0 Y9 b4 H1 k8 }# T

Re: ppp problem

yueyue papa — 2013-05-23T14:37:26

It is hard to know from your code. Provide the debug log with AT
information.




On Thu, May 23, 2013 at 10:32 PM, Ahmed Fayek <ahmad_fayek-/E1597aS9LQAvxtiuMwx3w< at >public.gmane.org> wrote:

_______________________________________________
lwip-users mailing list
lwip-users-qX2TKyscuCcdnm+yROfE0A< at >public.gmane.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

Re: ppp problem

Ahmed Fayek — 2013-05-23T14:32:27

yes modem is initialized corectly and PPP channel opened.

I have this code:

          int connected = 0;
          int setup = 0;
          int pd;
          const char *username = "myuser";
          const char *password = "mypassword";

          /* initialise lwIP. This creates a new thread, tcpip_thread, that
           * communicates with the pppInputThread (see below) */
          tcpip_init(tcpip_init_done, &setup);
          /*
          while (!setup) {
              DelayMS(100);
          }
        */
          /* initialise PPP. This needs to be done only once after boot up, to
           * initialize global variables, etc. */
          pppInit();

//initialize modem and open ppp
#if GSM
    while(!GSM_PPP_ConnectProcess());
#endif

          pd = pppOverSerialOpen(0, linkStatusCB, &connected);




________________________________
 From: yueyue papa <yueyuepapa-Re5JQEeQq