Re: [lwip-users] PPP - TCP

[Norberto R. de Goes Jr.]

Hi Sylvain, all rigth? I am here again...

Sorry, but the problem continues. Please but I have a suspicion. First I explain you the status.

1) My setup is very simillar to your. I have two VM´s (Ubuntu) in a Win-7 host. Serial

using the "Host pipe" VirtualBox method.

2) Today (jan/21/2016), I updated my lwip and lwip-contrib git-repos changing just the two files sent to you (lwipopts.h and simhost.c) in the ".../lwip-contrib/ports/unix/proj/unixsim" directory. Master branch.

3) In the VM#1, I run the pppd (linux), passive mode. Just after, the "simhost" app is executed in VM#2.  The link ppp is established, with no problem. I think that validates the serial and PPP connections.

The problem is with TCP mode. Today I use also the netcat tool like you. No problem (example in client side:   echo "TEST" | nc 192.168.123.21 7).

I have a suspicion when the client opens and uses the same connection to exchange several echo messages different from "netcat" behavior. I think. 

Thus I picked up a echo-client int the internet (thank to Silver Moon). When I use this "original client" the problem happens after 3 or plus echo messages like in my implementation.

But when I use the "changed client" (open and close socket for each echo message/prompt), no problem, all works fine.

Please, could you test the two attached clients files in your setup (on pppd server host) ? Is there sense in my suspicion?

OBS:

   > gcc echo_client_original.c -o echoClientOriginal

   > gcc echo_client_changed.c -o echoClientChanged

Thanks.

Norberto

2016-01-20 22:16 GMT-02:00 Sylvain Rochet <address@hidden>:

Hi Norberto,

On Mon, Jan 18, 2016 at 01:01:12PM -0200, Norberto R. de Goes Jr. wrote:
>
> I would like remember you the code used to reproduce the PPP/TCP error is
> practically the same of the lwip git-repos (lwip e contrib), last Friday
> (jan/15/2016) version, only the "simhost.c" and "lwipopts.h" files with
> little modifications (attached again), or better, I am not using my code to
> show you the reported problem, but the behavior is the same.

I tested your setup with VirtualBox using two Debian VMs and a serial
bridge using the "Host pipe" VirtualBox method.

I started the pppd deamon on one host this way:

/usr/sbin/pppd silent local nocrtscts noauth  \
    192.168.123.20:192.168.123.21 /dev/ttyS0

And the latest lwIP Git with your two changed files in the unixsim
project on the other host.


This is what I have on the pppd server host:

Jan 21 01:00:04 debian pppd[1797]: pppd 2.4.6 started by root, uid 0
Jan 21 01:00:04 debian pppd[1797]: Using interface ppp0
Jan 21 01:00:04 debian pppd[1797]: Connect: ppp0 <--> /dev/ttyS0
Jan 21 01:01:51 debian pppd[1797]: local  IP address 192.168.123.20
Jan 21 01:01:51 debian pppd[1797]: remote IP address 192.168.123.21

And what I have on the lwIP host:

# ./simhost
System initialized.
SNMP private MIB start, detecting sensors.
TFCP/IP initialized.
Applications started.
ppp_link_status_cb: PPPERR_NONE
   our_ip4addr = 192.168.123.21
   his_ip4addr = 192.168.123.20
   netmask     = 255.255.255.255


I checked TCP, UDP, using various tools over this link (netcat, telnet),
even at "full" speed, and it works perfectly fine.

Sylvain

_______________________________________________
lwip-users mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/lwip-users

--

Norberto R. de Goes Jr.
CPqD - DRC
Tel.: +55 19 3705-4241 / Fax: +55 19 3705-6125
address@hidden
www.cpqd.com.br

echo_client_original.c
Description: Text Data

echo_client_changed.c
Description: Text Data