Re: [lwip-users] LWIP Accept Failures

[Neil Burton]

Hi folks,

Just as an update it appears I've found the culprit for this problem. It's a
previously reported bug described in the archives here:

http://mail.gnu.org/archive/html/lwip-users/2003-01/msg02149.html

Neil B


> Neil,
>
> Are you collecting stats? Take a look at the lwip_stats structure and
> see if you are running out of any kind of memory. I would guess your
> problem could be in this area.
>
> David.
>
>
> Neil Burton wrote:
>
> >Hi all,
> >
> >I'm working with a port of LWIP onto a Z80182 platform, which until now
has
> >been very successful. However, recent testing has thrown up an
unpredictable
> >problem whereby LWIP seems to accept (briefly) an incoming TCP
connection,
> >and then almost immediately abort it. I TCPDUMP'd one such occasion on a
> >relevant gateway, with the following trace:
> >
> >15:22:24.988891 192.168.1.5.1169 > 10.128.0.15.8082: S
> >3696301131:3696301131(0) win 64240 <mss 1460,nop,nop,sackOK> (DF)
> >15:22:27.259441 10.128.0.15.8082 > 192.168.1.5.1169: S 2214116:2214116(0)
> >ack 3696301132 win 1024 <mss 128> (DF)
> >15:22:27.295033 192.168.1.5.1169 > 10.128.0.15.8082: . ack 1 win 64256
(DF)
> >15:22:27.332945 192.168.1.5.1169 > 10.128.0.15.8082: P 1:4(3) ack 1 win
> >64256 (DF)
> >15:22:27.958011 10.128.0.15.8082 > 192.168.1.5.1169: R 2:2(0) ack 1 win 0
> >(DF)
> >15:22:28.003770 10.128.0.15.8082 > 192.168.1.5.1169: R 2:2(0) ack 4 win 0
> >(DF)
> >15:23:30.068880 192.168.1.5.1170 > 10.128.0.15.8082: S
> >3711228507:3711228507(0) win 64240 <mss 1460,nop,nop,sackOK> (DF)
> >15:23:33.261482 192.168.1.5.1170 > 10.128.0.15.8082: S
> >3711228507:3711228507(0) win 64240 <mss 1460,nop,nop,sackOK> (DF)
> >15:23:33.640154 10.128.0.15.8082 > 192.168.1.5.1170: S 2320082:2320082(0)
> >ack 3711228508 win 1024 <mss 128> (DF)
> >15:23:33.681788 192.168.1.5.1170 > 10.128.0.15.8082: . ack 1 win 64256
(DF)
> >15:23:33.704224 10.128.0.15.8082 > 192.168.1.5.1170: S 2426049:2426049(0)
> >ack 3711228508 win 1024 <mss 128> (DF)
> >15:23:33.717342 192.168.1.5.1170 > 10.128.0.15.8082: P 1:4(3) ack
4294861330
> >win 64256 (DF)
> >15:23:33.740220 192.168.1.5.1170 > 10.128.0.15.8082: . ack 4294861330 win
> >64256 (DF)
> >15:23:34.378489 10.128.0.15.8082 > 192.168.1.5.1170: R
> >4294861331:4294861331(0) ack 1 win 0 (DF)
> >15:23:34.424260 10.128.0.15.8082 > 192.168.1.5.1170: R
> >4294861331:4294861331(0) ack 4 win 0 (DF)
> >15:23:34.464386 10.128.0.15.8082 > 192.168.1.5.1170: R
> >4294861331:4294861331(0) ack 4 win 0 (DF)
> >
> >192.168.1.5 is a Win2k server machine running one end of our application
(a
> >Win32 gateway app), and 10.128.0.15 is the remote, Z80-based platform
> >running LWIP. At the Windows end, the connection is established, and then
> >immediately reports reset by peer. Linux telnet attempting to connect to
the
> >LWIP platform shows the exact same symptoms. However, ICMP pings still
work
> >fine to the LWIP platform, and connections *from* LWIP into the Win2k
> >gateway app also work fine. Only the LWIP listener has a problem.
> >
> >Notably, none of the callbacks into my application are fired from LWIP
> >during this process. Whatever is happening, is happening entirely in the
> >LWIP system.
> >
> >This is my first post to the list so apologies if the information is
> >incomplete, and I didn't actually write the port though I did integrate
the
> >supplied code into my app. I'd appreciate any and all insight into the
> >nature of this problem while I'm digging around trying to solve it.
> >
> >Thanks in advance,
> >
> >Neil Burton