Re: [lwip-users] protocol issue

[Laurent BIERGE]

Hi again,

I've played with LWIP_DEPUG a bit and i'm now able to see what's happening.
Reminder: i try to use a LWIP UDP echo program which is on a remote equipment. 
I send UDP datagrams from my computer.
          UDP datagrams don't seem to reach the remote equipment 
          Actually i don't have all the rights on my computer.

So what's exactly happening is that when i send a UDP datagrams from my 
computer to the equipment, my computer send ARP request which are answered by 
LWIP correctly. The ARP cache on my computer is filled well after. It seems 
that even after that my computer want's to fill Netbios name service table. So 
my computer send UDP datagrams on port 137 and the table is filled well after.
What's strange is that the ARP request and Netbios request are sent 3 times in 
a row each time. Besides when i try to send another datagram the computer send 
ARP/Netbios request again (even if it's just one second after).

I don't know the ARP/Netbios routine at all. But my guess is that since i don't 
have all the rights on the computer, it try to fill his ARP/Netbios table but 
with no success. And sometimes it works (bypassing security for an instant) and 
the computer send all the datagrams at once.

So i know for sure that it doesn't concern LWIP now, but maybe you have an idea 
of what i'm facing ?

Regards,

Laurent  

Ps: By the way the "pcb issue" came from the Netbios request. So i was assuming 
that it was my datagrams (which are sent on the 6020 port) but it was Netbios 
datagrams (on port 137).




----- Mail original -----
De: "Simon Goldschmidt" <address@hidden>
À: "lwip-users" <address@hidden>
Envoyé: Mardi 19 Juillet 2016 12:25:58
Objet: Re: [lwip-users] raw UDP pcb issue(s)

Laurent BIERGE wrote:
> So to be sure, is there absolutly no way that lwip could trigger 
> ethernet_input with some king of delay ?

No. Any delay between packet arrival an ethernet_input can only come from your 
netif driver or from an incorrect port to your os or hardware; or by you (or 
some code) blocking the lwip thread for too long (but that won't duplicate the 
packets, just delay them, so 7* send -> 7* RX processing).


Simon

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