Re: [lwip-users] Stability of master branch code

[address@hidden]

Robert Deschambault wrote:
> Was there anything else you would like me to try? I have confirmed 
> that I am freeing the pbuf in our code.

While I have forgotton to answer (sorry for that), I really can't tell 
what to check next. When the MEMP_TCPIP_MSG_INPKT pool is empty, it 
should get filled again in tcpip_thread. This could either mean your 
sys_mbox implementation has a problem (e.g. the mbox is full but your 
sys_mbox_trypost() still returns ERR_OK), or tcpip_thread fails to pull 
these messages from the mbox (e.g. it doesn't run any more, this could 
just be a higher thread blocking it or stack overflow making it jump to 
invalid code but your netif ISR still works).

Maybe you should check why you can't access your device to see the stats 
counters. There could be more problems than just the 
MEMP_TCPIP_MSG_INPKT pool being empty (which is a rather normal thing on 
RX overload).

Simon