Re: [lwip-users] One active TCP connection at a time

[Jeffrey Nichols]

Hi Simon,

I think that the issue was that I was calling tcp_close() in tcp_err 
callback. At least I haven't seen it since I fixed that.

I had a similar deadlocking issue that I tracked down to running out of 
tcp pcbs, which I fixed by increasing MEMP_NUM_TCP_PCB from 5 to 30.

Everything seems to be stable now.

Thanks,
Jeff

On 11/4/2019 3:30 PM, address@hidden wrote:
> Am 31.10.2019 um 20:59 schrieb Jeffrey Nichols:
> > Replying to myself here. This did indeed cause issues.
> >
> > Calling tcp_close() on the listening pcb inside the tcp_accept callback
> > I occasionally get a deadlock inside tcp_remove_listener(), where the
> > linked list is circular: pcb->next is equal to pcb.
> >
> > Anyone have any ideas for how to debug/fix this?
>
> That should not happen. When getting such circular lists, most probably
> a threading issue is the cause.
>
> Regards,
> Simon
>
> _______________________________________________
> lwip-users mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/lwip-users