Re: [lwip-users] Windows 7 Startup crashes LwIP and Free RTOS

[Marco Jakobs]

Dear Kieran,

you may be right - with the debug function, the packets are still 
arriving in that way, but the timing inside the stack differs and some 
packets may be lost due to full buffers. So this is no way to get behind 
the problem.

I may try to disable my watchdog functions so i'll get no reset and 
maybe i can get in this state with the JTAG attached and debugger 
running. The problem is: What to do then - as i'm not in the internals 
of the stack. If there is somewhere some valueable information that may 
help, just tell me what i should look for and i'll try.

I've hammered the LwIP with generated broadcast packets every 20ms 
(using the packet generator of NetScanTools) - no problem at all. So it 
must be maybe a very tight timing. Is it theoretically possible that it 
did not yet return from the interrupt of a reception when the next 
packet arrives and this causes some trouble?


@Simon:
It's not a direct reboot - *i suspect LwIP is ending in an endless loop, 
not suspending its task*. As it has a higher priority in the RTOS of my 
implementation, the other tasks will not be called anymore, which leads 
to a panic of my watchdog task -> there the restart is made. As my 
watchdog task with the highest priority of all is still happily running, 
it underlines this theory!

Thank you all for looking at my EMAC routines, so i'll push this back.

But @Richard: Where can i find these improved routines, so i can verify 
the changes to the ones i'm using? I would be happy if you may send them 
to me or point me to a download.

Marco


Am 29.06.2010 22:47, schrieb Kieran Mansley:
>   this seems to prevent it also from happening.
>    
> In that case abandon the debug logs, but if possible keep logging error messages and assertion failures.  Try and get a wireshark capture that corresponds to the failure if you think it is related to particular traffic.  My guess is that it's a timing dependent crash that is triggered by receiving a few packets quickly, so wireshark might not help us much, so if it's difficult then don't worry.  The most useful way to solve the problem might be to try and work out how to provoke it and so make it more reproducible.
>
> Kieran
> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/lwip-users
>