Re: [lwip-users] lwIP hangs on transfers with many short packets

[multiplex]

Kieran Mansley wrote:
> 
> 
> On 26 May 2010, at 06:54, multiplex wrote:
> 
>> 
>> 
>> 
>> Kieran Mansley wrote:
>>> 
>>> On Thu, 2010-05-20 at 11:10 -0700, multiplex wrote:
>>>> Any suggestions ?
>>> 
>>> I would use the LWIP_STATS code to identify if there is some resource
>>> within lwIP that is running low.
>>> 
>>> Kieran
>>> 
>>> 
>> 
>> Thanks Kieran - Here are two printouts for the display_stats(). The first
>> one is when everything is still going well; the second when it hangs
> 
> Unfortunately there's nothing there to point a finger a what might have
> gone wrong.  All the resources seem fine: there is plenty of everything
> available, and it doesn't seem like anything has run out at any point.
> 
>> (note:
>> it is only lwIP that hangs - I still have serial - and if we disconnect
>> and
>> wait really really really long, the connection will closely nicely).
> 
> Can you explain that in more detail?  Can you be precise about what you
> mean by disconnect and how long a really really really long time is?  If
> lwIP has hung, I wouldn't expect it to do anything no matter how long you
> wait.
> 
> Thanks
> 
> Kieran
> 
> 

>From the system logs, I get the message that the connection was shut down
correctly - my interpretation was that, due to the other side disconnecting
when you close the telnet session, at a certain we hit th timeout (which for
TCP can be really long). 

Too bad - nothing seemed wrong to me either. Any chance we might be losing
ACKs? And we just start waiting, hoping for them to come back? If so - I am
currently looking at switching to UDP. 

One last thing. If we take Xilinx' lastezd and greatezd SDK 12.1 - we get
1.3.2 as well. Would that help, IYO?
-- 
View this message in context: 
http://old.nabble.com/lwIP-hangs-on-transfers-with-many-short-packets-tp28607115p28684958.html
Sent from the lwip-users mailing list archive at Nabble.com.