Re: [lwip-users] Upgrading to 0.6.1

[David Ryan]

Hi Leon,

Thanks for the quick response.  I updated to 0.6.2 and had another go. 
I'm getting the same response.  A second tcp data packet with a random 
number of the character sent in the first packet.  

pc    ----->  'x'  ------>   ps2        single character sent to ps2.
pc    <------ 'x' --------   ps2        single character returned.
pc    ------- ACK  ----> ps2        packet acknowledged
pc    <----- 'xxxx'  ------ ps2         random number of characters
returned in new packet.
pc    -------- ACK -----> ps2        packet acknowledged

Have you got any ideas of the most likely place this could happen?  The 
implementation is mult-threaded and uses the socket api.  So I'm 
assumming that this is in the socket send api, or tcp send code.  

There must be a portion of timing in there because the 'xxxx' returned 
in the example above can be between one additional 'x' character and 
about seven 'xxxxxxx'.  Its always exactly the same character sent. 
Packet one is sent correctly with one packet, packet two is sent with 
multiple characters.

I'm sure this is something stupid.  I'm going to go check my lwipopts.h 
file and make sure that I'm not missing anything.  Any ideas would be 
appreciated though.

David.

address@hidden wrote:

> Hello David,
>
> please check out "STABLE" branch from CVS
>
> These are two major bug fixes since 0.6.1:
>
> HISTORY
>
> (current)
>
>   ++ Bugfixes:
>
>   * TCP has been fixed to deal with the new use of the pbuf->ref
>     counter.
>
>   * DHCP dhcp_inform() crash bug fixed.
>
> (0.6.1)
>
>
> Regards,
>
> Leon Woestenberg
>
>
>
>
>
> *David Ryan <address@hidden>*
> Sent by: address@hidden
>
> 17-04-2003 03:10
> Please respond to Mailing list for lwIP users
>
>        
>         To:        lwip <address@hidden>
>         cc:        
>         Subject:        [lwip-users] Upgrading to 0.6.1
>
>
>
>
>
> Hi all,
>
> I've finally found time to do some testing of 0.6.1 since upgrading from
> 0.6.0-rc1 with my ps2 implementation.  I first discovered that I needed
> to set netif->hwaddr_len or ARPs would return with mostly junk.  After
> resolving that issue I came up against a very strange issue...
>
> My basic test is to connect to an echo server and then telnet to it from
> windows, and type in a few characters.  What I found was that a single
> character would result in a random number of characters in return.  I
> first checked my echo server to ensure that it was receiving one
> character and sending one character, which it did.  After that I started
> checking with Ethereal.  I found the following
>
> pc    ----->  'x'  ------>   ps2        single character sent to ps2.
> pc    <------ 'x' --------   ps2        single character returned.
> pc    ------- ACK  ----> ps2        packet acknowledged
> pc    <----- 'xxxx'  ------ ps2         random number of characters
> returned in new packet.
> pc    -------- ACK -----> ps2        packet acknowledged
>
> This was consistant with every packet sent.  My first assumption is that
> this is probably related to how the driver should interface with pbufs.
>  I haven't yet changed the driver from 0.6.0 code.   Is this likely, or
> should I be looking deeper into lwip?
>
> If you think it is the driver, can someone suggest which driver I should
> base my code?
>
> What other changes from 0.6.0 to 0.6.1 should I look out for?
>
> Thanks,
> David.
>
>
>
>
>
>
> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://mail.nongnu.org/mailman/listinfo/lwip-users
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://mail.nongnu.org/mailman/listinfo/lwip-users
>