Well, yeah I didn't mean I fixed the fragmentation problem, I just avoided it. :) Maybe I will try to set MTU to 1500 just to see what is happening.

Thanks!


On 8.3.2017 17:36, goldsimom wrote:
For your MTU, 1446 is correct. I'd rather fix the MTU to be 1500 though.
And you don't fix the fragmentation problem. I think this is fixed in 2.0.0 though.

Simonl)

Am 8. März 2017 12:09:24 MEZ schrieb pekez <nenad.pe...@rt-rk.com>:

    Well, just maybe I have succeeded to find the solution to this
    problem... However, since I don't know exactly how TCP works under
    the hood, maybe I could cause some other problems, but for now
    it's working. Nevertheless, I decided to reduce TCP MSS from 1460
    to 1446, and ip_frag() is not called any more at all. At least,
    not so far, I have been testing the whole night, so that's good
    starting point.

    I would be grateful if someone could explain me whether this
    change is valid, I mean, if that's not something I should be
    changing (reducing) for some reason?

    On 6.3.2017 11:09, pekez wrote:

    I forgot to mention that I use Xilinx lwIP example applications
    (XAPP 1026
    
<https://www.xilinx.com/support/documentation/application_notes/xapp1026.pdf>).
    Basically the only thing I have changed is in their TCP server
    app, instead of only receiving data, I made server only to send
    data (literally, instead of lwip_recv, I put lwip_send).

    Also, I don't know whether this means anything, but bug
    reproduction is kinda interesting. I am always able to bring up
    TCP client several consecutive times and successfully exchange
    data (ip_frag never gets called during the exchange) and close
    TCP client. Then I don't do anything for about 5-10 minutes. Then
    I bring up TCP client again, and TCP server get stuck in
    ip_frag() when start sending data.


    On 6.3.2017 10:57, pekez wrote:
    1) netif->mtu = XEMACPS_MTU - XEMACPS_HDR_SIZE;

    #define XEMACPS_MTU             1500U    /* max MTU size of
    Ethernet frame */
    #define XEMACPS_HDR_SIZE        14U    /* size of Ethernet
    header */

    2) how can I check exact length of pbufs?

    3) Yep, IP_FRAG_USES_STATIC_BUF and LWIP_NETIF_TX_SINGLE_PBUF
    are both 0.

    Thanks!

    On 6.3.2017 10:51, Simon Goldschmidt wrote:
    pekez wrote:
    Anyone? Not a single person can tell me something about this
    problem?
    Does not seem so.

    Could you give us more info:
    - what is netif->mtu set to
    - exact length of pbufs in a chaing > 1480
    - setting of IP_FRAG_USES_STATIC_BUF and
    LWIP_NETIF_TX_SINGLE_PBUF (I assume both are 0 - default)
    That way I could try to have a quick look.


    Simon

    _______________________________________________
    lwip-users mailing list
    lwip-users@nongnu.org
    https://lists.nongnu.org/mailman/listinfo/lwip-users


    _______________________________________________
    lwip-users mailing list
    lwip-users@nongnu.org
    https://lists.nongnu.org/mailman/listinfo/lwip-users



    _______________________________________________
    lwip-users mailing list
    lwip-users@nongnu.org
    https://lists.nongnu.org/mailman/listinfo/lwip-users


--
Diese Nachricht wurde von meinem Android-Gerät mit K-9 Mail gesendet.


_______________________________________________
lwip-users mailing list
lwip-users@nongnu.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

_______________________________________________
lwip-users mailing list
lwip-users@nongnu.org
https://lists.nongnu.org/mailman/listinfo/lwip-users

Reply via email to