IP Datagram reassembly issue -- need information on mbuf settings in driver before passing packets up to stack. - VxWorks

This is a discussion on IP Datagram reassembly issue -- need information on mbuf settings in driver before passing packets up to stack. - VxWorks ; Hi, Here is a brief description: END for dedicated h/w picks frames from ASIC and passes them up to IP layer via MUX. Max MTU on link is 2K. If I ping my node with payload it works (no reassembly ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: IP Datagram reassembly issue -- need information on mbuf settings in driver before passing packets up to stack.

  1. IP Datagram reassembly issue -- need information on mbuf settings in driver before passing packets up to stack.

    Hi,

    Here is a brief description:

    END for dedicated h/w picks frames from ASIC and passes them up to IP
    layer via MUX. Max MTU on link is 2K. If I ping my node with payload it works (no reassembly needed) but fails when I go beyond 2K
    (requires reassembly). Platform is VxWorks based and I am using mbufs
    to pass data between driver and protocol/s.

    I suspect that I am missing some settings of mbuf (mBlkHdr fields or
    length adjustments etc).

    Any clues?

    Thanks,
    Bhal


  2. Re: IP Datagram reassembly issue -- need information on mbuf settings in driver before passing packets up to stack.

    On Jun 19, 4:49 pm, Bhal wrote:
    > Hi,
    >
    > Here is a brief description:
    >
    > END for dedicated h/w picks frames from ASIC and passes them up to IP
    > layer via MUX. Max MTU on link is 2K. If I ping my node with > payload it works (no reassembly needed) but fails when I go beyond 2K
    > (requires reassembly). Platform is VxWorks based and I am using mbufs
    > to pass data between driver and protocol/s.
    >
    > I suspect that I am missing some settings of mbuf (mBlkHdr fields or
    > length adjustments etc).
    >
    > Any clues?
    >
    > Thanks,
    > Bhal


    I do not think END has any thing to do with fragments. What kind of
    data link layer you are using. May be it had an issue in handling
    frames with more than MTU.
    Check your pool configuration for buffer sizes and number of buffers,
    may be you have less number of 2K buffers to handle fragments.

    Ranga


+ Reply to Thread