SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


    [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

    RE: Urgent as Framing Hint



    All,
    
    What would the TCP API look like for a Framing TCP protocol?  I am willing
    to assume such a framing scheme is possible either by means of a TCP option,
    special checksums, urgent pointer redefinitions, or whatever clever concept
    makes framing the encapsulation PDUs possible.  The next issue is how do you
    employ this framing scheme within existing stacks.
    
    Here is my guess at this topic:
    
    One, modify the Socket Open to include a pointer to the Application Specific
    Data Steering & Parsing Routine together with a pointer to the Encapsulation
    Descriptors.  This binds the routine to the port.
    
    Two, create a signaling between the TCP stack and this routine so that
    buffer can be allocated, data directed, and upon the filling the allocated
    buffer, the TCP stack returns the TCP Socket receive.  The parsing is
    accomplished by the size of the allocated buffer.  There also needs to be a
    means to redefine mid-stream the size of this allocation buffer if
    truncation or extension is required or perhaps just a general error signal
    relayed to the application via the descriptor.
    
    Three, a definition of the Encapsulation Descriptor list used to communicate
    between the Application Specific Data Steering & Parsing Routine and the
    frame enhanced application.  Each item within the this descriptor listed
    would be updated atomically.  There would be a simple completion flag
    signaling completion by the ASDSPR so that the decriptor location can be
    reused.  There should also be an error field within the descriptor to
    indicate either communication or parsing errors were found to cause the
    application to then re-parse the returned data itself.  There may also be an
    advantage in including a offset to descriptor within the Socket receive
    function return.
    
    I have included a graphical outline of such a scheme.  Any ideas as to how a
    Framing TCP stack can be improved?  What details should be added or amended?
    
    Doug
    

    Framed TCP.gif



Home

Last updated: Tue Sep 04 01:06:12 2001
6315 messages in chronological order