FD.io VPP  v19.08-27-gf4dcae4
Vector Packet Processing
vlib_process_doc.h File Reference

Cooperative multi-tasking thread support.

Go to the source code of this file.

Detailed Description

Cooperative multi-tasking thread support.

Vlib provides a lightweight cooperative multi-tasking thread model. Context switching costs a setjmp/longjump pair. It's not unreasonable to put vlib threads to sleep for 10us.

The graph node scheduler invokes these processes in much the same way as traditional vector-processing run-to-completion graph nodes; plus-or-minus a setjmp/longjmp pair required to switch stacks. Simply set the vlib_node_registration_t type field to VLIB_NODE_TYPE_PROCESS. Process is a misnomer; these are threads.

As of this writing, the default stack size is 2<<15; 32kb. Initialize the node registration's process_log2_n_stack_bytes member as needed. The graph node dispatcher makes some effort to detect stack overrun. We map a no-access page below each thread stack.

Process node dispatch functions are expected to be while(1) { } loops which suspend when not otherwise occupied, and which must not run for unreasonably long periods of time. Unreasonably long is an application-dependent concept. Over the years, we have constructed frame-size sensitive control-plane nodes which will use a much higher fraction of the available CPU bandwidth when the frame size is low. Classic example: modifying forwarding tables. So long as the table-builder leaves the forwarding tables in a valid state, one can suspend the table builder to avoid dropping packets as a result of control-plane activity.

Process nodes can suspend for fixed amounts of time, or until another entity signals an event, or both. See the example below.

When running in VLIB process context, one must pay strict attention to loop invariant issues. If one walks a data structure and calls a function which may suspend, one had best know by construction that it cannot change. Often, it s best to simply make a snapshot copy of a data structure, walk the copy at leisure, then free the copy.

Here's an example:

    #define EXAMPLE_POLL_PERIOD 10.0

    static uword
    example_process (vlib_main_t * vm, vlib_node_runtime_t * rt,
                     vlib_frame_t * f)
      f64 poll_time_remaining;
      uword event_type, *event_data = 0;

      poll_time_remaining = EXAMPLE_POLL_PERIOD;
      while (1)
          int i;

Sleep until next periodic call due,
or until we receive event(s)

          poll_time_remaining =
            vlib_process_wait_for_event_or_clock (vm, poll_time_remaining);

          event_type = vlib_process_get_events (vm, &event_data);
          switch (event_type)
            case ~0:            // no events => timeout

            case EVENT1:
              for (i = 0; i < vec_len (event_data); i++)
                handle_event1 (mm, event_data[i]);

            case EVENT2:
              for (i = 0; i < vec_len (event_data); i++)
                handle_event2 (vm, event_data[i]);

... and so forth for each event type

This should never happen...
              clib_warning ("BUG: unhandled event type %d",
          vec_reset_length (event_data);

Timer expired, call periodic function
          if (vlib_process_suspend_time_is_zero (poll_time_remaining))
              example_periodic (vm);
              poll_time_remaining = EXAMPLE_POLL_PERIOD;
      return 0;

    static VLIB_REGISTER_NODE (example_node) = {
      .function = example_process,
      .name = "example-process",

In this example, the VLIB process node waits for an event to occur, or for 10 seconds to elapse. The code demuxes on the event type, calling the appropriate handler function.

Each call to vlib_process_get_events returns a vector of per-event-type data passed to successive vlib_process_signal_event calls; vec_len (event_data) >= 1. It is an error to process only event_data[0].

Resetting the event_data vector-length to 0 by calling vec_reset_length (event_data) - instead of calling vec_free (...)

  • means that the event scheme doesn t burn cycles continuously allocating and freeing the event data vector. This is a common coding pattern, well worth using when appropriate.

Definition in file vlib_process_doc.h.