diff mbox

[API-NEXT,PATCHv12,1/3] api: tm: add tm API definitions

Message ID 1447805799-31047-2-git-send-email-bill.fischofer@linaro.org
State Accepted
Commit 90a32ca40fe25d73ee9bdbf263f8dd522ca8db87
Headers show

Commit Message

Bill Fischofer Nov. 18, 2015, 12:16 a.m. UTC
This introduces an API for configuring and using Traffic Management
systems.

The purpose of this API is as a general packet scheduling system that
accepts packets from input queues and applies strict priority
scheduling, weighted fair queuing scheduling and/or bandwidth controls
to decide which input packet should be chosen as the next output
packet and when this output packet can be sent onwards.

Signed-off-by: Barry Spinney <spinney@ezchip.com>
Signed-off-by: Bill Fischofer <bill.fischofer@linaro.org>
---
 include/odp.h                                      |    1 +
 include/odp/api/packet.h                           |   69 +
 include/odp/api/traffic_mngr.h                     | 1611 ++++++++++++++++++++
 platform/linux-generic/Makefile.am                 |    3 +
 .../linux-generic/include/odp/plat/packet_types.h  |   11 +
 .../include/odp/plat/traffic_mngr_types.h          |  185 +++
 platform/linux-generic/include/odp/traffic_mngr.h  |   35 +
 .../linux-generic/include/odp_packet_internal.h    |    5 +
 8 files changed, 1920 insertions(+)
 create mode 100644 include/odp/api/traffic_mngr.h
 create mode 100644 platform/linux-generic/include/odp/plat/traffic_mngr_types.h
 create mode 100644 platform/linux-generic/include/odp/traffic_mngr.h
diff mbox

Patch

diff --git a/include/odp.h b/include/odp.h
index 2adcb8b..4a93c23 100644
--- a/include/odp.h
+++ b/include/odp.h
@@ -55,6 +55,7 @@  extern "C" {
 #include <odp/random.h>
 #include <odp/errno.h>
 #include <odp/thrmask.h>
+#include <odp/traffic_mngr.h>
 #include <odp/spinlock_recursive.h>
 #include <odp/rwlock_recursive.h>
 #include <odp/std_clib.h>
diff --git a/include/odp/api/packet.h b/include/odp/api/packet.h
index 85cc9c9..9c63b5f 100644
--- a/include/odp/api/packet.h
+++ b/include/odp/api/packet.h
@@ -48,6 +48,26 @@  extern "C" {
  * Invalid packet segment
  */
 
+ /**
+  * @typedef odp_packet_color_t
+  * Color of packet for shaper/drop processing
+  */
+
+ /**
+  * @def ODP_PACKET_GREEN
+  * Packet is green
+  */
+
+ /**
+  * @def ODP_PACKET_YELLOW
+  * Packet is yellow
+  */
+
+ /**
+  * @def ODP_PACKET_RED
+  * Packet is red
+  */
+
 /*
  *
  * Alloc and free
@@ -729,6 +749,55 @@  odp_packet_seg_t odp_packet_last_seg(odp_packet_t pkt);
  */
 odp_packet_seg_t odp_packet_next_seg(odp_packet_t pkt, odp_packet_seg_t seg);
 
+/**
+ * Get packet color
+ *
+ * @param pkt Packet handle
+ * @return packet color
+ */
+odp_packet_color_t odp_packet_color(odp_packet_t pkt);
+
+/**
+ * Set packet color
+ *
+ * @param pkt Packet handle
+ * @param color Color to set
+ */
+void odp_packet_color_set(odp_packet_t pkt, odp_packet_color_t color);
+
+/**
+ * Get drop eligible status
+ *
+ * @param pkt Packet handle
+ * @return Packet drop eligibility status
+ * @retval 0 Packet is not drop eligible
+ * @retval 1 Packet is drop
+ */
+odp_bool_t odp_packet_drop_eligible(odp_packet_t pkt);
+
+/**
+ * Set drop eligible status
+ *
+ * @param pkt Packet handle
+ * @param status Drop eligibility status
+ */
+void odp_packet_drop_eligible_set(odp_packet_t pkt, odp_bool_t status);
+
+/**
+ * Get shaper length adjustment
+ *
+ * @param pkt Packet handle
+ * @return Shaper adjustment (-128..127)
+ */
+int8_t odp_packet_shaper_len_adjust(odp_packet_t pkt);
+
+/**
+ * Set shaper length adjustment
+ *
+ * @param pkt Packet handle
+ * @param adj Signed adjustment value
+ */
+void odp_packet_shaper_len_adjust_set(odp_packet_t pkt, int8_t adj);
 
 /*
  *
diff --git a/include/odp/api/traffic_mngr.h b/include/odp/api/traffic_mngr.h
new file mode 100644
index 0000000..2459a8b
--- /dev/null
+++ b/include/odp/api/traffic_mngr.h
@@ -0,0 +1,1611 @@ 
+/** Copyright (c) 2015, Linaro Limited
+ * All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#ifndef ODP_TRAFFIC_MNGR_H_
+#define ODP_TRAFFIC_MNGR_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <odp/std_types.h>
+#include <odp/packet_io.h>
+
+/**
+ * @file
+ *
+ */
+
+/** @defgroup odp_traffic_mngr ODP TRAFFIC MNGR
+ * @{
+ *
+ * An API for configuring and using Traffic Management systems
+ *
+ * This file forms a simple interface for creating, configuring and using
+ * Traffic Management (TM) subsystems.  By TM subsystem it is meant a general
+ * packet scheduling system that accepts packets from input queues and applies
+ * strict priority scheduling, weighted fair queueing scheduling and/or
+ * bandwidth controls to decide which input packet should be chosen as the
+ * next output packet and when this output packet can be sent onwards.
+ *
+ * A given platform supporting this TM API could support one or more pure
+ * hardware based packet scheduling systems, one or more pure software
+ * based systems or one or more hybrid systems - where because of
+ * hardware constraints some of the packet scheduling is done in hardware
+ * and some is done in software.  In addition, there may also be additional
+ * API's beyond those described here for (a) controlling advanced capabilities
+ * supported by specific hardware, software or hybrid subsystems or (b)
+ * dealing with constraints and limitations of specific implementations.
+ * The intention here is to be the simplest API that covers the vast majority
+ * of packet scheduling requirements.
+ *
+ * Often a TM subsystem's output(s) will be directly connected
+ * to a device's physical (or virtual) output interfaces/links, in which case
+ * sometimes such a system will be called an Egress Packet Scheduler or an
+ * Output Link Shaper, etc..  While the TM subsystems configured by this API
+ * can be used in such a way, this API equally well supports the ability to
+ * have the TM subsystem's outputs connect to other TM subsystem input queues
+ * or general software queues or even some combination of these three cases.
+ *
+ * <H2>TM Algorithms</H2>
+ *
+ * The packet scheduling/dropping techniques that can be applied to input
+ * traffic include any mixture of the following:
+ * <ol>
+ * <li> Strict Priority scheduling.
+ * <li> Weighted Fair Queueing scheduling (WFQ).
+ * <li> Bandwidth Shaping.
+ * <li> Weighted Random Early Discard (WRED).
+ * </ol>
+ * Note that Bandwidth Shaping is the only feature that can cause packets
+ * to be "delayed", and Weighted Random Early Discard is the only feature
+ * (other than input queues becoming full) that can cause packets to be
+ * dropped.
+ *
+ * <H3>Strict Priority Scheduling</H3>
+ * Strict Priority Scheduling (or just priority for short), is a technique
+ * where input queues and the packets from them, are assigned a priority
+ * value in the range 0 .. ODP_TM_MAX_PRIORITIES - 1.  At all times packets
+ * the the smallest priority value will be chosen ahead of packets with a
+ * numerically larger priority value.  This is called strict priority
+ * scheduling because the algorithm strictly enforces the scheduling of
+ * higher priority packets over lower priority packets.
+ *
+ * <H3>Bandwidth Shaping</H3>
+ * Bandwidth Shaping (or often just Shaping) is the term used here for the
+ * idea of controlling packet rates using single rate and/or dual rate token
+ * bucket algorithms.  For single rate shaping a rate (the commit rate) and
+ * a "burst size" (the maximum commit count) are configured.  Then an
+ * internal signed integer counter called the commitCnt is maintained such
+ * that if the commitCnt is positive then packets are eligible to be sent.
+ * When such a packet is actually sent then its commitCnt is decremented
+ * (usually by its length, but one could decrement by 1 for each packet
+ * instead).  The commitCnt is then incremented periodically based upon the
+ * configured rate, so that this technique causes the traffic to be limited
+ * to the commit rate over the long term, while allowing some ability to
+ * exceed this rate for a very short time (based on the burst size) in order
+ * to catch up if the traffic input temporarily drops below the commit rate.
+ *
+ * Dual Rate Shaping is designed to allow  certain traffic flows to fairly
+ * send more than their assigned commit rate when the  scheduler has excess
+ * capacity.  The idea being that it may be better to allow some types of
+ * traffic to send more than their committed bandwidth rather than letting
+ * the TM outputs be idle.  The configuration of Dual Rate Shaping requires
+ * additionally a peak rate and a peak burst size.  The peak rate must be
+ * greater than the related comls mit rate, but the burst sizes have no similar
+ * constraint.  Also for every input priority that has Dual Rate shaping
+ * enabled, there needs to be an additional equal or lower priority (equal or
+ * higher numeric priority value) assigned.  Then if the traffic exceeds its
+ * commit rate but not its peak rate, the "excess" traffic will be sent at the
+ * lower priority level - which by the strict priority algorithm should
+ * cause no degradation of the higher priority traffic, while allowing for
+ * less idle outputs.
+ *
+ * <H3>Weighted Fair Queuing</H3>
+ * Weighted Fair Queuing (WFQ) is used to arbitrate amongst multiple input
+ * packets with the same priority.  Each input can be assigned a weight in the
+ * range MIN_WFQ_WEIGHT..MAX_WFQ_WEIGHT (nominally 1..255) that affects the way
+ * the algorithm chooses the next packet.  If all of the weights are equal AND
+ * all of the input packets are the same length then the algorithm is
+ * equivalent to a round robin scheduling.  If all of the weights are equal
+ * but the packets have different lengths then the WFQ algorithm will attempt
+ * to choose the packet such that inputs each get a fair share of the
+ * bandwidth - in other words it implements a weighted round robin algorithm
+ * where the weighting is based on frame length.
+ *
+ * When the input weights are not all equal and the input packet lengths vary
+ * then the WFQ algorithm will schedule packets such that the packet with
+ * the lowest "Virtual Finish Time" is chosen first.  An input packet's
+ * Virtual Finish Time is roughly calculated based on the WFQ object's base
+ * Virtual Finish Time when the packet becomes the first packet in its queue
+ * plus its frame length divided by its weight.
+ * @code
+ * virtualFinishTime = wfqVirtualTimeBase + (pktLength / wfqWeight)
+ * @endcode
+ * In a system running at full capacity with no bandwidth limits - over the
+ * long term - each input fan-in's average transmit rate will be the same
+ * fraction of the output bandwidth as the fraction of its weight divided by
+ * the sum of all of the WFQ fan-in weights.  Hence larger WFQ weights result
+ * in better "service" for a given fan-in.
+ * @code
+ * totalWfqWeight = 0;
+ * for (each fan-in entity - fanIn - feeding this WFQ scheduler)
+ *     totalWfqWeight += fanIn->sfqWeight;
+ *
+ * fanIn->avgTransmitRate = avgOutputRate * fanIn->sfqWeight / totalWfqWeight;
+ * @endcode
+ *
+ * <H3>Weighted Random Early Discard</H3>
+ * The Weighted Random Early Discard (WRED) algorithm deals with the situation
+ * where an input packet rate exceeds some output rate (including
+ * the case where Bandwidth Shaping limits some output rates).  Without WRED
+ * enabled and configured, the TM system will just implement a tail dropping
+ * scheme whereby whichever packet is unlucky enough to arrive when an TM
+ * input queue is full will be discarded regardless of priority or any other
+ * consideration.
+ * WRED allows one to configure the system to use a better/fairer algorithm
+ * than simple tail dropping.  It works by measuring the "fullness" of
+ * various packet queues and converting this percentage into a probability
+ * of random packet dropping with the help of some configurable parameters.
+ * Then a random number is picked and together with the drop probability,
+ * a decision is made to accept the packet or drop it.
+ * A basic parameterization of WRED requires three parameters:
+ * <ol>
+ * <li> the maximum queue level (which could be either a maximum number of
+ *      packets or a maximum amount of memory (i.e. bytes/buffers) used),
+ * <li> a starting threshold - which is a number in the range 0..100
+ *      representing a percentage of the maximum queue level at which the
+ *      drop probability becomes non-zero,
+ * <li> a drop probability - which is a number in the range 0..100
+ *      representing a probability (0 means no drop and 100 means
+ *      certain drop) - which is used when the queue is near 100% full.
+ * </ol>
+ *
+ * Note that all packet drops for a TM system only occur when a new packet
+ * arrives at a given TM system input queue.  At that time either the WRED
+ * algorithm, if enabled for this input queue, or the "input queue full"
+ * tail drop algorithm will make a drop/no drop decision.  After this point,
+ * any packets not dropped, will at some point be sent out a TM output -
+ * assuming that the topology is fully connected and enabled.
+ *
+ * <H2>Hierarchical Scheduling and tm_nodes</H2>
+ * This API supports the ability to do Hierarchical Scheduling whereby the
+ * final scheduling decision is controlled by equal priority schedulers,
+ * strict priority multiplexers, bandwidth shapers - at multiple levels - all
+ * forming a tree rooted at a single egress object.  In other words, all
+ * tm_queues and tm_nodes have the property that their logical "output" feeds
+ * into one fan-in of a subsequent tm_node or egresss object - forming a proper
+ * tree.  See the following link -
+ * <A HREF="diagram1.svg">Example Tm_node</A> - for an example.
+ *
+ * Multi-level/hierarchical scheduling adds both great control and significant
+ * complexity.  Logically, despite the implication of the tm_node tree
+ * diagrams, there are no queues between the levels of hierarchy.  Instead all
+ * packets are held in their input queue, until such time that the totality of
+ * all of the tm_nodes in the single path from input queue to output object
+ * agrees that this packet should be the next to be chosen to leave the TM
+ * system through the output object "portal".  Hence what flows from level to
+ * level is the "local choice" of what packet/tm_queue should next be
+ * serviced.
+ *
+ * <H3>tm_nodes</H3>
+ * Tm_nodes are the main "entity"/object that a TM system is composed of.
+ * Each tm_node is a mini-TM subsystem of its own, but the interconnection
+ * and interplay of a multi-level "tree" of tm_nodes can allow the user
+ * to specify some very sophisticated behaviours.
+ * Each tm_node can contain a set of scheduler (one per strict priority level),
+ * a strict priority multiplexer, a bandwidth shaper and a WRED component - or
+ * a subset of these.
+ *
+ * In its full generality an tm_node consists of a set of "fan-in" connections
+ * to preceding tm_queues or tm_nodes.  The fan-in for a single tm_node
+ * can range from 1 to many many thousands.  This fan-in is divided first
+ * into a WFQ scheduler per priority level. So if 4 priority levels are
+ * implemented by this tm_node, there would be 4 WFQ schedulers - each with
+ * its own unique fan-in.  After the WFQ schedulers a priority chooser comes
+ * next - where it will always choose the highest priority WFQ output
+ * available.  The output of the priority chooser then feeds a bandwidth
+ * shaper function which then finally uses the shaper's propagation table
+ * to determine its output packet and its priority.  This output could
+ * then be remapped via a priority map profile and then becomes one of the
+ * input fan-in to perhaps another level of tm_nodes, and so on.
+ *
+ * During this process it is important to remember that the bandwidth shaping
+ * function never causes packets to be dropped.  Instead all packet drops
+ * occur because of tm_queue fullness or be running the WRED algorithm
+ * at the time a new packet attempts to be appended to the end of some
+ * input queue.
+ *
+ * The WRED profile associated with an tm_node considers the entire set of
+ * tm_queues feeding directly or indirectly into it as its measure of
+ * queue fullness.
+ *
+ * <H3>tm_queues</H3>
+ * tm_queues are the second major type of "entity"/object that a TM
+ * system is composed of.  All packets MUST first enter the TM system via
+ * some tm_queue.  Then logically, the head packets of all of the tm_queues
+ * are examined simultaneously by the entire TM system, and ONE tm_queue is
+ * chosen send its head packet out of the TM system's egress.  Abstractly
+ * packets stay in the tm_queue until they are chosen at which time they are
+ * instantly transferred from tm_queue to/through the corresponding TM egress.
+ * It is also important to note that packets in the same tm_queue MUST always
+ * stay in order.  In other words, the second packet in an tm_queue must never
+ * leave the TM system through a TM egress spigot before the first packet has
+ * left the system.  So tm_queue packet order must always be maintained.
+ *
+ * <H3>TM egress</H3>
+ * Note that TM egress objects are NOT referred to as queues, because in
+ * many/most cases they don't have multi-packet structure but instead are
+ * viewed as a port/spigot through which the TM system schedules and finally
+ * transfers input packets through.
+ *
+ * <H2>Ideal versus Actual Behavior</H2>
+ * It is important to recognize the difference between the "abstract"
+ * mathematical model of the prescribed behavior and real implementations.
+ * The model describes the Ideal, but theoretically desired behavior, but such
+ * an Ideal is generally not practical to implement.  Instead, one understands
+ * that virtually all Real TM systems attempt to approximate the Ideal behavior
+ * as given by the TM configuration as best as they can - while still
+ * attaining high packet processing performance.  The idea is that instead of
+ * trying too hard to be "perfect" at the granularity of say microseconds, it
+ * may be better to instead try to match the long term Ideal behavior over a
+ * much more reasonable period of time like a millisecond.  It is generally
+ * better to have a stable implementation that when averaged over a period of
+ * several milliseconds matches the Ideal behavior very closely than to have
+ * an implementation that is perhaps more accurate over a period of
+ * microseconds, but whose millisecond averaged behavior drifts away from the
+ * Ideal case.
+ *
+ * <H2>Other TM Concepts</H2>
+ *
+ * <H3>Profiles</H3>
+ * This specification often packages related TM system parameters into
+ * records/objects called profiles.  These profiles can then be associated with
+ * various entities like tm_nodes and tm_queue's.  This way the amount of
+ * storage associated with setting related parameters can be reduced and
+ * in addition it is common to re-use the same set of parameter set over
+ * and over again, and also to be able to change the parameter set once
+ * and have it affect lots of entities with which it is associated with/applied
+ * to.
+ *
+ * <H3>Absolute Limits versus odp_tm_capability_t</H3>
+ * This header file defines some constants representing the absolute maximum
+ * settings for any TM system, though in most cases a TM system can (and
+ * should) be created/instantiated with smaller values, since lower values
+ * will often result in faster operation and/or less memory used.
+ */
+
+/**
+ * @def ODP_TM_MAX_NUM_SYSTEMS
+ * The maximum number of TM systems that may be created.  On some platforms
+ * this might be much more limited to as little as one hardware TM system.
+ */
+
+/**
+ * @def ODP_TM_MAX_PRIORITIES
+ * The largest range of priorities that any TM system can support.  All strict
+ * priority values MUST in the range 0..ODP_TM_MAX_PRIORITIES-1.
+ */
+
+/**
+ * @def ODP_TM_MAX_LEVELS
+ * The largest range of tm_node levels that any TM system can support.  Hence
+ * all tm_node level values MUST be in the range 0..ODP_TM_MAX_LEVELS-1.
+ * Smaller tm_node levels are associated with tm_nodes closer to the TM system
+ * egress.
+ */
+
+/**
+ * @def ODP_TM_MIN_SCHED_WEIGHT
+ * The smallest SCHED weight is 1 (i.e. 0 is not a legal WFQ/WRR value).
+ */
+
+/**
+ * @def ODP_TM_MAX_SCHED_WEIGHT
+ * The largest weight any TM system can support (at least from a configuration
+ * standpoint).  A given TM system could have a smaller value.
+ */
+
+/**
+ * @def ODP_TM_MAX_TM_QUEUES
+ * The largest number of tm_queues that can handled by any one TM system.
+ */
+
+/**
+ * @def ODP_TM_MAX_NUM_OUTPUTS
+ * The largest number of outputs that can be configured for any one TM system.
+ */
+
+/**
+ * @def ODP_TM_MAX_NUM_TM_NODES
+ * The largest number of tm_nodes that can be in existence for any one TM
+ * system.
+ */
+
+/**
+ * @def ODP_TM_MAX_TM_NODE_FANIN
+ * The largest number of fan-in "inputs" that can be simultaneously connected
+ * to a single tm_node.
+ * @todo Does this need to be as large as ODP_TM_MAX_TM_QUEUES?
+ */
+
+/**
+ * @def ODP_TM_MIN_SHAPER_BW
+ * The largest amound of bandwidth that any shaper's peak or commit rate can
+ * be set to. It is in units of 1000 bytes/second.
+ */
+
+/**
+ * @def ODP_TM_MAX_SHAPER_BW
+ * The largest amound of bandwidth that any shaper's peak or commit rate can
+ * be set to. It is in units of 1000 bytes/second.
+ */
+
+/**
+ * @def ODP_NUM_SHAPER_COLORS
+ * The number of enumeration values defined in the odp_tm_shaper_color_t type.
+ */
+
+/**
+ * @def ODP_TM_INVALID_PRIORITY
+ * Used to indicate an invalid priority value.
+ */
+
+/**
+ * @typedef odp_tm_percent_t
+ * Is used when specifying fields that are percentages.  It is a fixed point
+ * integer whose units are 1/100 of a percent.  Hence 100% is represented as
+ * the integer value 10000.  Note that because it is often used as a ratio of
+ * the current queue value and maximum queue threshold, it can be > 100%, but
+ * in any event will never be larger than 500% (i.e. it MUST be capped at
+ * 50000).
+ */
+
+/**
+ * @typedef odp_tm_t
+ * Each odp_tm_t value represents a specific TM system.  Almost all functions
+ * in this API require a odp_tm_t value - either directly as a function
+ * parameter or indirectly by having another ODP TM handle value as a function
+ * parameter.
+ */
+
+/**
+ * @typedef odp_tm_queue_t
+ * Each odp_tm_queue_t value is an opaque ODP handle representing a specific
+ * tm_queue within a specific TM system.
+ */
+
+/**
+ * @typedef odp_tm_node_t
+ * Each odp_tm_queue_t value is an opaque ODP handle representing a specific
+ * tm node within a specific TM system.
+ */
+
+/**
+ * @typedef odp_tm_shaper_t
+ * Each odp_tm_shaper_t value is an opaque ODP handle representing a specific
+ * shaper profile usable across all TM systems described by this API.  A given
+ * shaper profile can then be attached to any tm_queue or tm_node.
+ */
+
+/**
+ * @typedef odp_tm_sched_t
+ * Each odp_tm_sched_t value is an opaque ODP handle representing a specific
+ * tm_node scheduler profile usable across all TM systems described by this
+ * API.  A given tm_node scheduler profile can then be attached to any
+ * tm_node.
+ */
+
+/**
+ * @typedef odp_tm_threshold_t
+ * Each odp_tm_threshold_t value is an opaque ODP handle representing a
+ * specific queue threshold profile usable across all TM systems described by
+ * this API.  A given queue threshold profile can then be attached to any
+ * tm_queue or tm_node.
+ */
+
+/**
+ * @typedef odp_tm_wred_t
+ * Each odp_tm_wred_t value is an opaque ODP handle representing a specific
+ * WRED profile usable across all TM systems described by this API.  A given
+ * WRED profile can then be attached to any tm_queue or tm_node.
+ */
+
+/**
+ * @def ODP_TM_INVALID
+ * Constant that can be used with any ODP TM handle type and indicates that
+ * this value does NOT represent a valid TM object.
+ */
+
+/**
+ * @def ODP_TM_ROOT
+ * Constant that is used to refer to the egress/root node of the TM subsystem's
+ * tree/hierarchy of nodes.
+ */
+
+/** The odp_tm_capability_t type is used to describe the feature set and limits
+ * of a TM system.  It is passed to the odp_tm_create() function indirectly
+ * by being part of the odp_tm_params_t record.
+ */
+typedef struct {
+       /** max_tm_queues specifies the maximum number of tm_queues that can
+	* be in existence for this TM System.
+	*/
+	uint32_t max_tm_queues;
+
+       /** max_fanin_per_level specifies the maximum number of fan_in link
+	* to any given scheduler (whether weighted or using fair queueing or
+	* round robin) belonging to tm_nodes at the given level.
+	*/
+	uint32_t max_fanin_per_level[ODP_TM_MAX_LEVELS];
+
+       /** max_priority specifies the maximum number of strict priority
+	* levels used by any tm_queue or tm_node.  Note that any given
+	* tm_queue or tm_node can use a subset of these levels.  max_priority
+	* must be in the range 0..ODP_TM_MAX_PRIORITIES - 1.  Note that lower
+	* numeric values represent higher (more important or time critical)
+	* priorities.
+	*/
+	uint8_t max_priority;
+
+       /** max_levels specifies that maximum number of levels of hierarchical
+	* scheduling allowed by this TM System.  This is a count of the
+	* tm_node stages and does not include tm_queues or tm_egress objects.
+	* Hence any given tm_node will have associated tm_node_level in the
+	* range 0 to max_levels - 1, where tm_node's at level 0 output's only
+	* go to egress objects and tm_nodes whose level is max_levels - 1
+	* have their fan_in only from tm_queues.
+	*/
+	uint8_t max_levels;
+
+       /** tm_queue_shaper_supported indicates that the tm_queues support
+	* proper TM shaping.  Note that TM Shaping is NOT the same thing as
+	* Ingress Metering/Policing as specified by RFC 2697 (A Single Rate
+	* Three Color Marker) or RFC 2698 (A Two Rate Three Color Marker).
+	* These RFC's can be used for a Diffserv traffic conditioner, or
+	* other ingress policing.  They make no mention of and have no
+	* algorithms for delaying packets - which is what TM shapers are
+	* expected to do.
+	*/
+	odp_bool_t tm_queue_shaper_supported;
+
+       /** tm_node_shaper_supported indicates that the tm_nodes (at least for
+	* some hierarchical levels) support proper T < M shaping.
+	*/
+	odp_bool_t tm_node_shaper_supported;
+
+       /** red_supported indicates that the tm_queues support some form of
+	* Random Early Discard.
+	*/
+	odp_bool_t red_supported;
+
+       /** hierarchical_red_supported indicates that this TM system supports
+	* some form of RED where the queue fullness of tm_nodes contributes
+	* to the overall RED DROP/NO-DROP decision.
+	*/
+	odp_bool_t hierarchical_red_supported;
+
+       /** weights_supported indicates that the tm_node schedulers (at least
+	* for some hierarchical levels) can have their different weights for
+	* their fan-ins.
+	*/
+	odp_bool_t weights_supported;
+
+       /** fair_queuing_supported indicates the the tm_node schedulers (at
+	* least for some hierarchical levels) can implement WFQ or FQ
+	* scheduling disciplines, otherwise these schedulers can only
+	* implement WRR or RR algorithms,
+	*/
+	odp_bool_t fair_queuing_supported;
+} odp_tm_capability_t;
+
+/** The odp_tm_egress_fcn_t type defines the parameter profile of the egress
+ * function callback.  Using an egress function callback is just one of several
+ * ways of getting packets out from an egress spigot.
+ *
+ */
+typedef void (*odp_tm_egress_fcn_t) (odp_packet_t odp_pkt);
+
+/** The tm_egress_kind_e enumeration type is used to indicate the kind of
+ * egress object ("spigot") associated with this TM system.  Most of these
+ * kinds are optional - with ODP_TM_EGRESS_PKT_IO being the only mandatory
+ * kind.  The TM_EGRESS_FN - if implemented - is useful for testing the TM
+ * subsystem, and users are warned that its performance might be limited.
+ */
+typedef enum {
+	ODP_TM_EGRESS_PKT_IO,
+	ODP_TM_EGRESS_FN,
+} odp_tm_egress_kind_t;
+
+/** The odp_tm_egress_t type is used to describe that type of "egress spigot"
+ * associated with this TM system.  It is passed to the odp_tm_create()
+ * function indirectly by being part of the odp_tm_params_t record.
+ */
+typedef struct {
+	odp_tm_egress_kind_t egress_kind; /**< Union discriminator */
+
+	union {
+		odp_pktio_t pktio;
+		odp_tm_egress_fcn_t egress_fcn;
+	};
+} odp_tm_egress_t;
+
+/** The odp_tm_params_t record type is used to hold extra parameters when
+ * calling the odp_tm_create() function.
+ * Since it is expected that implementations might augment this record type
+ * with platform specific additional fields - it is required that
+ * odp_tm_params_init() be called on variables of this type before any of the
+ * fields are filled in.
+ */
+typedef struct {
+	odp_tm_capability_t capability; /**< capability record */
+	odp_tm_egress_t egress; /**< describes the egress "spigot" */
+} odp_tm_params_t;
+
+/** odp_tm_capability_init() must be called to initialize any
+ * odp_tm_capability_t record before it is first used or assigned to.
+ *
+ * @param[in] capability  A pointer to an odp_tm_capability_t record which
+ *                        is to be initialized.
+ */
+void odp_tm_capability_init(odp_tm_capability_t *capability);
+
+/** odp_tm_params_init() must be called to initialize any
+ * odp_tm_params_t record before it is first used or assigned to.
+ *
+ * @param[in] params  A pointer to an odp_tm_params_t record which
+ *                    is to be initialized.
+ */
+void odp_tm_params_init(odp_tm_params_t *params);
+
+/** Create/instantiate a TM Packet Scheduling system.
+ *
+ * @param[in] name    The name to be assigned to this TM system.  Cannot be
+ *                    NULL, and also must be unique amongst all other TM system
+ *                    names.
+ * @param[in] params  The params to be used when creating this TM system.
+ * @return            Returns ODP_TM_INVALID upon failure, otherwise the newly
+ *                    created TM system's odp_tm_t handle is returned.
+ */
+odp_tm_t odp_tm_create(const char *name, odp_tm_params_t *params);
+
+/** Find a pre-existing TM Packet Scheduling system.  This function can be
+ * used either to find a TM system created previously with odp_tm_create OR
+ * get the odp_tm_t of a built-in TM system - usually based on HW. In this
+ * later case the format of the name used to refer to a specific built-in
+ * hardware TM system may be platform dependent, but in any case a name of
+ * "HW_TM_%u" where the number starts at 1, can be used to find a built-in
+ * system independently of the best capability match.  If name is NULL then
+ * the existing (built-in or created by odp_tm_create) TM system that best
+ * matches capability is returned.
+ *
+ * @param[in] name        If NULL then only uses the capability parameter to
+ *                        find a closest match, otherwise if the name is
+ *                        matched by an existing TM system it is returned.
+ * @param[in] capability  Used when the name is NULL (in which
+ *                        case the closest match is returned) or when the
+ *                        name is not-NULL, but doesn't match
+ *                        any existing TM system in which case the
+ *                        capability is used to find the FIRST
+ *                        TM system matching exactly these limits.
+ * @return                If an existing TM system (built-in or previously
+ *                        created via odp_tm_create) is found, its
+ *                        odp_tm_t value is returned, otherwise
+ *                        ODP_TM_INVALID is returned.
+ */
+odp_tm_t odp_tm_find(const char *name, odp_tm_capability_t *capability);
+
+/** odp_tm_capability() can be used to query the actual limits of a given TM
+ * system.  This function can be used for both built-in TM systems AND TM
+ * system's created via odp_tm_create().
+ *
+ * @param[in]  odp_tm      The odp_tm_t value of the TM system to be
+ *                         queried.
+ * @param[out] capability  A pointer to a odp_tm_capability_t record
+ *                         where the actual limits used by the TM system are
+ *                         copied into.  Note that these limits do NOT
+ *                         have to match the capability passed in if
+ *                         a TM system was created by odp_tm_create,
+ *                         but of course these limits in some cases could
+ *                         be larger.
+ * @return                 Returns 0 upon success, < 0 upon failure (which
+ *                         indicates that the odp_tm value did not
+ *                         exist).
+ */
+int odp_tm_capability(odp_tm_t odp_tm, odp_tm_capability_t *capability);
+
+/** odp_tm_destroy() may be used to destroy TM systems created via
+ * odp_tm_create().  It generally CANNOT be used to destroy built-in TM
+ * systems.  Also some platforms MAY not support destroying of TM systems
+ * created via odp_tm_create() under certain conditions.  For example a given
+ * platform may require that the TM system be first "drained" of all of its
+ * queued packets before it will accept a odp_tm_destroy() call.
+ *
+ * In general calling odp_tm_destroy() on an active TM system does not
+ * guarantee anything about the disposition of any packets queued within the
+ * TM system, other than EVENTUALLY these packets will be either sent (in ANY
+ * order) or freed.
+ *
+ * @param[in] odp_tm  The odp_tm_t value of the TM system to be destroyed (and
+ *                    hence destroyed (and hence freed).
+ * @return            0 upon success, < 0 upon failure.
+ */
+int odp_tm_destroy(odp_tm_t odp_tm);
+
+/** Shaper profile types and functions */
+
+/** Possible values of running the shaper algorithm.  ODP_TM_SHAPER_GREEN
+ * means that the traffic is within the commit specification (rate and burst
+ * size), ODP_TM_SHAPER_YELLOW means that the traffic is within the peak
+ * specification (rate and burst size) and ODP_TM_SHAPER_RED means that the
+ * traffic is exceeding both its commit and peak specifications.  Note that
+ * packets can also have an assigned <b> packet color</b> of ODP_PACKET_GREEN,
+ * ODP_PACKET_YELLOW or ODP_PACKET_RED which has a different meaning and
+ * purpose than the shaper colors.
+ */
+typedef enum {
+	ODP_TM_SHAPER_GREEN, ODP_TM_SHAPER_YELLOW, ODP_TM_SHAPER_RED
+} odp_tm_shaper_color_t;
+
+/** The odp_tm_shaper_params_t record type is used to supply the parameters
+ * associated with a shaper profile.  Since it is expected that
+ * implementations might augment this record type with platform specific
+ * additional fields - it is required that odp_tm_shaper_params_init() be
+ * called on variables of this type before any of the fields are filled in.
+ */
+typedef struct {
+       /** The committed information rate for this shaper profile.  The units
+	* for this integer are always in bits per second.
+	*/
+	uint64_t commit_bps;
+
+       /** The peak information rate for this shaper profile.  The units for
+	* this integer are always in bits per second.
+	*/
+	uint64_t peak_bps;
+
+       /** The commit burst tolerance for this shaper profile.  The units for
+	* this field are always bits.  This value sets an upper limit for the
+	* size of the commitCnt.
+	*/
+	uint32_t commit_burst;
+
+       /** The peak burst tolerance for this shaper profile.  The units for
+	* this field are always bits.  This value sets an upper limit for the
+	* size of the peakCnt.
+	*/
+	uint32_t peak_burst;
+
+       /** The shaper_len_adjust is a value between -128 and 127 which is
+	* directly added to the frame_len of a packet associated with this
+	* profile.  The frame_len would normally include the outermost
+	* Ethernet header (DA, SA, ...) through to the outermost Ethernet CRC
+	* inclusive.  Hence this field - when non-zero - will usually be set
+	* to a value approximating the "time" (in units of bytes) taken by
+	* the Ethernet preamble and Inter Frame Gap.  Traditionally this
+	* would be the value 20 (8 + 12), but in same cases can be as low as
+	* 9 (4 + 5).
+	*/
+	int8_t shaper_len_adjust;
+
+       /** If dual_rate is TRUE it indicates the desire for the
+	* implementation to use dual rate shaping for packets associated with
+	* this profile.  The precise semantics of dual rate shaping are
+	* implementation specific, but in any case require a non-zero set of
+	* both commit and peak parameters.
+	*/
+	odp_bool_t dual_rate;
+} odp_tm_shaper_params_t;
+
+/** odp_tm_shaper_params_init() must be called to initialize any
+ * odp_tm_shaper_params_t record before it is first used or assigned to.
+ *
+ * @param[in] params  A pointer to an odp_tm_shaper_params_t record which
+ *                    is to be initialized.
+ */
+void odp_tm_shaper_params_init(odp_tm_shaper_params_t *params);
+
+/** odp_tm_shaper_create() creates a shaper profile object, which can
+ * subsequently be attached to any number (including zero) of tm_queues
+ * or tm_nodes.
+ *
+ * @param[in] name    Optional name associated with this shaper profile.  Can
+ *                    be NULL.  If non-NULL must be unique amongst the set of
+ *                    all other shaper profiles.
+ * @param[in] params  The profile parameters.  See comments associated with
+ *                    the odp_tm_shaper_params_t for more details.
+ * @return            Returns ODP_TM_INVALID upon failure, or the newly
+ *                    allocated odp_tm_shaper_t value representing this
+ *                    profile object.
+ */
+odp_tm_shaper_t odp_tm_shaper_create(const char *name,
+				     odp_tm_shaper_params_t *params);
+
+/** odp_tm_shaper_params_read() "gets" the current set of values associated
+ * with the specified shaper profile object, and copies them into the supplied
+ * record.
+ *
+ * @param[in]  shaper_profile  Specifies the shaper profile object whose
+ *                             values are to be read.
+ * @param[out] params          A pointer to an odp_tm_shaper_params_t record
+ *                             where the current shaper profile object values
+ *                             are copied to.
+ * @return                     Returns < 0 upon failure or 0 upon success.
+ */
+int odp_tm_shaper_params_read(odp_tm_shaper_t shaper_profile,
+			      odp_tm_shaper_params_t *params);
+
+/** odp_tm_shaper_params_update() "sets" the current set of values associated
+ * with the specified shaper profile object.  In addition, this call has the
+ * effect that all tm_input's and tm_nodes that are associated (attached?)
+ * with this shaper profile object will be updated with the new values.
+ *
+ * @param[in] shaper_profile  Specifies the shaper profile object whose
+ *                            values are to be set.
+ * @param[in] params          A pointer to an odp_tm_shaper_params_t record
+ *                            where the new shaper profile object values
+ *                            are taken from.
+ * @return                    Returns < 0 upon failure or 0 upon success.
+ */
+int odp_tm_shaper_params_update(odp_tm_shaper_t shaper_profile,
+				odp_tm_shaper_params_t *params);
+
+/** odp_tm_shaper_lookup() can be used to find the shaper profile object
+ * created with the specified name.
+ *
+ * @param[in] name  Name of a previously created shaper profile.  Cannot be
+ *                  NULL.
+ * @return          Returns ODP_TM_INVALID upon failure, or the shaper
+ *                  profile handle created with this name.
+ */
+odp_tm_shaper_t odp_tm_shaper_lookup(const char *name);
+
+/** Scheduler Profiles - types and functions */
+
+/** The odp_tm_sched_mode_t type is used to control whether a tm_node
+ * scheduler takes into account packet lengths (by setting the sched_mode to
+ * ODP_TM_BYTE_BASED_WEIGHTS) or instead treat packets with different lengths
+ * the same (by setting the sched_mode to ODP_TM_FRAME_BASED_WEIGHTS).
+ * Normally the sched_mode will be set to ODP_TM_BYTE_BASED_WEIGHTS, otherwise
+ * the scheduler becomes a weighted round robin scheduler.
+ */
+typedef enum {
+	ODP_TM_BYTE_BASED_WEIGHTS, /**< Use the packet length in
+				      scheduler algorithm */
+	ODP_TM_FRAME_BASED_WEIGHTS /**< Ignore the packet length */
+} odp_tm_sched_mode_t;
+
+/** The odp_tm_sched_params_t record type is used to supply the parameters
+ * associated with a scheduler profile.  Since it is expected that
+ * implementations might augment this record type with platform specific
+ * additional fields - it is required that odp_tm_sched_params_init() be
+ * called on variables of this type before any of the fields are filled in.
+ */
+typedef struct {
+       /** sched_modes indicates whether weighted scheduling should be used
+	* or not - on a priority basis.
+	*/
+	odp_tm_sched_mode_t sched_modes[ODP_TM_MAX_PRIORITIES];
+
+       /** In the case that sched_modes for a given strict priority level
+	* indicates the use of weighted scheduling, this field supplies the
+	* weighting factors.  The weights - when defined - are used such that
+	* the (adjusted) frame lengths are divided by these 8-bit weights
+	* (i.e. they are divisors and not multipliers).  Consequently a
+	* weight of 0 (when sched_mode is ODP_TM_BYTE_BASED_WEIGHTS) is
+	* illegal.
+	*/
+	uint8_t sched_weights[ODP_TM_MAX_PRIORITIES];
+} odp_tm_sched_params_t;
+
+/** odp_tm_sched_params_init() must be called to initialize any
+ * odp_tm_sched_params_t record before it is first used or assigned to.
+ *
+ * @param[in] params  A pointer to an odp_tm_sched_params_t record which
+ *                    is to be initialized.
+ */
+void odp_tm_sched_params_init(odp_tm_sched_params_t *params);
+
+/** odp_tm_sched_create() creates a scheduler profile object, which can
+ * subsequently be attached to any number (including zero) of tm_nodes.
+ *
+ * @param[in] name    Optional name associated with this scheduler profile.
+ *                    Can be NULL.  If non-NULL must be unique amongst the
+ *                    set of all other scheduler profiles.
+ * @param[in] params  The profile parameters.  See comments associated with
+ *                    the odp_tm_sched_params_t for more details.
+ * @return            Returns ODP_TM_INVALID upon failure, or the newly
+ *                    allocated odp_tm_sched_t value representing this profile
+ *                    object.
+ */
+odp_tm_sched_t odp_tm_sched_create(const char *name,
+				   odp_tm_sched_params_t *params);
+
+/** odp_tm_sched_params_read() "gets" the current set of values associated
+ * with the specified scheduler profile object, and copies them into the
+ * supplied record.
+ *
+ * @param[in]  sched_profile  Specifies the scheduler profile whose values
+ *                            are to be read.
+ * @param[out] params         A pointer to an odp_tm_sched_params_t record
+ *                            where the current scheduler profile object
+ *                            values are copied to.
+ * @return                    Returns < 0 upon failure or 0 upon success.
+ */
+int odp_tm_sched_params_read(odp_tm_sched_t sched_profile,
+			     odp_tm_sched_params_t *params);
+
+/** odp_tm_sched_params_update() "sets" the current set of values associated
+ * with the specified scheduler profile object.  In addition, this call has
+ * the effect that all tm_nodes that are associated (attached?) with this
+ * Scheduler profile object will be updated with the new values.
+ *
+ * @param[in] sched_profile   Specifies the Scheduler profile object whose
+ *                            values are to be set.
+ * @param[in] params          A pointer to an odp_tm_sched_params_t record
+ *                            where the new scheduler profile object values
+ *                            are taken from.
+ * @return                    Returns < 0 upon failure or 0 upon success.
+ */
+int odp_tm_sched_params_update(odp_tm_sched_t sched_profile,
+			       odp_tm_sched_params_t *params);
+
+/** odp_tm_sched_lookup() can be used to find the scheduler profile object
+ * created with the specified name.
+ *
+ * @param[in] name  Name of a previously created scheduler profile.  Cannot be
+ *                  NULL.
+ * @return          Returns ODP_TM_INVALID upon failure, or the scheduler
+ *                  profile handle created with this name.
+ */
+odp_tm_sched_t odp_tm_sched_lookup(const char *name);
+
+/** Queue Threshold Profiles - types and functions */
+
+/** The odp_tm_threshold_params_t record type is used to supply the parameters
+ * associated with a queue thresholds profile.  Since it is expected that
+ * implementations might augment this record type with platform specific
+ * additional fields - it is required that odp_tm_threshold_params_init() be
+ * called on variables of this type before any of the fields are filled in
+ */
+typedef struct {
+	uint64_t max_pkts; /**<  max pkt cnt for this threshold profile */
+	uint64_t max_bytes; /**<  max byte cnt for this threshold profile */
+	odp_bool_t enable_max_pkts; /**<  TRUE if max_pkts is valid */
+	odp_bool_t enable_max_bytes; /**<  TRUE if max_bytes is valid */
+} odp_tm_threshold_params_t;
+
+/** odp_tm_threshold_params_init() must be called to initialize any
+ * odp_tm_threshold_params_t record before it is first used or assigned to.
+ *
+ * @param[in] params  A pointer to an odp_tm_threshold_params_t record which
+ *                    is to be initialized.
+ */
+void odp_tm_threshold_params_init(odp_tm_threshold_params_t *params);
+
+/** odp_tm_threshold_create() creates a queue threshold profile object, which
+ * can subsequently be attached to any number (including zero) of tm_queues or
+ * tm_nodes.
+ *
+ * @param[in] name    Optional name associated with this queue threshold
+ *                    profile.  Can be NULL.  If non-NULL must be unique
+ *                    amongst the set of all other queue threshold profiles.
+ * @param[in] params  The profile parameters.  See comments associated with
+ *                    the odp_tm_threshold_params_t for more details.
+ * @return            Returns ODP_TM_INVALID upon failure, or the newly
+ *                    allocated odp_tm_threshold_t value representing this
+ *                    profile object.
+ */
+odp_tm_threshold_t odp_tm_threshold_create(const char *name,
+					   odp_tm_threshold_params_t *params);
+
+/** odp_tm_thresholds_params_read() "gets" the current set of values associated
+ * with the specified queue thresholds profile object, and copies them into the
+ * supplied record.
+ *
+ * @param[in]  threshold_profile  Specifies the queue thresholds profile
+ *                                object whose values are to be read.
+ * @param[out] params             A pointer to an odp_tm_threshold_params_t
+ *                                record where the current queue thresholds
+ *                                profile object values are copied to.
+ * @return                        Returns < 0 upon failure or 0 upon success.
+ */
+int odp_tm_thresholds_params_read(odp_tm_threshold_t threshold_profile,
+				  odp_tm_threshold_params_t *params);
+
+/** odp_tm_thresholds_params_update() "sets" the current set of values
+ * associated with the specified queue thresholds profile object.  In addition,
+ * this call has the effect that all tm_input's and tm_nodes that are
+ * associated (attached?) with this queue thresholds profile object will be
+ * updated with the new values.
+ *
+ * @param[in] threshold_profile  Specifies the queue thresholds profile
+ *                               object whose values are to be set.
+ * @param[in] params             A pointer to an odp_tm_threshold_params_t
+ *                               record where the current queue thresholds
+ *                               profile object values are taken from.
+ * @return                       Returns < 0 upon failure or 0 upon success.
+ */
+int odp_tm_thresholds_params_update(odp_tm_threshold_t threshold_profile,
+				    odp_tm_threshold_params_t *params);
+
+/** odp_tm_thresholds_lookup() can be used to find the queue thresholds
+ * profile object created with the specified name.
+ *
+ * @param[in] name  Name of a previously created queue thresholds profile.
+ *                  Cannot be NULL.
+ * @return          Returns ODP_TM_INVALID upon failure, or the queue
+ *                  thresholds profile handle created with this name.
+ */
+odp_tm_threshold_t odp_tm_thresholds_lookup(const char *name);
+
+/** WRED Profiles - types and functions */
+
+/** The odp_tm_wred_params_t record type is used to supply the parameters
+ * associated with a Random Early Discard profile.  Since it is expected that
+ * implementations might augment this record type with platform specific
+ * additional fields - it is required that odp_tm_wred_params_init() be called
+ * on variables of this type before any of the fields are filled in.
+ */
+typedef struct {
+       /** When min_threshold is set to zero then single-slope WRED is
+	* enabled, as described in the description of med_threshold.
+	* Otherwise dual-slope WRED is enabled whereby the behavior depends
+	* on which of the following three cases exists:
+	* <ol> <li> queue
+	* fullness < min_threshold.  In this case the drop probability is
+	* zero.
+	* <li> min_threshold <= queue fullness < med_threshold.  In
+	* this case the drop probability increases linearly from zero until
+	* it reaches med_drop_prob at a queue fullness equal to
+	* med_threshold.
+	* <li> med_threshold <= queue fullness.  In this case
+	* the drop probability increases linearly from med_drop_prob when the
+	* queue fullness equals med_threshold until it reaches 100% with a
+	* drop probability of max_drop_prob.  </ol>
+	*/
+	odp_tm_percent_t min_threshold;
+
+       /** The meaning of med_threshold depends upon whether single-slope or
+	* dual-slope WRED is being used or not.  When min_threshold is 0 then
+	* single-slope WRED is enabled in which case the med_threshold value
+	* represents (as a percentage of max queue fullness) the point at
+	* which the drop probability starts increasing linearly from 0 until
+	* it becomes equal to max_drop_prob when the queue fullness reaches
+	* 100%.  See min_threshold comments for the case of dual-slope WRED.
+	*/
+	odp_tm_percent_t med_threshold;
+
+       /** The med_drop_prob is only used when dual-slope WRED is being used,
+	* in which case med_drop_prob MUST be < max_drop_prob.  See
+	* min_threshold comments for more details.
+	*/
+	odp_tm_percent_t med_drop_prob;
+
+       /** The max_drop_prob equals the drop probability when the queue
+	* fullness almost equals 100%.  Of course once the queue fullness is
+	* >= 100% of the max queue fullness, the drop probability
+	* discontinuously becomes 100%.
+	*/
+	odp_tm_percent_t max_drop_prob;
+
+       /** When enable_wred is false, all tm_queues and tm_nodes that are
+	* attached to this profile will not take part in a Random Early
+	* Discard algorithm.
+	*/
+	odp_bool_t enable_wred;
+
+       /** When use_byte_fullness is true then WRED will use queue memory
+	* usage as the fullness criterion, otherwise when use_byte_fullness
+	* is false, WRED will use the queue length (i.e. the number of
+	* packets in the queue) as the fullness criterion.  Often will be set
+	* to true for WRED profiles applied to tm_queues and set to false for
+	* WRED profiles applied to tm_nodes.
+	*/
+	odp_bool_t use_byte_fullness;
+} odp_tm_wred_params_t;
+
+/** odp_tm_wred_params_init() must be called to initialize any
+ * odp_tm_wred_params_t record before it is first used or assigned to.
+ *
+ * @param[in] params  A pointer to an odp_tm_wred_params_t record which
+ *                    is to be initialized.
+ */
+void odp_tm_wred_params_init(odp_tm_wred_params_t *params);
+
+/** odp_tm_wred_create() creates a WRED (Weighted Random Early Discard)
+ * profile object, which can subsequently be attached to any number (including
+ * zero) of tm_queues or tm_nodes.
+ *
+ * @param[in] name    Optional name associated with this WRED profile.  Can
+ *                    be NULL.  If non-NULL must be unique amongst the set of
+ *                    all other WRED profiles.
+ * @param[in] params  The profile parameters.  See comments associated with the
+ *                    odp_tm_wred_params_t for more details.
+ * @return            Returns ODP_TM_INVALID upon failure, or the newly
+ *                    allocated odp_tm_wred_t value representing this profile
+ *                    object.
+ */
+odp_tm_wred_t odp_tm_wred_create(const char *name,
+				 odp_tm_wred_params_t *params);
+
+/** odp_tm_wred_params_read() "gets" the current set of values associated
+ * with the specified WRED profile object, and copies them into the supplied
+ * record.
+ *
+ * @param[in]  wred_profile  Specifies the WRED profile object whose
+ *                           values are to be read.
+ * @param[out] params        A pointer to an odp_tm_wred_params_t record
+ *                           where the current WRED profile object values
+ *                           are copied to.
+ * @return                   Returns < 0 upon failure or 0 upon success.
+ */
+int odp_tm_wred_params_read(odp_tm_wred_t wred_profile,
+			    odp_tm_wred_params_t *params);
+
+/** odp_tm_wred_params_update() "sets" the current set of values associated
+ * with the specified WRED profile object.  In addition, this call has the
+ * effect that all tm_input's and tm_nodes that are associated (attached?)
+ * with this WRED profile object will be updated with the new values.
+ *
+ * @param[in] wred_profile  Specifies the WRED profile object whose
+ *                          values are to be set.
+ * @param[in] params        A pointer to an odp_tm_wred_params_t record
+ *                          where the new WRED profile object values
+ *                          are taken from.
+ * @return                  Returns < 0 upon failure or 0 upon success.
+ */
+int odp_tm_wred_params_update(odp_tm_wred_t wred_profile,
+			      odp_tm_wred_params_t *params);
+
+/** odp_tm_wred_lookup() can be used to find the WRED profile object created
+ * with the specified name.
+ *
+ * @param[in] name  Name of a previously created WRED profile.  Cannot be
+ *                  NULL.
+ * @return          Returns ODP_TM_INVALID upon failure, or the WRED
+ *                  profile handle created with this name.
+ */
+odp_tm_wred_t odp_tm_wred_lookup(const char *name);
+
+/** The odp_tm_node_params_t record type is used to hold extra parameters when
+ * calling the odp_tm_node_create() function.  Many of these fields are
+ * optional EXCEPT for max_fanin and level.  Also since it is expected that
+ * implementations might augment this record type with platform specific
+ * additional fields - it is required that odp_tm_node_params_init() be called
+ * on variables of this type before any of the fields are filled in.
+ */
+typedef struct {
+       /** The max_fan_in sets tha maximum number of src tm_queues and
+	* producer tm_nodes that can be simultaneously be connected to this
+	* tm_node as their destination.
+	*/
+	uint32_t max_fanin;
+
+       /**> @todo uint8_t num_priorities; ? */
+
+       /** The shaper profile to be associated with this tm_node.  Can be
+	* ODP_TM_INVALID and can also be set and changed post-creation via
+	* odp_tm_node_shaper_config();
+	*/
+	odp_tm_shaper_t shaper_profile;
+
+       /** The threshold profile to be used in setting the max queue fullness
+	* for WRED and/or tail drop?  Can be ODP_TM_INVALID and can also be
+	* set and changed post-creation via odp_tm_node_threshold_config().
+	*/
+	odp_tm_threshold_t threshold_profile;
+
+       /** The WRED profile(s) to be associated with this tm_node.  Any or
+	* all array elements can be ODP_TM_INVALID and can also be set and
+	* changed post-creation via odp_tm_node_wred_config().
+	*/
+	odp_tm_wred_t wred_profile[ODP_NUM_PACKET_COLORS];
+
+       /** The level (or tm_node stage) sets the level for this tm_node It
+	* must be in range 0..max_levels-1.  Note that the tm_node topology
+	* is constrained such that only tm_node outputs with numerically
+	* greater levels may be connected to the fan-in of tm_node's with
+	* numerically smaller levels.
+	*/
+	uint8_t level;
+} odp_tm_node_params_t;
+
+/** odp_tm_node_params_init() must be called to initialize any
+ * odp_tm_node_params_t record before it is first used or assigned to.
+ *
+ * @param[in] params  A pointer to an odp_tm_node_params_t record which
+ *                    is to be initialized.
+ */
+void odp_tm_node_params_init(odp_tm_node_params_t *params);
+
+/** Create an tm_node with a specific set of implemented strict priority
+ * levels as given by the priorities array parameter.  The set of priority
+ * levels does not have to "contiguous", but the "priorities" values for all
+ * indexes > max_priority MUST be FALSE.  Note that the set of implemented
+ * strict priority levels for an tm_node cannot be changed after tm_node
+ * creation.  The level parameter MUST be in the range 0..max_level - 1.
+ *
+ * @param[in] odp_tm  Odp_tm is used to identify the TM system into which this
+ *                    odp_tm_node object is created.
+ * @param[in] name    Optional name that can be used later later to find this
+ *                    same odp_tm_node_t.  Can be NULL, otherwise must be
+ *                    unique across all odp_tm_node objects.
+ * @param[in] params  A pointer to a record holding (an extensible) set of
+ *                    properties/attributes of this tm_node.
+ * @return            Returns ODP_TM_INVALID upon failure, otherwise returns
+ *                    a valid odp_tm_node_t handleif successful.
+ */
+odp_tm_node_t odp_tm_node_create(odp_tm_t odp_tm, const char *name,
+				 odp_tm_node_params_t *params);
+
+/** The odp_tm_node_shaper_config() function is used to dynamically set or
+ * change the shaper profile associated with this tm_node.
+ *
+ * @param[in] tm_node         Specifies the tm_node to be changed.
+ * @param[in] shaper_profile  Specifies the shaper profile that should
+ *                            now be used for the shaper entity within the
+ *                            given tm_node.  Note that it is legal to specify
+ *                            ODP_TM_INVALID indicating that this tm_node
+ *                            no longer implements a shaper function.
+ * @return                    Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_node_shaper_config(odp_tm_node_t tm_node,
+			      odp_tm_shaper_t shaper_profile);
+
+/** The odp_tm_node_sched_config() function is used to dynamically set or
+ * change the scheduler profile associated with a tm_node.
+ *
+ * @param[in] tm_node         Specifies the tm_node to be changed.
+ * @param[in] tm_fan_in_node  Specifies which of the specified tm_node's
+ *                            fan-in's weights etc are to be changed. The
+ *                            fan-in is indentified by the "producer"/parent
+ *                            tm_node actually connected to this fan-in.
+ * @param[in] sched_profile   Specifies the scheduler profile that should
+ *                            now be used for the WFQ/RR entity within the
+ *                            given tm_node.
+ * @return                    Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_node_sched_config(odp_tm_node_t tm_node,
+			     odp_tm_node_t tm_fan_in_node,
+			     odp_tm_sched_t sched_profile);
+
+/** The odp_tm_node_threshold_config() function is used to dynamically set or
+ * change the queue threshold profile associated with this tm_node.
+ *
+ * @param[in] tm_node             Specifies the tm_node to be changed.
+ * @param[in] thresholds_profile  Specifies the queue threshold profile that
+ *                                should now be used for the given tm_node.
+ * @return                        Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_node_threshold_config(odp_tm_node_t tm_node,
+				 odp_tm_threshold_t thresholds_profile);
+
+/** The odp_tm_node_wred_config() function is used to dynamically set or
+ * change the WRED profile associated with this tm_node or tm_node/pkt_color
+ * combination.
+ *
+ * @param[in] tm_node       Specifies the tm_node to be changed.
+ * @param[in] pkt_color     Specifies the pkt_color that this profile is to be
+ *                          used with.  Can also be the special value
+ *                          ALL_PKT_COLORS.
+ * @param[in] wred_profile  Specifies the WRED profile that should now be used
+ *                          by this tm_queue, when processing pkts of this
+ *                          pkt_color.  It can be the value ODP_TM_INVALID
+ *                          indicating that this tm_queue/pkt_color combination
+ *                          no longer implements WRED.
+ * @return                  Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_node_wred_config(odp_tm_node_t tm_node,
+			    odp_packet_color_t pkt_color,
+			    odp_tm_wred_t wred_profile);
+
+/** odp_tm_node_lookup() can be used to find the tm_node object created with
+ * the specified name.
+ *
+ * @param[in] odp_tm  Odp_tm is used to identify the TM system into which this
+ *                    odp_tm_node object is created.
+ * @param[in] name    Name of a previously created tm_node.  Cannot be
+ *                    NULL.
+ * @return            Returns ODP_TM_INVALID upon failure, or the tm_node
+ *                    handle created with this name.
+ */
+odp_tm_node_t odp_tm_node_lookup(odp_tm_t odp_tm, const char *name);
+
+/** The odp_tm_queue_params_t record type is used to hold extra parameters
+ * when calling the odp_tm_queue_create() function.  Many of these fields are
+ * optional EXCEPT for priority.  Also since it is expected that
+ * implementations might augment this record type with platform specific
+ * additional fields - it is required that odp_tm_queue_params_init() be
+ * called on variables of this type before any of the fields are filled in.
+ */
+typedef struct {
+       /** The shaper profile to be associated with this tm_queue.  Can be
+	* ODP_TM_INVALID and can also be set and changed post-creation via
+	* odp_tm_queue_shaper_config();
+	*/
+	odp_tm_shaper_t shaper_profile;
+
+       /** The threshold profile to be used in setting the max queue fullness
+	* for WRED and/or tail drop?  Can be ODP_TM_INVALID and can also be
+	* set and changed post-creation via odp_tm_queue_threshold_config().
+	*/
+	odp_tm_threshold_t threshold_profile;
+
+       /** The WRED profile(s) to be associated with this tm_queue.  Any or
+	* all array elements can be ODP_TM_INVALID and can also be set and
+	* changed post-creation via odp_tm_queue_wred_config().
+	*/
+	odp_tm_wred_t wred_profile[ODP_NUM_PACKET_COLORS];
+
+       /** The strict priority level assigned to packets in this tm_queue -
+	* in other words all packets associated with a given tm_queue MUST
+	* have the same single strict priority level and this level must be
+	* in the range 0..max_priority.
+	*/
+	uint8_t priority;
+} odp_tm_queue_params_t;
+
+/** odp_tm_queue_params_init() must be called to initialize any
+ * odp_tm_queue_params_t record before it is first used or assigned to.
+ *
+ * @param[in] params  A pointer to an odp_tm_queue_params_t record which
+ *                    is to be initialized.
+ */
+void odp_tm_queue_params_init(odp_tm_queue_params_t *params);
+
+/** Create an tm_queue object.  One can specify the maximum queue limits
+ * either as a maximum number of packets in the queue OR as a maximum number
+ * of bytes in the queue, or if both are specified, then whichever limit is
+ * hit first.  Note that in the case of specifying the maximum queue memory
+ * size as bytes, the system is free to instead convert this byte value into a
+ * number of buffers and instead limit the queue memory usage by buffer counts
+ * versus strictly using byte counts.
+ *
+ * @param[in] odp_tm  Odp_tm is used to identify the TM system into which this
+ *                    odp_tm_queue object is created.
+ * @param[in] params  A pointer to a record holding (an extensible) set of
+ *                    properties/attributes of this tm_queue.
+ * @return            Returns ODP_TM_INVALID upon failure, otherwise a valid
+ *                    odp_tm_queue_t handle.
+ */
+odp_tm_queue_t odp_tm_queue_create(odp_tm_t odp_tm,
+				   odp_tm_queue_params_t *params);
+
+/** The odp_tm_queue_shaper_config() function is used to dynamically set
+ * or change the shaper profile associated with this tm_queue.
+ *
+ * @param[in] tm_queue        Specifies the tm_queue to be changed.
+ * @param[in] shaper_profile  Specifies the shaper profile that should now be
+ *                            used for shaping the tm_queue's packet stream.
+ *                            Note that it is legal to specify ODP_TM_INVALID
+ *                            indicating that this tm_queue no longer
+ *                            implements a shaper function.
+ * @return                    Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_queue_shaper_config(odp_tm_queue_t tm_queue,
+			       odp_tm_shaper_t shaper_profile);
+
+/** The odp_tm_queue_sched_config() function is used to dynamically set or
+ * change the scheduler profile associated with a tm_node.  Note that despite
+ * the name, this function affects a tm_node scheduler - specifically the
+ * scheduler fan-in when such fan-in comes from an tm_queue.
+ *
+ * @param[in] tm_node         Specifies the tm_node to be changed.
+ * @param[in] tm_fan_in_queue Specifies which of the specified tm_node's
+ *                            fan-in's weights etc are to be changed. The
+ *                            fan-in is indentified by the "producer"/parent
+ *                            tm_queue actually connected to this fan-in.
+ * @param[in] sched_profile   Specifies the scheduler profile that should
+ *                            now be used for the WFQ/RR entity within the
+ *                            given tm_node.
+ * @return                    Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_queue_sched_config(odp_tm_node_t tm_node,
+			      odp_tm_queue_t tm_fan_in_queue,
+			      odp_tm_sched_t sched_profile);
+
+/** The odp_tm_queue_threshold_config() function is used to dynamically set or
+ * change the queue threshold profile associated with this tm_queue.
+ *
+ * @param[in] tm_queue            Specifies the tm_queue to be changed.
+ * @param[in] thresholds_profile  Specifies the queue threshold profile that
+ *                                should now be used for the given tm_queue.
+ * @return                        Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_queue_threshold_config(odp_tm_queue_t tm_queue,
+				  odp_tm_threshold_t thresholds_profile);
+
+/** odp_tm_queue_wred_config() function is used to dynamically set or change
+ * the WRED profile associated with this tm_queue or tm_queue/pkt_color
+ * combination.
+ *
+ * @param[in] tm_queue      Specifies the tm_queue to be changed.
+ * @param[in] pkt_color     Specifies the pkt_color that this profile is to be
+ *                          used with.  Can also be the special value
+ *                          ALL_PKT_COLORS.
+ * @param[in] wred_profile  Specifies the WRED profile that should now be used
+ *                          by this tm_queue, when processing pkts of this
+ *                          pkt_color.  It can be the value ODP_TM_INVALID
+ *                          indicating that this tm_queue/pkt_color combination
+ *                          no longer implements WRED.
+ * @return                  Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_queue_wred_config(odp_tm_queue_t tm_queue,
+			     odp_packet_color_t pkt_color,
+			     odp_tm_wred_t wred_profile);
+
+/** Topology setting functions */
+
+/** Connects the "output" of the src_tm_node to be a "producer" of the given
+ * dst_tm_node.  Note that an ODP_TM_ROOT handle passed in for the
+ * dst_tm_node implies connection to the egress/root object of this TM system.
+ *
+ * @param[in] src_tm_node  odp_tm_node_t handle of the tm_node whose output is
+ *                         to be connected to the fan-in of the next tm_node
+ *                         as represented by the dst_tm_node.
+ * @param[in] dst_tm_node  odp_tm_node_t handle of the tm_node object that will
+ *                         receive all of the pkt_descs from the src tm_node
+ *                         output.  If ODP_TM_ROOT, then attachment is to
+ *                         the root egress object/spigot.
+ * @return                 0 upon success, < 0 on failure.
+ */
+int odp_tm_node_connect(odp_tm_node_t src_tm_node, odp_tm_node_t dst_tm_node);
+
+/** The odp_queue_connect() function connects the indicated tm_queue to a
+ * parent tm_node or to the egress/root node.  The tm_queue will then become
+ * one of the dst node's fan-in set.
+ *
+ * @param[in] tm_queue     Specifies the tm_queue.
+ * @param[in] dst_tm_node  odp_tm_node_t handle of the tm_node object that will
+ *                         receive all of the pkt_descs from the src tm_node
+ *                         output.  If ODP_TM_ROOT, then attachment is to
+ *                         the root egress object/spigot.
+ * @return                 Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_queue_connect(odp_tm_queue_t tm_queue, odp_tm_node_t dst_tm_node);
+
+/** Input API */
+
+/** The odp_tm_enq() function is used to add packets to a given TM system.
+ * Note that the System Metadata associated with the pkt needed by the TM
+ * system is (a) a drop_eligible bit, (b) a two bit "pkt_color", (c) a 16-bit
+ * pkt_len, and MAYBE? (d) a signed 8-bit shaper_len_adjust.
+ *
+ * If there is a non-zero shaper_len_adjust, then it is added to the pkt_len
+ * after any non-zero shaper_len_adjust that is part of the shaper profile.
+ *
+ * The pkt_color bits are a result of some earlier Metering/Marking/Policing
+ * processing (typically ingress based), and should not be confused with the
+ * shaper_color produced from the TM shaper entities within the tm_inputs and
+ * tm_nodes.
+ *
+ * @param[in] tm_queue  Specifies the tm_queue (and indirectly the TM system).
+ * @param[in] pkt       Handle to a packet.
+ * @return              Returns 0 upon success, < 0 upon failure. One of the
+ *                      more common failure reasons is WRED dropage.
+ */
+int odp_tm_enq(odp_tm_queue_t tm_queue, odp_packet_t pkt);
+
+/** The odp_tm_enq_with_cnt() function behaves identically to odp_tm_enq(),
+ * except that it also returns (an approximation to?) the current tm_queue
+ * packet queue count.
+ *
+ * @param[in] tm_queue  Specifies the tm_queue (and indirectly the TM system).
+ * @param[in] pkt       Handle to a packet.
+ * @return              Returns the number of packets previously enqueued on
+ *                      this tm_queue upon success, < 0 upon failure.
+ */
+int odp_tm_enq_with_cnt(odp_tm_queue_t tm_queue, odp_packet_t pkt);
+
+/** Dynamic state query functions */
+
+/** The following bit mask constants are used to refine the queue query
+ * functions defined below.
+ */
+#define ODP_TM_QUERY_PKT_CNT     0x01   /**<  The total_pkt_cnt value */
+#define ODP_TM_QUERY_BYTE_CNT    0x02   /**<  The total_byte_cnt value */
+#define ODP_TM_QUERY_THRESHOLDS  0x04   /**<  The thresholds??? */
+
+/** The odp_tm_queue_info_t record type is used to return the various counts
+ * as requested by functions like odp_tm_queue_query() and
+ * odp_tm_total_query().
+ */
+typedef struct {
+       /** The total_pkt_cnt field is the total number of packets currently
+	* stored/associated with the requested set of tm_queues.  Note that
+	* because the packet queues are potentially being manipulated by
+	* multiple cpu's, the values here are only accurate when the tm
+	* system is "stopped" (i.e. the egress spigot is stopped and no
+	* odp_tm_enq calls are taking place).  Implementations are free to
+	* batch update these counters - up to a dozen or so packets.
+	*/
+	uint64_t total_pkt_cnt;
+
+       /** If the requested set of tm_queues has an odp_tm_threshold_t
+	* profile associated with it, then this is the max_pkt_cnt set in the
+	* profile params.  Returning this field is a convenience to the ODP
+	* programmer, enabling them to quickly see how the total_pkt_cnt
+	* compares to the maximum packet count threshold.  Note that there is
+	* no requirement that total_pkt_cnt be <= max_pkt_cnt.
+	*/
+	uint64_t max_pkt_cnt;
+
+       /** The total_byte_cnt can either be the actual number of bytes used
+	* or an approximation of the number of bytes used based upon the
+	* number of fixed sized buffers used multiplied by the buffer size.
+	* In both cases the total_byte_cnt should correspond to the same set
+	* of packets that were counted above.  For instance, if the
+	* total_pkt_cnt is updated in a batch, then the total_byte_cnt should
+	* also be updated in the same batch.  The approx_byte_cnt field below
+	* indicates whether the total_byte_cnt is buffer count based or not.
+	* In the case that the number of bytes used by a packet is rounded up
+	* to a 2, 4, 8, or 16 byte boundary, it is recommended that
+	* approx_byte_cnt be false.  It is implementation dependent whether
+	* the byte count of a packet includes the CRC, but it is recommended
+	* that it not include headroom, preamble or IPG.  Of course when the
+	* buffer counting method is used, it is expected that any headroom in
+	* the first buffer is implicitly included.  Finally in the case of
+	* variable length pkt based buffering, instead of taking the
+	* total_pkt_cnt and multiplying it by the maximum ethernet packet
+	* size, it is recommended that byte_cnt_valid be FALSE - even when
+	* query_flags includes ODP_TM_QUERY_BYTE_CNT.
+	*/
+	uint64_t total_byte_cnt;
+
+       /** If the requested set of tm_queues has an odp_tm_threshold_t
+	* profile associated with it, then this is the max_byte_cnt set in
+	* the profile params.  Returning this field is a convenience to the
+	* ODP programmer, enabling them to quickly see how the total_byte_cnt
+	* compares to the maximum byte count threshold.  Note that there is
+	* no requirement that total_byte_cnt be <= max_byte_cnt.
+	*/
+	uint64_t max_byte_cnt;
+
+       /** The following boolean values indicate which of the counts above
+	* are valid.  Invalid count values must be 0.
+	*/
+	odp_bool_t total_pkt_cnt_valid;  /**< TRUE if total_pkt_cnt is valid */
+	odp_bool_t max_pkt_cnt_valid;    /**< TRUE if max_pkt_cnt is valid */
+	odp_bool_t total_byte_cnt_valid; /**< TRUE if total_byte_cnt is valid */
+	odp_bool_t max_byte_cnt_valid;   /**< TRUE if max_byte_cnt is valid */
+
+       /** The approx_byte_cnt is TRUE if the total_byte_cnt field is valid
+	* AND if the buffer counting method is used.
+	*/
+	odp_bool_t approx_byte_cnt;
+} odp_tm_queue_info_t;
+
+/** The odp_tm_queue_query() function can be used to check a single tm_queue's
+ * queue utilization.  The query_flags indicate whether or not packet counts,
+ * byte counts or both are being requested.  It is an error to request
+ * neither.  The implementation may still return both sets of counts
+ * regardless of query_flags if the cost of returning all the counts is
+ * comparable to the cost of checking the query_flags.
+ *
+ * @param[in]  tm_queue     Specifies the tm_queue (and indirectly the
+ *                          TM system).
+ * @param[out] query_flags  A set of flag bits indicating which counters are
+ *                          being requested to be returned in the info record.
+ * @param[out] info         Pointer to an odp_tm_queue_info_t record where the
+ *                          requested queue info is returned.
+ * @return                  Returns 0 upon success, < 0 upon failure.
+ */
+int odp_tm_queue_query(odp_tm_queue_t tm_queue,
+		       uint32_t query_flags,
+		       odp_tm_queue_info_t *info);
+
+/** The odp_tm_priority_query() function can be used to check the queue
+ * utilization of all tm_queue's with the given priority.  The query_flags
+ * indicate whether or not packet counts, byte counts or both are being
+ * requested.  It is an error to request neither.  The implementation may
+ * still return both sets of counts regardless of query_flags if the cost of
+ * returning all the counts is comparable to the cost of checking the
+ * query_flags.
+ *
+ * @param[in]  odp_tm       Specifies the TM system.
+ * @param[in]  priority     Supplies the strict priority level used to specify
+ *                          which tm_queues are included in the info values.
+ * @param[out] query_flags  A set of flag bits indicating which counters are
+ *                          being requested to be returned in the info record.
+ * @param[out] info         Pointer to an odp_tm_queue_info_t record where the
+ *                          requested queue info is returned.
+ * @return                  Returns 0 upon success, < 0 upon failure.
+ */
+int odp_tm_priority_query(odp_tm_t odp_tm, uint8_t priority,
+			  uint32_t query_flags, odp_tm_queue_info_t *info);
+
+/** The odp_tm_total_query() function can be used to check the queue
+ * utilization of all tm_queue's in a single TM system.  The query_flags
+ * indicate whether or not packet counts, byte counts or both are being
+ * requested.  It is an error to request neither.  The implementation may
+ * still return both sets of counts regardless of query_flags if the cost of
+ * returning all the counts is comparable to the cost of checking the
+ * query_flags.
+ *
+ * @param[in]  odp_tm       Specifies the TM system.
+ * @param[out] query_flags  A set of flag bits indicating which counters are
+ *                          being requested to be returned in the info record.
+ * @param[out] info         Pointer to an odp_tm_queue_info_t record where the
+ *                          requested queue info is returned.
+ * @return                  Returns 0 upon success, < 0 upon failure.
+ */
+int odp_tm_total_query(odp_tm_t odp_tm, uint32_t query_flags,
+		       odp_tm_queue_info_t *info);
+
+/** The odp_tm_priority_threshold_config() function is only used to associate
+ * a maximum packet count and/or a maximum byte count with a strict priority
+ * level - for the benefit of the odp_tm_priority_query() function.  It has no
+ * semantic effects other than returning these queue threshold values in the
+ * odp_tm_queue_info_t record.
+ *
+ * @param[in] odp_tm              Specifies the TM system.
+ * @param[in] priority            Supplies the strict priority level that
+ *                                the threshold profile params are associated
+ *                                with.
+ * @param[in] thresholds_profile  Specifies the queue threshold profile that
+ *                                should now be associated with the supplied
+ *                                strict priority level.
+ * @return                        Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_priority_threshold_config(odp_tm_t odp_tm, uint8_t priority,
+				     odp_tm_threshold_t thresholds_profile);
+
+/** The odp_tm_total_threshold_config() function is only used to associate a
+ * maximum packet count and/or a maximum byte count with a TM system - for the
+ * benefit of the odp_tm_total_query() function.  It has no semantic effects
+ * other than returning these queue threshold values in the
+ * odp_tm_queue_info_t record.
+ *
+ * @param[in] odp_tm              Specifies the TM system.
+ * @param[in] thresholds_profile  Specifies the queue threshold profile that
+ *                                should now be used for the entire TM
+ *                                system.
+ * @return                        Returns 0 upon success and < 0 upon failure.
+ */
+int odp_tm_total_threshold_config(odp_tm_t odp_tm,
+				  odp_tm_threshold_t thresholds_profile);
+
+/** Misc functions */
+
+/** The odp_tm_periodic_update function is a placeholder for any external
+ * source of periodic events.  In some cases the TM system may already have an
+ * internal built-in source of periodic events - in which case calling this
+ * function has no effect.
+ */
+void odp_tm_periodic_update(void);
+
+/** The odp_tm_is_idle function is used to determine if the specified ODP
+ * traffic management system still has "work" to do (i.e. has at least one
+ * non-empty tm_queue and perhaps some outstanding timers etc).  This function
+ * can be used by test programs and ODP applications that wish to know when
+ * TM system has completed its work - presumably after they have stopped
+ * sending in new pkts.  Note that this function should not be called often
+ * since for some implementations this call could take a fairly long time
+ * to execute!
+ *
+ * @param[in] odp_tm  Specifies the TM system.
+ * @return            Returns 1 if the TM system is idle and 0 otherwise.
+ */
+odp_bool_t odp_tm_is_idle(odp_tm_t odp_tm);
+
+/** The odp_tm_stats_print function is used to write implementation-defined
+ * information about the specified TM system to the ODP log. The intended use
+ * is for debugging.
+ *
+ * @param[in] odp_tm  Specifies the TM system.
+ */
+void odp_tm_stats_print(odp_tm_t odp_tm);
+
+/**
+ * @}
+ */
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/platform/linux-generic/Makefile.am b/platform/linux-generic/Makefile.am
index a6b6029..143167f 100644
--- a/platform/linux-generic/Makefile.am
+++ b/platform/linux-generic/Makefile.am
@@ -51,6 +51,7 @@  odpinclude_HEADERS = \
 		  $(srcdir)/include/odp/ticketlock.h \
 		  $(srcdir)/include/odp/time.h \
 		  $(srcdir)/include/odp/timer.h \
+		  $(srcdir)/include/odp/traffic_mngr.h \
 		  $(srcdir)/include/odp/version.h
 
 odpplatincludedir= $(includedir)/odp/plat
@@ -78,6 +79,7 @@  odpplatinclude_HEADERS = \
 		  $(srcdir)/include/odp/plat/ticketlock_types.h \
 		  $(srcdir)/include/odp/plat/time_types.h \
 		  $(srcdir)/include/odp/plat/timer_types.h \
+		  $(srcdir)/include/odp/plat/traffic_mngr_types.h \
 		  $(srcdir)/include/odp/plat/version_types.h
 
 odpapiincludedir= $(includedir)/odp/api
@@ -122,6 +124,7 @@  odpapiinclude_HEADERS = \
 		  $(top_srcdir)/include/odp/api/ticketlock.h \
 		  $(top_srcdir)/include/odp/api/time.h \
 		  $(top_srcdir)/include/odp/api/timer.h \
+		  $(top_srcdir)/include/odp/api/traffic_mngr.h \
 		  $(top_srcdir)/include/odp/api/version.h
 
 noinst_HEADERS = \
diff --git a/platform/linux-generic/include/odp/plat/packet_types.h b/platform/linux-generic/include/odp/plat/packet_types.h
index 45cb801..c3be633 100644
--- a/platform/linux-generic/include/odp/plat/packet_types.h
+++ b/platform/linux-generic/include/odp/plat/packet_types.h
@@ -36,6 +36,17 @@  typedef ODP_HANDLE_T(odp_packet_seg_t);
 
 #define ODP_PACKET_SEG_INVALID _odp_cast_scalar(odp_packet_seg_t, 0xffffffff)
 
+/** odp_packet_color_t assigns names to the various pkt "colors" */
+typedef enum {
+	ODP_PACKET_GREEN = 0,
+	ODP_PACKET_YELLOW = 1,
+	ODP_PACKET_RED = 2,
+	ODP_PACKET_ALL_COLORS = 3,
+} odp_packet_color_t;
+
+/** Sets the maximum number of pkt "colors" */
+#define ODP_NUM_PACKET_COLORS 3
+
 /** Get printable format of odp_packet_t */
 static inline uint64_t odp_packet_to_u64(odp_packet_t hdl)
 {
diff --git a/platform/linux-generic/include/odp/plat/traffic_mngr_types.h b/platform/linux-generic/include/odp/plat/traffic_mngr_types.h
new file mode 100644
index 0000000..52df64b
--- /dev/null
+++ b/platform/linux-generic/include/odp/plat/traffic_mngr_types.h
@@ -0,0 +1,185 @@ 
+/* Copyright (c) 2015, Linaro Limited
+ * All rights reserved.
+ *
+ * SPDX-License-Identifier:     BSD-3-Clause
+ */
+
+/**
+ * @file
+ *
+ * ODP traffic mngr
+ */
+
+#ifndef ODP_TRAFFIC_MNGR_TYPES_H_
+#define ODP_TRAFFIC_MNGR_TYPES_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <odp/std_types.h>
+#include <odp/plat/strong_types.h>
+
+/** @addtogroup odp_traffic_mngr
+ *  Macros and operations on a TM system.
+ *  @{
+ */
+
+/** The ODP_TM_MAX_NUM_SYSTEMS constant specifies the maximum number of TM
+ * systems that may be created.  On some platforms this might be much more
+ * limited to as little as one hardware TM system.
+ */
+#define ODP_TM_MAX_NUM_SYSTEMS   64
+
+/** The ODP_TM_MAX_PRIORITIES constant specifies the largest range of
+ * priorities that any TM system can support.  All strict priority values MUST
+ * in the range 0..ODP_TM_MAX_PRIORITIES-1.
+ */
+#define ODP_TM_MAX_PRIORITIES  16
+
+/** The ODP_TM MAX_LEVELS constant specifies the largest range of
+ * tm_node levels that any TM system can support.  Hence all tm_node level
+ * values MUST be in the range 0..ODP_TM_MAX_LEVELS-1.  Smaller tm_node
+ * levels are associated with tm_nodes closer to the TM system egress.
+ */
+#define ODP_TM_MAX_LEVELS  8
+
+/**
+ * The smallest SCHED weight is 1 (i.e. 0 is not a legal WFQ/WRR value).
+ */
+#define ODP_TM_MIN_SCHED_WEIGHT  1
+
+/** The ODP_TM_MAX_SCHED_WEIGHT constant is the largest weight any TM system
+ * can support (at least from a configuration standpoint).  A given TM system
+ * could have a smaller value.
+ */
+#define ODP_TM_MAX_SCHED_WEIGHT  255
+
+/** The ODP_TM_MAX_TM_QUEUES constant is the largest number of tm_queues
+ * that can handled by any one TM system.
+ */
+#define ODP_TM_MAX_TM_QUEUES  (16 * 1024 * 1024)
+
+/** The ODP_TM_MAX_NUM_OUTPUTS constant is the largest number of outputs that
+ * can be configured for any one TM system.
+ */
+#define ODP_TM_MAX_NUM_OUTPUTS  256
+
+/** The ODP_TM_MAX_NUM_TM_NODES constant is the largest number of tm_nodes that
+ * can be in existence for any one TM system.
+ */
+#define ODP_TM_MAX_NUM_TM_NODES  (1024 * 1024)
+
+/** The ODP_TM_MAX_TM_NODE_FANIN constant is the largest number of fan-in
+ * "inputs" that can be simultaneously connected to a single tm_node.
+ * *TBD* Does this need to be as large as ODP_TM_MAX_TM_QUEUES? *TBD*
+ */
+#define ODP_TM_MAX_TM_NODE_FANIN  (1024 * 1024)
+
+/** The ODP_TM_MIN_SHAPER_BW constant is the smallest amount of bandwidth that
+ * can a shaper's peak or commit rate can be set to.  It is in units of
+ * 1000 bytes/second so that it and the ODP_TM_MAX_SHAPER_BW can both fit in
+ * 32 bits.
+ */
+#define ODP_TM_MIN_SHAPER_BW  1
+
+/** The ODP_TM_MAX_SHAPER_BW constant is the largest amound of bandwidth that
+ * any shaper's peak or commit rate can be set to.  It is in units of
+ * 1000 bytes/second so that it and the ODP_TM_MIN_SHAPER_BW can both fit in
+ * 32 bits.
+ */
+#define ODP_TM_MAX_SHAPER_BW  12500000
+
+/** The ODP_NUM_SHAPER_COLORS constant just counts the number of enumeration
+ * values defined in the odp_tm_shaper_color_t type.
+ */
+#define ODP_NUM_SHAPER_COLORS  3
+
+/** The INVALID_PRIORITY constant is used when one needs to indicate an
+ * invalid priority value.
+ */
+#define ODP_TM_INVALID_PRIORITY  255
+
+/** The odp_tm_percent_t type is used when specifying fields that are
+ * percentages.  It is a fixed point integer whose units are 1/100 of a
+ * percent.  Hence 100% is represented as the integer value 10000.  Note
+ * that because it is often used as a ratio of the current queue value and
+ * maximum queue threshold, it can be > 100%, but in any event will never
+ * be larger than 500% (i.e. it MUST be capped at 50000).
+ */
+typedef uint16_t odp_tm_percent_t;
+
+/** The odp_tm_handle_t type is a generic type that can stand for any of the
+ * other ODP_TM handle types.
+ */
+typedef uint64_t odp_tm_handle_t;
+
+/** Each odp_tm_t value represents a specific TM system.  Almost all
+ * functions in this API require a odp_tm_t value - either directly
+ * as a function parameter or indirectly by having another ODP TM handle value
+ * as a function parameter.
+ */
+typedef odp_tm_handle_t odp_tm_t;
+
+/** Each odp_tm_queue_t value is an opaque ODP handle representing a specific
+ * tm_queue within a specific TM system.
+ */
+typedef odp_tm_handle_t odp_tm_queue_t;
+
+/** Each odp_tm_node_t value is an opaque ODP handle representing a specific
+ * tm_node within a specific TM system.
+ */
+typedef odp_tm_handle_t odp_tm_node_t;
+
+/** Each odp_tm_shaper_t value is an opaque ODP handle representing a specific
+ * shaper profile usable across all TM systems described by this API.  A given
+ * shaper profile can then be attached to any tm_queue or tm_node.
+ */
+typedef odp_tm_handle_t odp_tm_shaper_t;
+
+/** Each odp_tm_sched_t value is an opaque ODP handle representing a specific
+ * tm_node scheduler profile usable across all TM systems described by this
+ * API.  A given tm_node scheduler profile can then be attached to any tm_node.
+ */
+typedef odp_tm_handle_t odp_tm_sched_t;
+
+/** Each odp_tm_threshold_t value is an opaque ODP handle representing a
+ * specific queue threshold profile usable across all TM systems described by
+ * this API.  A given queue threshold profile can then be attached to any
+ * tm_queue or tm_node.
+ */
+typedef odp_tm_handle_t odp_tm_threshold_t;
+
+/** Each odp_tm_wred_t value is an opaque ODP handle representing a specific
+ * WRED profile usable across all TM systems described by this API.  A given
+ * WRED profile can then be attached to any tm_queue or tm_node.
+ */
+typedef odp_tm_handle_t odp_tm_wred_t;
+
+/** The ODP_TM_INVALID constant can be used with any ODP TM handle type and
+ * indicates that this value does NOT represent a valid TM object.
+ */
+#define ODP_TM_INVALID  0
+
+/**
+ * @def ODP_TM_ROOT
+ * Constant that is used to refer to the egress/root node of the TM subsystem's
+ * tree/hierarchy of nodes.
+ */
+#define ODP_TM_ROOT 0
+
+/** Get printable format of odp_queue_t */
+static inline uint64_t odp_tm_handle_to_u64(odp_tm_handle_t hdl)
+{
+	return hdl;
+}
+
+/**
+ * @}
+ */
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/platform/linux-generic/include/odp/traffic_mngr.h b/platform/linux-generic/include/odp/traffic_mngr.h
new file mode 100644
index 0000000..3aa6267
--- /dev/null
+++ b/platform/linux-generic/include/odp/traffic_mngr.h
@@ -0,0 +1,35 @@ 
+/* Copyright (c) 2015, Linaro Limited
+ * All rights reserved.
+ *
+ * SPDX-License-Identifier:     BSD-3-Clause
+ */
+
+/**
+ * @file
+ *
+ * ODP Traffic manager
+ */
+
+#ifndef ODP_PLAT_TRAFFIC_MNGR_H_
+#define ODP_PLAT_TRAFFIC_MNGR_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/** @ingroup odp_traffic_mngr
+ *  @{
+ */
+
+/**
+ * @}
+ */
+
+#include <odp/plat/traffic_mngr_types.h>
+#include <odp/api/traffic_mngr.h>
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/platform/linux-generic/include/odp_packet_internal.h b/platform/linux-generic/include/odp_packet_internal.h
index 6f1521c..a1dd375 100644
--- a/platform/linux-generic/include/odp_packet_internal.h
+++ b/platform/linux-generic/include/odp_packet_internal.h
@@ -62,6 +62,9 @@  typedef union {
 		uint32_t tcpopt:1;    /**< TCP options present */
 		uint32_t sctp:1;      /**< SCTP */
 		uint32_t icmp:1;      /**< ICMP */
+
+		uint32_t color:2;     /**< Packet color for traffic mgmt */
+		uint32_t nodrop:1;    /**< Drop eligibility status */
 	};
 } input_flags_t;
 
@@ -103,6 +106,8 @@  typedef union {
 		uint32_t l3_chksum:1;     /**< L3 chksum override */
 		uint32_t l4_chksum_set:1; /**< L3 chksum bit is valid */
 		uint32_t l4_chksum:1;     /**< L4 chksum override  */
+
+		int8_t shaper_len_adj;    /**< adjustment for traffic mgr */
 	};
 } output_flags_t;