doc: remove trailing white space
[dpdk.git] / doc / guides / sample_app_ug / l3_forward.rst
index 2635f40..07c8d44 100644 (file)
@@ -1,46 +1,26 @@
-..  BSD LICENSE
-    Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
-    All rights reserved.
-
-    Redistribution and use in source and binary forms, with or without
-    modification, are permitted provided that the following conditions
-    are met:
-
-    * Redistributions of source code must retain the above copyright
-    notice, this list of conditions and the following disclaimer.
-    * Redistributions in binary form must reproduce the above copyright
-    notice, this list of conditions and the following disclaimer in
-    the documentation and/or other materials provided with the
-    distribution.
-    * Neither the name of Intel Corporation nor the names of its
-    contributors may be used to endorse or promote products derived
-    from this software without specific prior written permission.
-
-    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-    "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-    A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-    OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-    SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-    LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-    DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-    THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-    (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-    OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+..  SPDX-License-Identifier: BSD-3-Clause
+    Copyright(c) 2010-2014 Intel Corporation.
 
 L3 Forwarding Sample Application
 ================================
 
 
 L3 Forwarding Sample Application
 ================================
 
-The L3 Forwarding application is a simple example of packet processing using the DPDK.
+The L3 Forwarding application is a simple example of packet processing using
+DPDK to demonstrate usage of poll and event mode packet I/O mechanism.
 The application performs L3 forwarding.
 
 Overview
 --------
 
 The application performs L3 forwarding.
 
 Overview
 --------
 
-The application demonstrates the use of the hash and LPM libraries in the DPDK to implement packet forwarding.
-The initialization and run-time paths are very similar to those of the :doc:`l2_forward_real_virtual`.
-The main difference from the L2 Forwarding sample application is that the forwarding decision
-is made based on information read from the input packet.
+The application demonstrates the use of the hash and LPM libraries in the DPDK
+to implement packet forwarding using poll or event mode PMDs for packet I/O.
+The initialization and run-time paths are very similar to those of the
+:doc:`l2_forward_real_virtual` and :doc:`l2_forward_event`.
+The main difference from the L2 Forwarding sample application is that optionally
+packet can be Rx/Tx from/to eventdev instead of port directly and forwarding
+decision is made based on information read from the input packet.
+
+Eventdev can optionally use S/W or H/W (if supported by platform) scheduler
+implementation for packet I/O based on run time parameters.
 
 The lookup method is either hash-based or LPM-based and is selected at run time. When the selected lookup method is hash-based,
 a hash object is used to emulate the flow classification stage.
 
 The lookup method is either hash-based or LPM-based and is selected at run time. When the selected lookup method is hash-based,
 a hash object is used to emulate the flow classification stage.
@@ -62,28 +42,9 @@ In the sample application, hash-based forwarding supports IPv4 and IPv6. LPM-bas
 Compiling the Application
 -------------------------
 
 Compiling the Application
 -------------------------
 
-To compile the application:
-
-#.  Go to the sample application directory:
-
-    .. code-block:: console
-
-        export RTE_SDK=/path/to/rte_sdk
-        cd ${RTE_SDK}/examples/l3fwd
-
-#.  Set the target (a default target is used if not specified). For example:
-
-    .. code-block:: console
-
-        export RTE_TARGET=x86_64-native-linuxapp-gcc
-
-    See the *DPDK Getting Started Guide* for possible RTE_TARGET values.
-
-#.  Build the application:
-
-    .. code-block:: console
+To compile the sample application see :doc:`compiling`.
 
 
-        make
+The application is located in the ``l3fwd`` sub-directory.
 
 Running the Application
 -----------------------
 
 Running the Application
 -----------------------
@@ -101,6 +62,10 @@ The application has a number of command line options::
                              [--hash-entry-num]
                              [--ipv6]
                              [--parse-ptype]
                              [--hash-entry-num]
                              [--ipv6]
                              [--parse-ptype]
+                             [--per-port-pool]
+                             [--mode]
+                             [--eventq-sched]
+                             [--event-eth-rxqs]
 
 Where,
 
 
 Where,
 
@@ -129,6 +94,15 @@ Where,
 
 * ``--parse-ptype:`` Optional, set to use software to analyze packet type. Without this option, hardware will check the packet type.
 
 
 * ``--parse-ptype:`` Optional, set to use software to analyze packet type. Without this option, hardware will check the packet type.
 
+* ``--per-port-pool:`` Optional, set to use independent buffer pools per port. Without this option, single buffer pool is used for all ports.
+
+* ``--mode:`` Optional, Packet transfer mode for I/O, poll or eventdev.
+
+* ``--eventq-sched:`` Optional, Event queue synchronization method, Ordered, Atomic or Parallel. Only valid if --mode=eventdev.
+
+* ``--event-eth-rxqs:`` Optional, Number of ethernet RX queues per device. Only valid if --mode=eventdev.
+
+
 For example, consider a dual processor socket platform with 8 physical cores, where cores 0-7 and 16-23 appear on socket 0,
 while cores 8-15 and 24-31 appear on socket 1.
 
 For example, consider a dual processor socket platform with 8 physical cores, where cores 0-7 and 16-23 appear on socket 0,
 while cores 8-15 and 24-31 appear on socket 1.
 
@@ -159,6 +133,51 @@ In this command:
 |          |           |           |                                     |
 +----------+-----------+-----------+-------------------------------------+
 
 |          |           |           |                                     |
 +----------+-----------+-----------+-------------------------------------+
 
+To use eventdev mode with sync method **ordered** on above mentioned environment,
+Following is the sample command:
+
+.. code-block:: console
+
+    ./build/l3fwd -l 0-3 -n 4 -w <event device> -- -p 0x3 --eventq-sched=ordered
+
+or
+
+.. code-block:: console
+
+    ./build/l3fwd -l 0-3 -n 4 -w <event device> -- -p 0x03 --mode=eventdev --eventq-sched=ordered
+
+In this command:
+
+*   -w option whitelist the event device supported by platform. Way to pass this device may vary based on platform.
+
+*   The --mode option defines PMD to be used for packet I/O.
+
+*   The --eventq-sched option enables synchronization menthod of event queue so that packets will be scheduled accordingly.
+
+If application uses S/W scheduler, it uses following DPDK services:
+
+*   Software scheduler
+*   Rx adapter service function
+*   Tx adapter service function
+
+Application needs service cores to run above mentioned services. Service cores
+must be provided as EAL parameters along with the --vdev=event_sw0 to enable S/W
+scheduler. Following is the sample command:
+
+.. code-block:: console
+
+    ./build/l3fwd -l 0-7 -s 0xf0000 -n 4 --vdev event_sw0 -- -p 0x3 --mode=eventdev --eventq-sched=ordered
+
+In case of eventdev mode, *--config* option is not used for ethernet port
+configuration. Instead each ethernet port will be configured with mentioned
+setup:
+
+*   Single Rx/Tx queue
+
+*   Each Rx queue will be connected to event queue via Rx adapter.
+
+*   Each Tx queue will be connected via Tx adapter.
+
 Refer to the *DPDK Getting Started Guide* for general information on running applications and
 the Environment Abstraction Layer (EAL) options.
 
 Refer to the *DPDK Getting Started Guide* for general information on running applications and
 the Environment Abstraction Layer (EAL) options.
 
@@ -168,7 +187,7 @@ Explanation
 -----------
 
 The following sections provide some explanation of the sample application code. As mentioned in the overview section,
 -----------
 
 The following sections provide some explanation of the sample application code. As mentioned in the overview section,
-the initialization and run-time paths are very similar to those of the :doc:`l2_forward_real_virtual`.
+the initialization and run-time paths are very similar to those of the :doc:`l2_forward_real_virtual` and :doc:`l2_forward_event`.
 The following sections describe aspects that are specific to the L3 Forwarding sample application.
 
 Hash Initialization
 The following sections describe aspects that are specific to the L3 Forwarding sample application.
 
 Hash Initialization
@@ -288,7 +307,7 @@ The get_ipv4_dst_port() function is shown below:
         int ret = 0;
         union ipv4_5tuple_host key;
 
         int ret = 0;
         union ipv4_5tuple_host key;
 
-        ipv4_hdr = (uint8_t *)ipv4_hdr + offsetof(struct ipv4_hdr, time_to_live);
+        ipv4_hdr = (uint8_t *)ipv4_hdr + offsetof(struct rte_ipv4_hdr, time_to_live);
 
         m128i data = _mm_loadu_si128(( m128i*)(ipv4_hdr));
 
 
         m128i data = _mm_loadu_si128(( m128i*)(ipv4_hdr));
 
@@ -316,10 +335,10 @@ The key code snippet of simple_ipv4_fwd_4pkts() is shown below:
     {
         // ...
 
     {
         // ...
 
-        data[0] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[0], unsigned char *) + sizeof(struct ether_hdr) + offsetof(struct ipv4_hdr, time_to_live)));
-        data[1] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[1], unsigned char *) + sizeof(struct ether_hdr) + offsetof(struct ipv4_hdr, time_to_live)));
-        data[2] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[2], unsigned char *) + sizeof(struct ether_hdr) + offsetof(struct ipv4_hdr, time_to_live)));
-        data[3] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[3], unsigned char *) + sizeof(struct ether_hdr) + offsetof(struct ipv4_hdr, time_to_live)));
+        data[0] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[0], unsigned char *) + sizeof(struct rte_ether_hdr) + offsetof(struct rte_ipv4_hdr, time_to_live)));
+        data[1] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[1], unsigned char *) + sizeof(struct rte_ether_hdr) + offsetof(struct rte_ipv4_hdr, time_to_live)));
+        data[2] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[2], unsigned char *) + sizeof(struct rte_ether_hdr) + offsetof(struct rte_ipv4_hdr, time_to_live)));
+        data[3] = _mm_loadu_si128(( m128i*)(rte_pktmbuf_mtod(m[3], unsigned char *) + sizeof(struct rte_ether_hdr) + offsetof(struct rte_ipv4_hdr, time_to_live)));
 
         key[0].xmm = _mm_and_si128(data[0], mask0);
         key[1].xmm = _mm_and_si128(data[1], mask0);
 
         key[0].xmm = _mm_and_si128(data[0], mask0);
         key[1].xmm = _mm_and_si128(data[1], mask0);
@@ -352,9 +371,14 @@ for LPM-based lookups is done by the get_ipv4_dst_port() function below:
 .. code-block:: c
 
     static inline uint16_t
 .. code-block:: c
 
     static inline uint16_t
-    get_ipv4_dst_port(struct ipv4_hdr *ipv4_hdr, uint16_t portid, lookup_struct_t *ipv4_l3fwd_lookup_struct)
+    get_ipv4_dst_port(struct rte_ipv4_hdr *ipv4_hdr, uint16_t portid, lookup_struct_t *ipv4_l3fwd_lookup_struct)
     {
         uint8_t next_hop;
 
         return ((rte_lpm_lookup(ipv4_l3fwd_lookup_struct, rte_be_to_cpu_32(ipv4_hdr->dst_addr), &next_hop) == 0)? next_hop : portid);
     }
     {
         uint8_t next_hop;
 
         return ((rte_lpm_lookup(ipv4_l3fwd_lookup_struct, rte_be_to_cpu_32(ipv4_hdr->dst_addr), &next_hop) == 0)? next_hop : portid);
     }
+
+Eventdev Driver Initialization
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Eventdev driver initialization is same as L2 forwarding eventdev application.
+Refer :doc:`l2_forward_event` for more details.