-.. BSD LICENSE
- Copyright 2012-2015 6WIND S.A.
-
- 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 6WIND S.A. 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 2012 6WIND S.A.
+ Copyright 2015 Mellanox Technologies, Ltd
MLX4 poll mode driver library
=============================
The MLX4 poll mode driver library (**librte_pmd_mlx4**) implements support
-for **Mellanox ConnectX-3 EN** 10/40 Gbps adapters as well as their virtual
-functions (VF) in SR-IOV context.
+for **Mellanox ConnectX-3** and **Mellanox ConnectX-3 Pro** 10/40 Gbps adapters
+as well as their virtual functions (VF) in SR-IOV context.
Information and documentation about this family of adapters can be found on
the `Mellanox website <http://www.mellanox.com>`_. Help is also provided by
which remain functional, although they stop receiving unicast packets as
long as they share the same MAC address.
-Compiling librte_pmd_mlx4 causes DPDK to be linked against libibverbs.
-
-Features and limitations
-------------------------
-
-- RSS, also known as RCA, is supported. In this mode the number of
- configured RX queues must be a power of two.
-- VLAN filtering is supported.
-- Link state information is provided.
-- Promiscuous mode is supported.
-- All multicast mode is supported.
-- Multiple MAC addresses (unicast, multicast) can be configured.
-- Scattered packets are supported for TX and RX.
+The :ref:`flow_isolated_mode` is supported.
-.. break
-
-- RSS hash key cannot be modified.
-- Hardware counters are not implemented (they are software counters).
-- Checksum offloads are not supported yet.
+Compiling librte_pmd_mlx4 causes DPDK to be linked against libibverbs.
Configuration
-------------
Toggle compilation of librte_pmd_mlx4 itself.
-- ``CONFIG_RTE_LIBRTE_MLX4_DEBUG`` (default **n**)
+- ``CONFIG_RTE_IBVERBS_LINK_DLOPEN`` (default **n**)
- Toggle debugging code and stricter compilation flags. Enabling this option
- adds additional run-time checks and debugging messages at the cost of
- lower performance.
+ Build PMD with additional code to make it loadable without hard
+ dependencies on **libibverbs** nor **libmlx4**, which may not be installed
+ on the target system.
-- ``CONFIG_RTE_LIBRTE_MLX4_SGE_WR_N`` (default **4**)
+ In this mode, their presence is still required for it to run properly,
+ however their absence won't prevent a DPDK application from starting (with
+ ``CONFIG_RTE_BUILD_SHARED_LIB`` disabled) and they won't show up as
+ missing with ``ldd(1)``.
- Number of scatter/gather elements (SGEs) per work request (WR). Lowering
- this number improves performance but also limits the ability to receive
- scattered packets (packets that do not fit a single mbuf). The default
- value is a safe tradeoff.
+ It works by moving these dependencies to a purpose-built rdma-core "glue"
+ plug-in which must either be installed in a directory whose name is based
+ on ``CONFIG_RTE_EAL_PMD_PATH`` suffixed with ``-glue`` if set, or in a
+ standard location for the dynamic linker (e.g. ``/lib``) if left to the
+ default empty string (``""``).
-- ``CONFIG_RTE_LIBRTE_MLX4_MAX_INLINE`` (default **0**)
+ This option has no performance impact.
- Amount of data to be inlined during TX operations. Improves latency but
- lowers throughput.
+- ``CONFIG_RTE_IBVERBS_LINK_STATIC`` (default **n**)
-- ``CONFIG_RTE_LIBRTE_MLX4_TX_MP_CACHE`` (default **8**)
+ Embed static flavor of the dependencies **libibverbs** and **libmlx4**
+ in the PMD shared library or the executable static binary.
- Maximum number of cached memory pools (MPs) per TX queue. Each MP from
- which buffers are to be transmitted must be associated to memory regions
- (MRs). This is a slow operation that must be cached.
+- ``CONFIG_RTE_LIBRTE_MLX4_DEBUG`` (default **n**)
- This value is always 1 for RX queues since they use a single MP.
+ Toggle debugging code and stricter compilation flags. Enabling this option
+ adds additional run-time checks and debugging messages at the cost of
+ lower performance.
-- ``CONFIG_RTE_LIBRTE_MLX4_SOFT_COUNTERS`` (default **1**)
+This option is available in meson:
- Toggle software counters. No counters are available if this option is
- disabled since hardware counters are not supported.
+- ``ibverbs_link`` can be ``static``, ``shared``, or ``dlopen``.
Environment variables
~~~~~~~~~~~~~~~~~~~~~
-- ``MLX4_INLINE_RECV_SIZE``
+- ``MLX4_GLUE_PATH``
+
+ A list of directories in which to search for the rdma-core "glue" plug-in,
+ separated by colons or semi-colons.
- A nonzero value enables inline receive for packets up to that size. May
- significantly improve performance in some cases but lower it in
- others. Requires careful testing.
+ Only matters when compiled with ``CONFIG_RTE_IBVERBS_LINK_DLOPEN``
+ enabled and most useful when ``CONFIG_RTE_EAL_PMD_PATH`` is also set,
+ since ``LD_LIBRARY_PATH`` has no effect in this case.
Run-time configuration
~~~~~~~~~~~~~~~~~~~~~~
-- The only constraint when RSS mode is requested is to make sure the number
- of RX queues is a power of two. This is a hardware requirement.
-
- librte_pmd_mlx4 brings kernel network interfaces up during initialization
because it is affected by their state. Forcing them down prevents packets
reception.
- **ethtool** operations on related kernel interfaces also affect the PMD.
+- ``port`` parameter [int]
+
+ This parameter provides a physical port to probe and can be specified multiple
+ times for additional ports. All ports are probed by default if left
+ unspecified.
+
+- ``mr_ext_memseg_en`` parameter [int]
+
+ A nonzero value enables extending memseg when registering DMA memory. If
+ enabled, the number of entries in MR (Memory Region) lookup table on datapath
+ is minimized and it benefits performance. On the other hand, it worsens memory
+ utilization because registered memory is pinned by kernel driver. Even if a
+ page in the extended chunk is freed, that doesn't become reusable until the
+ entire memory is freed.
+
+ Enabled by default.
+
Kernel module parameters
~~~~~~~~~~~~~~~~~~~~~~~~
- **-1**: force device-managed flow steering (DMFS).
- **-7**: configure optimized steering mode to improve performance with the
- following limitation: Ethernet frames with the port MAC address as the
- destination cannot be received, even in promiscuous mode. Additional MAC
- addresses can still be set by ``rte_eth_dev_mac_addr_addr()``.
+ following limitation: VLAN filtering is not supported with this mode.
+ This is the recommended mode in case VLAN filter is not needed.
+
+Limitations
+-----------
+
+- For secondary process:
+
+ - Forked secondary process not supported.
+ - External memory unregistered in EAL memseg list cannot be used for DMA
+ unless such memory has been registered by ``mlx4_mr_update_ext_mp()`` in
+ primary process and remapped to the same virtual address in secondary
+ process. If the external memory is registered by primary process but has
+ different virtual address in secondary process, unexpected error may happen.
+
+- CRC stripping is supported by default and always reported as "true".
+ The ability to enable/disable CRC stripping requires OFED version
+ 4.3-1.5.0.0 and above or rdma-core version v18 and above.
+
+- TSO (Transmit Segmentation Offload) is supported in OFED version
+ 4.4 and above.
Prerequisites
-------------
allocations and initialization. The following dependencies are not part of
DPDK and must be installed separately:
-- **libibverbs**
+- **libibverbs** (provided by rdma-core package)
User space verbs framework used by librte_pmd_mlx4. This library provides
a generic interface between the kernel and low-level user space drivers
resources allocations) to be managed by the kernel and fast operations to
never leave user space.
-- **libmlx4**
+- **libmlx4** (provided by rdma-core package)
Low-level user space driver library for Mellanox ConnectX-3 devices,
it is automatically loaded by libibverbs.
This library basically implements send/receive calls to the hardware
queues.
-- **Kernel modules** (mlnx-ofed-kernel)
+- **Kernel modules**
They provide the kernel-side verbs API and low level device drivers that
manage actual hardware initialization and resources sharing with user
Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
licensed.
-Currently supported by DPDK:
+Depending on system constraints and user preferences either RDMA core library
+with a recent enough Linux kernel release (recommended) or Mellanox OFED,
+which provides compatibility with older releases.
-- Mellanox OFED **2.4-1**.
-- Firmware version **2.33.5000** and higher.
+Current RDMA core package and Linux kernel (recommended)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-Getting Mellanox OFED
-~~~~~~~~~~~~~~~~~~~~~
+- Minimal Linux kernel version: 4.14.
+- Minimal RDMA core version: v15 (see `RDMA core installation documentation`_).
+
+- Starting with rdma-core v21, static libraries can be built::
+
+ cd build
+ CFLAGS=-fPIC cmake -DIN_PLACE=1 -DENABLE_STATIC=1 -GNinja ..
+ ninja
+
+.. _`RDMA core installation documentation`: https://raw.githubusercontent.com/linux-rdma/rdma-core/master/README.md
-While these libraries and kernel modules are available on OpenFabrics
-Alliance's `website <https://www.openfabrics.org/>`_ and provided by package
-managers on most distributions, this PMD requires Ethernet extensions that
-may not be supported at the moment (this is a work in progress).
+If rdma-core libraries are built but not installed, DPDK makefile can link them,
+thanks to these environment variables:
-`Mellanox OFED
-<http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers>`_
-includes the necessary support and should be used in the meantime. For DPDK,
-only libibverbs, libmlx4, mlnx-ofed-kernel packages and firmware updates are
-required from that distribution.
+ - ``EXTRA_CFLAGS=-I/path/to/rdma-core/build/include``
+ - ``EXTRA_LDFLAGS=-L/path/to/rdma-core/build/lib``
+ - ``PKG_CONFIG_PATH=/path/to/rdma-core/build/lib/pkgconfig``
+
+.. _Mellanox_OFED_as_a_fallback:
+
+Mellanox OFED as a fallback
+~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- `Mellanox OFED`_ version: **4.4, 4.5, 4.6**.
+- firmware version: **2.42.5000** and above.
+
+.. _`Mellanox OFED`: http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers
.. note::
this DPDK release was developed and tested against is strongly
recommended. Please check the `prerequisites`_.
-Getting libibverbs and libmlx4 from DPDK.org
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Installing Mellanox OFED
+^^^^^^^^^^^^^^^^^^^^^^^^
+
+1. Download latest Mellanox OFED.
+
+2. Install the required libraries and kernel modules either by installing
+ only the required set, or by installing the entire Mellanox OFED:
+
+ For bare metal use::
+
+ ./mlnxofedinstall --dpdk --upstream-libs
+
+ For SR-IOV hypervisors use::
+
+ ./mlnxofedinstall --dpdk --upstream-libs --enable-sriov --hypervisor
+
+ For SR-IOV virtual machine use::
+
+ ./mlnxofedinstall --dpdk --upstream-libs --guest
+
+3. Verify the firmware is the correct one::
+
+ ibv_devinfo
+
+4. Set all ports links to Ethernet, follow instructions on the screen::
+
+ connectx_port_config
-Based on Mellanox OFED, optimized libibverbs and libmlx4 versions can be
-optionally downloaded from DPDK.org:
+5. Continue with :ref:`section 2 of the Quick Start Guide <QSG_2>`.
-`<http://www.dpdk.org/download/mlx4>`_
+.. _qsg:
-Some enhancements are done for better performance with DPDK applications and
-are not merged upstream yet.
+Quick Start Guide
+-----------------
-Since it is partly achieved by tuning compilation options to disable features
-not needed by DPDK, linking these libraries statically and avoid system-wide
-installation is the preferred method.
+1. Set all ports links to Ethernet::
-Installation documentation is available from the above link.
+ PCI=<NIC PCI address>
+ echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port0"
+ echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port1"
+
+ .. note::
+
+ If using Mellanox OFED one can permanently set the port link
+ to Ethernet using connectx_port_config tool provided by it.
+ :ref:`Mellanox_OFED_as_a_fallback`:
+
+.. _QSG_2:
+
+2. In case of bare metal or hypervisor, configure optimized steering mode
+ by adding the following line to ``/etc/modprobe.d/mlx4_core.conf``::
+
+ options mlx4_core log_num_mgm_entry_size=-7
+
+ .. note::
+
+ If VLAN filtering is used, set log_num_mgm_entry_size=-1.
+ Performance degradation can occur on this case.
+
+3. Restart the driver::
+
+ /etc/init.d/openibd restart
+
+ or::
+
+ service openibd restart
+
+4. Compile DPDK and you are ready to go. See instructions on
+ :ref:`Development Kit Build System <Development_Kit_Build_System>`
+
+Performance tuning
+------------------
+
+1. Verify the optimized steering mode is configured::
+
+ cat /sys/module/mlx4_core/parameters/log_num_mgm_entry_size
+
+2. Use the CPU near local NUMA node to which the PCIe adapter is connected,
+ for better performance. For VMs, verify that the right CPU
+ and NUMA node are pinned according to the above. Run::
+
+ lstopo-no-graphics
+
+ to identify the NUMA node to which the PCIe adapter is connected.
+
+3. If more than one adapter is used, and root complex capabilities allow
+ to put both adapters on the same NUMA node without PCI bandwidth degradation,
+ it is recommended to locate both adapters on the same NUMA node.
+ This in order to forward packets from one to the other without
+ NUMA performance penalty.
+
+4. Disable pause frames::
+
+ ethtool -A <netdev> rx off tx off
+
+5. Verify IO non-posted prefetch is disabled by default. This can be checked
+ via the BIOS configuration. Please contact you server provider for more
+ information about the settings.
+
+.. note::
+
+ On some machines, depends on the machine integrator, it is beneficial
+ to set the PCI max read request parameter to 1K. This can be
+ done in the following way:
+
+ To query the read request size use::
+
+ setpci -s <NIC PCI address> 68.w
+
+ If the output is different than 3XXX, set it by::
+
+ setpci -s <NIC PCI address> 68.w=3XXX
+
+ The XXX can be different on different systems. Make sure to configure
+ according to the setpci output.
+
+6. To minimize overhead of searching Memory Regions:
+
+ - '--socket-mem' is recommended to pin memory by predictable amount.
+ - Configure per-lcore cache when creating Mempools for packet buffer.
+ - Refrain from dynamically allocating/freeing memory in run-time.
Usage example
-------------
This section demonstrates how to launch **testpmd** with Mellanox ConnectX-3
devices managed by librte_pmd_mlx4.
-#. Load the kernel modules:
-
- .. code-block:: console
+#. Load the kernel modules::
modprobe -a ib_uverbs mlx4_en mlx4_core mlx4_ib
+ Alternatively if MLNX_OFED is fully installed, the following script can
+ be run::
+
+ /etc/init.d/openibd restart
+
.. note::
User space I/O kernel modules (uio and igb_uio) are not used and do
not have to be loaded.
#. Make sure Ethernet interfaces are in working order and linked to kernel
- verbs. Related sysfs entries should be present:
-
- .. code-block:: console
+ verbs. Related sysfs entries should be present::
ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
- Example output:
-
- .. code-block:: console
+ Example output::
eth2
eth3
eth4
eth5
-#. Optionally, retrieve their PCI bus addresses for whitelisting:
-
- .. code-block:: console
+#. Optionally, retrieve their PCI bus addresses for whitelisting::
{
for intf in eth2 eth3 eth4 eth5;
} |
sed -n 's,.*/\(.*\),-w \1,p'
- Example output:
-
- .. code-block:: console
+ Example output::
-w 0000:83:00.0
-w 0000:83:00.0
There are only two distinct PCI bus addresses because the Mellanox
ConnectX-3 adapters installed on this system are dual port.
-#. Request huge pages:
-
- .. code-block:: console
+#. Request huge pages::
echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
-#. Start testpmd with basic parameters:
-
- .. code-block:: console
-
- testpmd -c 0xff00 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
+#. Start testpmd with basic parameters::
- Example output:
+ testpmd -l 8-15 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
- .. code-block:: console
+ Example output::
[...]
EAL: PCI device 0000:83:00.0 on NUMA socket 1