2 Copyright 2012 6WIND S.A.
3 Copyright 2015 Mellanox
5 Redistribution and use in source and binary forms, with or without
6 modification, are permitted provided that the following conditions
9 * Redistributions of source code must retain the above copyright
10 notice, this list of conditions and the following disclaimer.
11 * Redistributions in binary form must reproduce the above copyright
12 notice, this list of conditions and the following disclaimer in
13 the documentation and/or other materials provided with the
15 * Neither the name of 6WIND S.A. nor the names of its
16 contributors may be used to endorse or promote products derived
17 from this software without specific prior written permission.
19 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
31 MLX4 poll mode driver library
32 =============================
34 The MLX4 poll mode driver library (**librte_pmd_mlx4**) implements support
35 for **Mellanox ConnectX-3** and **Mellanox ConnectX-3 Pro** 10/40 Gbps adapters
36 as well as their virtual functions (VF) in SR-IOV context.
38 Information and documentation about this family of adapters can be found on
39 the `Mellanox website <http://www.mellanox.com>`_. Help is also provided by
40 the `Mellanox community <http://community.mellanox.com/welcome>`_.
42 There is also a `section dedicated to this poll mode driver
43 <http://www.mellanox.com/page/products_dyn?product_family=209&mtag=pmd_for_dpdk>`_.
47 Due to external dependencies, this driver is disabled by default. It must
48 be enabled manually by setting ``CONFIG_RTE_LIBRTE_MLX4_PMD=y`` and
51 Implementation details
52 ----------------------
54 Most Mellanox ConnectX-3 devices provide two ports but expose a single PCI
55 bus address, thus unlike most drivers, librte_pmd_mlx4 registers itself as a
56 PCI driver that allocates one Ethernet device per detected port.
58 For this reason, one cannot white/blacklist a single port without also
59 white/blacklisting the others on the same device.
61 Besides its dependency on libibverbs (that implies libmlx4 and associated
62 kernel support), librte_pmd_mlx4 relies heavily on system calls for control
63 operations such as querying/updating the MTU and flow control parameters.
65 For security reasons and robustness, this driver only deals with virtual
66 memory addresses. The way resources allocations are handled by the kernel
67 combined with hardware specifications that allow it to handle virtual memory
68 addresses directly ensure that DPDK applications cannot access random
69 physical memory (or memory that does not belong to the current process).
71 This capability allows the PMD to coexist with kernel network interfaces
72 which remain functional, although they stop receiving unicast packets as
73 long as they share the same MAC address.
75 Compiling librte_pmd_mlx4 causes DPDK to be linked against libibverbs.
83 These options can be modified in the ``.config`` file.
85 - ``CONFIG_RTE_LIBRTE_MLX4_PMD`` (default **n**)
87 Toggle compilation of librte_pmd_mlx4 itself.
89 - ``CONFIG_RTE_LIBRTE_MLX4_DEBUG`` (default **n**)
91 Toggle debugging code and stricter compilation flags. Enabling this option
92 adds additional run-time checks and debugging messages at the cost of
95 - ``CONFIG_RTE_LIBRTE_MLX4_TX_MP_CACHE`` (default **8**)
97 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
98 which buffers are to be transmitted must be associated to memory regions
99 (MRs). This is a slow operation that must be cached.
101 This value is always 1 for RX queues since they use a single MP.
103 Run-time configuration
104 ~~~~~~~~~~~~~~~~~~~~~~
106 - librte_pmd_mlx4 brings kernel network interfaces up during initialization
107 because it is affected by their state. Forcing them down prevents packets
110 - **ethtool** operations on related kernel interfaces also affect the PMD.
112 - ``port`` parameter [int]
114 This parameter provides a physical port to probe and can be specified multiple
115 times for additional ports. All ports are probed by default if left
118 Kernel module parameters
119 ~~~~~~~~~~~~~~~~~~~~~~~~
121 The **mlx4_core** kernel module has several parameters that affect the
122 behavior and/or the performance of librte_pmd_mlx4. Some of them are described
125 - **num_vfs** (integer or triplet, optionally prefixed by device address
128 Create the given number of VFs on the specified devices.
130 - **log_num_mgm_entry_size** (integer)
132 Device-managed flow steering (DMFS) is required by DPDK applications. It is
133 enabled by using a negative value, the last four bits of which have a
136 - **-1**: force device-managed flow steering (DMFS).
137 - **-7**: configure optimized steering mode to improve performance with the
138 following limitation: VLAN filtering is not supported with this mode.
139 This is the recommended mode in case VLAN filter is not needed.
144 This driver relies on external libraries and kernel drivers for resources
145 allocations and initialization. The following dependencies are not part of
146 DPDK and must be installed separately:
148 - **libibverbs** (provided by rdma-core package)
150 User space verbs framework used by librte_pmd_mlx4. This library provides
151 a generic interface between the kernel and low-level user space drivers
154 It allows slow and privileged operations (context initialization, hardware
155 resources allocations) to be managed by the kernel and fast operations to
156 never leave user space.
158 - **libmlx4** (provided by rdma-core package)
160 Low-level user space driver library for Mellanox ConnectX-3 devices,
161 it is automatically loaded by libibverbs.
163 This library basically implements send/receive calls to the hardware
168 They provide the kernel-side verbs API and low level device drivers that
169 manage actual hardware initialization and resources sharing with user
172 Unlike most other PMDs, these modules must remain loaded and bound to
175 - mlx4_core: hardware driver managing Mellanox ConnectX-3 devices.
176 - mlx4_en: Ethernet device driver that provides kernel network interfaces.
177 - mlx4_ib: InifiniBand device driver.
178 - ib_uverbs: user space driver for verbs (entry point for libibverbs).
180 - **Firmware update**
182 Mellanox OFED releases include firmware updates for ConnectX-3 adapters.
184 Because each release provides new features, these updates must be applied to
185 match the kernel modules and libraries they come with.
189 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
192 Depending on system constraints and user preferences either RDMA core library
193 with a recent enough Linux kernel release (recommended) or Mellanox OFED,
194 which provides compatibility with older releases.
196 Current RDMA core package and Linux kernel (recommended)
197 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
199 - Minimal Linux kernel version: 4.14.
200 - Minimal RDMA core version: v15 (see `RDMA core installation documentation`_).
202 .. _`RDMA core installation documentation`: https://raw.githubusercontent.com/linux-rdma/rdma-core/master/README.md
204 .. _Mellanox_OFED_as_a_fallback:
206 Mellanox OFED as a fallback
207 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
209 - `Mellanox OFED`_ version: **4.2**.
210 - firmware version: **2.42.5000** and above.
212 .. _`Mellanox OFED`: http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers
216 Several versions of Mellanox OFED are available. Installing the version
217 this DPDK release was developed and tested against is strongly
218 recommended. Please check the `prerequisites`_.
220 Installing Mellanox OFED
221 ^^^^^^^^^^^^^^^^^^^^^^^^
223 1. Download latest Mellanox OFED.
225 2. Install the required libraries and kernel modules either by installing
226 only the required set, or by installing the entire Mellanox OFED:
230 .. code-block:: console
232 ./mlnxofedinstall --dpdk --upstream-libs
234 For SR-IOV hypervisors use:
236 .. code-block:: console
238 ./mlnxofedinstall --dpdk --upstream-libs --enable-sriov --hypervisor
240 For SR-IOV virtual machine use:
242 .. code-block:: console
244 ./mlnxofedinstall --dpdk --upstream-libs --guest
246 3. Verify the firmware is the correct one:
248 .. code-block:: console
252 4. Set all ports links to Ethernet, follow instructions on the screen:
254 .. code-block:: console
258 5. Continue with :ref:`section 2 of the Quick Start Guide <QSG_2>`.
263 * Mellanox(R) ConnectX(R)-3 Pro 40G MCX354A-FCC_Ax (2*40G)
270 1. Set all ports links to Ethernet
272 .. code-block:: console
274 PCI=<NIC PCI address>
275 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port0"
276 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port1"
280 If using Mellanox OFED one can permanently set the port link
281 to Ethernet using connectx_port_config tool provided by it.
282 :ref:`Mellanox_OFED_as_a_fallback`:
286 2. In case of bare metal or hypervisor, configure optimized steering mode
287 by adding the following line to ``/etc/modprobe.d/mlx4_core.conf``:
289 .. code-block:: console
291 options mlx4_core log_num_mgm_entry_size=-7
295 If VLAN filtering is used, set log_num_mgm_entry_size=-1.
296 Performance degradation can occur on this case.
298 3. Restart the driver:
300 .. code-block:: console
302 /etc/init.d/openibd restart
306 .. code-block:: console
308 service openibd restart
310 4. Compile DPDK and you are ready to go. See instructions on
311 :ref:`Development Kit Build System <Development_Kit_Build_System>`
316 1. Verify the optimized steering mode is configured:
318 .. code-block:: console
320 cat /sys/module/mlx4_core/parameters/log_num_mgm_entry_size
322 2. Use the CPU near local NUMA node to which the PCIe adapter is connected,
323 for better performance. For VMs, verify that the right CPU
324 and NUMA node are pinned according to the above. Run:
326 .. code-block:: console
330 to identify the NUMA node to which the PCIe adapter is connected.
332 3. If more than one adapter is used, and root complex capabilities allow
333 to put both adapters on the same NUMA node without PCI bandwidth degradation,
334 it is recommended to locate both adapters on the same NUMA node.
335 This in order to forward packets from one to the other without
336 NUMA performance penalty.
338 4. Disable pause frames:
340 .. code-block:: console
342 ethtool -A <netdev> rx off tx off
344 5. Verify IO non-posted prefetch is disabled by default. This can be checked
345 via the BIOS configuration. Please contact you server provider for more
346 information about the settings.
350 On some machines, depends on the machine integrator, it is beneficial
351 to set the PCI max read request parameter to 1K. This can be
352 done in the following way:
354 To query the read request size use:
356 .. code-block:: console
358 setpci -s <NIC PCI address> 68.w
360 If the output is different than 3XXX, set it by:
362 .. code-block:: console
364 setpci -s <NIC PCI address> 68.w=3XXX
366 The XXX can be different on different systems. Make sure to configure
367 according to the setpci output.
372 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-3
373 devices managed by librte_pmd_mlx4.
375 #. Load the kernel modules:
377 .. code-block:: console
379 modprobe -a ib_uverbs mlx4_en mlx4_core mlx4_ib
381 Alternatively if MLNX_OFED is fully installed, the following script can
384 .. code-block:: console
386 /etc/init.d/openibd restart
390 User space I/O kernel modules (uio and igb_uio) are not used and do
391 not have to be loaded.
393 #. Make sure Ethernet interfaces are in working order and linked to kernel
394 verbs. Related sysfs entries should be present:
396 .. code-block:: console
398 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
402 .. code-block:: console
409 #. Optionally, retrieve their PCI bus addresses for whitelisting:
411 .. code-block:: console
414 for intf in eth2 eth3 eth4 eth5;
416 (cd "/sys/class/net/${intf}/device/" && pwd -P);
419 sed -n 's,.*/\(.*\),-w \1,p'
423 .. code-block:: console
432 There are only two distinct PCI bus addresses because the Mellanox
433 ConnectX-3 adapters installed on this system are dual port.
435 #. Request huge pages:
437 .. code-block:: console
439 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
441 #. Start testpmd with basic parameters:
443 .. code-block:: console
445 testpmd -l 8-15 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
449 .. code-block:: console
452 EAL: PCI device 0000:83:00.0 on NUMA socket 1
453 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
454 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_0" (VF: false)
455 PMD: librte_pmd_mlx4: 2 port(s) detected
456 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:b7:50
457 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:b7:51
458 EAL: PCI device 0000:84:00.0 on NUMA socket 1
459 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
460 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_1" (VF: false)
461 PMD: librte_pmd_mlx4: 2 port(s) detected
462 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:ba:b0
463 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:ba:b1
464 Interactive-mode selected
465 Configuring Port 0 (socket 0)
466 PMD: librte_pmd_mlx4: 0x867d60: TX queues number update: 0 -> 2
467 PMD: librte_pmd_mlx4: 0x867d60: RX queues number update: 0 -> 2
468 Port 0: 00:02:C9:B5:B7:50
469 Configuring Port 1 (socket 0)
470 PMD: librte_pmd_mlx4: 0x867da0: TX queues number update: 0 -> 2
471 PMD: librte_pmd_mlx4: 0x867da0: RX queues number update: 0 -> 2
472 Port 1: 00:02:C9:B5:B7:51
473 Configuring Port 2 (socket 0)
474 PMD: librte_pmd_mlx4: 0x867de0: TX queues number update: 0 -> 2
475 PMD: librte_pmd_mlx4: 0x867de0: RX queues number update: 0 -> 2
476 Port 2: 00:02:C9:B5:BA:B0
477 Configuring Port 3 (socket 0)
478 PMD: librte_pmd_mlx4: 0x867e20: TX queues number update: 0 -> 2
479 PMD: librte_pmd_mlx4: 0x867e20: RX queues number update: 0 -> 2
480 Port 3: 00:02:C9:B5:BA:B1
481 Checking link statuses...
482 Port 0 Link Up - speed 10000 Mbps - full-duplex
483 Port 1 Link Up - speed 40000 Mbps - full-duplex
484 Port 2 Link Up - speed 10000 Mbps - full-duplex
485 Port 3 Link Up - speed 40000 Mbps - full-duplex