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_DEBUG_BROKEN_VERBS`` (default **n**)
97 Mellanox OFED versions earlier than 4.2 may return false errors from
98 Verbs object destruction APIs after the device is plugged out.
99 Enabling this option replaces assertion checks that cause the program
100 to abort with harmless debugging messages as a workaround.
101 Relevant only when CONFIG_RTE_LIBRTE_MLX4_DEBUG is enabled.
103 - ``CONFIG_RTE_LIBRTE_MLX4_TX_MP_CACHE`` (default **8**)
105 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
106 which buffers are to be transmitted must be associated to memory regions
107 (MRs). This is a slow operation that must be cached.
109 This value is always 1 for RX queues since they use a single MP.
111 Run-time configuration
112 ~~~~~~~~~~~~~~~~~~~~~~
114 - librte_pmd_mlx4 brings kernel network interfaces up during initialization
115 because it is affected by their state. Forcing them down prevents packets
118 - **ethtool** operations on related kernel interfaces also affect the PMD.
120 - ``port`` parameter [int]
122 This parameter provides a physical port to probe and can be specified multiple
123 times for additional ports. All ports are probed by default if left
126 Kernel module parameters
127 ~~~~~~~~~~~~~~~~~~~~~~~~
129 The **mlx4_core** kernel module has several parameters that affect the
130 behavior and/or the performance of librte_pmd_mlx4. Some of them are described
133 - **num_vfs** (integer or triplet, optionally prefixed by device address
136 Create the given number of VFs on the specified devices.
138 - **log_num_mgm_entry_size** (integer)
140 Device-managed flow steering (DMFS) is required by DPDK applications. It is
141 enabled by using a negative value, the last four bits of which have a
144 - **-1**: force device-managed flow steering (DMFS).
145 - **-7**: configure optimized steering mode to improve performance with the
146 following limitation: VLAN filtering is not supported with this mode.
147 This is the recommended mode in case VLAN filter is not needed.
152 This driver relies on external libraries and kernel drivers for resources
153 allocations and initialization. The following dependencies are not part of
154 DPDK and must be installed separately:
156 - **libibverbs** (provided by rdma-core package)
158 User space verbs framework used by librte_pmd_mlx4. This library provides
159 a generic interface between the kernel and low-level user space drivers
162 It allows slow and privileged operations (context initialization, hardware
163 resources allocations) to be managed by the kernel and fast operations to
164 never leave user space.
166 - **libmlx4** (provided by rdma-core package)
168 Low-level user space driver library for Mellanox ConnectX-3 devices,
169 it is automatically loaded by libibverbs.
171 This library basically implements send/receive calls to the hardware
176 They provide the kernel-side verbs API and low level device drivers that
177 manage actual hardware initialization and resources sharing with user
180 Unlike most other PMDs, these modules must remain loaded and bound to
183 - mlx4_core: hardware driver managing Mellanox ConnectX-3 devices.
184 - mlx4_en: Ethernet device driver that provides kernel network interfaces.
185 - mlx4_ib: InifiniBand device driver.
186 - ib_uverbs: user space driver for verbs (entry point for libibverbs).
188 - **Firmware update**
190 Mellanox OFED releases include firmware updates for ConnectX-3 adapters.
192 Because each release provides new features, these updates must be applied to
193 match the kernel modules and libraries they come with.
197 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
200 Depending on system constraints and user preferences either RDMA core library
201 with a recent enough Linux kernel release (recommended) or Mellanox OFED,
202 which provides compatibility with older releases.
204 Current RDMA core package and Linux kernel (recommended)
205 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
207 - Minimal Linux kernel version: 4.14.
208 - Minimal RDMA core version: v15 (see `RDMA core installation documentation`_).
210 .. _`RDMA core installation documentation`: https://raw.githubusercontent.com/linux-rdma/rdma-core/master/README.md
212 .. _Mellanox_OFED_as_a_fallback:
214 Mellanox OFED as a fallback
215 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
217 - `Mellanox OFED`_ version: **4.2**.
218 - firmware version: **2.42.5000** and above.
220 .. _`Mellanox OFED`: http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers
224 Several versions of Mellanox OFED are available. Installing the version
225 this DPDK release was developed and tested against is strongly
226 recommended. Please check the `prerequisites`_.
228 Installing Mellanox OFED
229 ^^^^^^^^^^^^^^^^^^^^^^^^
231 1. Download latest Mellanox OFED.
233 2. Install the required libraries and kernel modules either by installing
234 only the required set, or by installing the entire Mellanox OFED:
238 .. code-block:: console
240 ./mlnxofedinstall --dpdk --upstream-libs
242 For SR-IOV hypervisors use:
244 .. code-block:: console
246 ./mlnxofedinstall --dpdk --upstream-libs --enable-sriov --hypervisor
248 For SR-IOV virtual machine use:
250 .. code-block:: console
252 ./mlnxofedinstall --dpdk --upstream-libs --guest
254 3. Verify the firmware is the correct one:
256 .. code-block:: console
260 4. Set all ports links to Ethernet, follow instructions on the screen:
262 .. code-block:: console
266 5. Continue with :ref:`section 2 of the Quick Start Guide <QSG_2>`.
271 * Mellanox(R) ConnectX(R)-3 Pro 40G MCX354A-FCC_Ax (2*40G)
278 1. Set all ports links to Ethernet
280 .. code-block:: console
282 PCI=<NIC PCI address>
283 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port0"
284 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port1"
288 If using Mellanox OFED one can permanently set the port link
289 to Ethernet using connectx_port_config tool provided by it.
290 :ref:`Mellanox_OFED_as_a_fallback`:
294 2. In case of bare metal or hypervisor, configure optimized steering mode
295 by adding the following line to ``/etc/modprobe.d/mlx4_core.conf``:
297 .. code-block:: console
299 options mlx4_core log_num_mgm_entry_size=-7
303 If VLAN filtering is used, set log_num_mgm_entry_size=-1.
304 Performance degradation can occur on this case.
306 3. Restart the driver:
308 .. code-block:: console
310 /etc/init.d/openibd restart
314 .. code-block:: console
316 service openibd restart
318 4. Compile DPDK and you are ready to go. See instructions on
319 :ref:`Development Kit Build System <Development_Kit_Build_System>`
324 1. Verify the optimized steering mode is configured:
326 .. code-block:: console
328 cat /sys/module/mlx4_core/parameters/log_num_mgm_entry_size
330 2. Use the CPU near local NUMA node to which the PCIe adapter is connected,
331 for better performance. For VMs, verify that the right CPU
332 and NUMA node are pinned according to the above. Run:
334 .. code-block:: console
338 to identify the NUMA node to which the PCIe adapter is connected.
340 3. If more than one adapter is used, and root complex capabilities allow
341 to put both adapters on the same NUMA node without PCI bandwidth degradation,
342 it is recommended to locate both adapters on the same NUMA node.
343 This in order to forward packets from one to the other without
344 NUMA performance penalty.
346 4. Disable pause frames:
348 .. code-block:: console
350 ethtool -A <netdev> rx off tx off
352 5. Verify IO non-posted prefetch is disabled by default. This can be checked
353 via the BIOS configuration. Please contact you server provider for more
354 information about the settings.
358 On some machines, depends on the machine integrator, it is beneficial
359 to set the PCI max read request parameter to 1K. This can be
360 done in the following way:
362 To query the read request size use:
364 .. code-block:: console
366 setpci -s <NIC PCI address> 68.w
368 If the output is different than 3XXX, set it by:
370 .. code-block:: console
372 setpci -s <NIC PCI address> 68.w=3XXX
374 The XXX can be different on different systems. Make sure to configure
375 according to the setpci output.
380 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-3
381 devices managed by librte_pmd_mlx4.
383 #. Load the kernel modules:
385 .. code-block:: console
387 modprobe -a ib_uverbs mlx4_en mlx4_core mlx4_ib
389 Alternatively if MLNX_OFED is fully installed, the following script can
392 .. code-block:: console
394 /etc/init.d/openibd restart
398 User space I/O kernel modules (uio and igb_uio) are not used and do
399 not have to be loaded.
401 #. Make sure Ethernet interfaces are in working order and linked to kernel
402 verbs. Related sysfs entries should be present:
404 .. code-block:: console
406 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
410 .. code-block:: console
417 #. Optionally, retrieve their PCI bus addresses for whitelisting:
419 .. code-block:: console
422 for intf in eth2 eth3 eth4 eth5;
424 (cd "/sys/class/net/${intf}/device/" && pwd -P);
427 sed -n 's,.*/\(.*\),-w \1,p'
431 .. code-block:: console
440 There are only two distinct PCI bus addresses because the Mellanox
441 ConnectX-3 adapters installed on this system are dual port.
443 #. Request huge pages:
445 .. code-block:: console
447 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
449 #. Start testpmd with basic parameters:
451 .. code-block:: console
453 testpmd -l 8-15 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
457 .. code-block:: console
460 EAL: PCI device 0000:83:00.0 on NUMA socket 1
461 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
462 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_0" (VF: false)
463 PMD: librte_pmd_mlx4: 2 port(s) detected
464 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:b7:50
465 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:b7:51
466 EAL: PCI device 0000:84:00.0 on NUMA socket 1
467 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
468 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_1" (VF: false)
469 PMD: librte_pmd_mlx4: 2 port(s) detected
470 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:ba:b0
471 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:ba:b1
472 Interactive-mode selected
473 Configuring Port 0 (socket 0)
474 PMD: librte_pmd_mlx4: 0x867d60: TX queues number update: 0 -> 2
475 PMD: librte_pmd_mlx4: 0x867d60: RX queues number update: 0 -> 2
476 Port 0: 00:02:C9:B5:B7:50
477 Configuring Port 1 (socket 0)
478 PMD: librte_pmd_mlx4: 0x867da0: TX queues number update: 0 -> 2
479 PMD: librte_pmd_mlx4: 0x867da0: RX queues number update: 0 -> 2
480 Port 1: 00:02:C9:B5:B7:51
481 Configuring Port 2 (socket 0)
482 PMD: librte_pmd_mlx4: 0x867de0: TX queues number update: 0 -> 2
483 PMD: librte_pmd_mlx4: 0x867de0: RX queues number update: 0 -> 2
484 Port 2: 00:02:C9:B5:BA:B0
485 Configuring Port 3 (socket 0)
486 PMD: librte_pmd_mlx4: 0x867e20: TX queues number update: 0 -> 2
487 PMD: librte_pmd_mlx4: 0x867e20: RX queues number update: 0 -> 2
488 Port 3: 00:02:C9:B5:BA:B1
489 Checking link statuses...
490 Port 0 Link Up - speed 10000 Mbps - full-duplex
491 Port 1 Link Up - speed 40000 Mbps - full-duplex
492 Port 2 Link Up - speed 10000 Mbps - full-duplex
493 Port 3 Link Up - speed 40000 Mbps - full-duplex