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.
80 - Multi arch support: x86_64 and POWER8.
81 - Link state information is provided.
82 - Scattered packets are supported for TX and RX.
83 - Inner L3/L4 (IP, TCP and UDP) TX/RX checksum offloading and validation.
84 - Outer L3 (IP) TX/RX checksum offloading and validation for VXLAN frames.
93 These options can be modified in the ``.config`` file.
95 - ``CONFIG_RTE_LIBRTE_MLX4_PMD`` (default **n**)
97 Toggle compilation of librte_pmd_mlx4 itself.
99 - ``CONFIG_RTE_LIBRTE_MLX4_DEBUG`` (default **n**)
101 Toggle debugging code and stricter compilation flags. Enabling this option
102 adds additional run-time checks and debugging messages at the cost of
105 - ``CONFIG_RTE_LIBRTE_MLX4_DEBUG_BROKEN_VERBS`` (default **n**)
107 Mellanox OFED versions earlier than 4.2 may return false errors from
108 Verbs object destruction APIs after the device is plugged out.
109 Enabling this option replaces assertion checks that cause the program
110 to abort with harmless debugging messages as a workaround.
111 Relevant only when CONFIG_RTE_LIBRTE_MLX4_DEBUG is enabled.
113 - ``CONFIG_RTE_LIBRTE_MLX4_MAX_INLINE`` (default **0**)
115 Amount of data to be inlined during TX operations. Improves latency but
118 - ``CONFIG_RTE_LIBRTE_MLX4_TX_MP_CACHE`` (default **8**)
120 Maximum number of cached memory pools (MPs) per TX queue. Each MP from
121 which buffers are to be transmitted must be associated to memory regions
122 (MRs). This is a slow operation that must be cached.
124 This value is always 1 for RX queues since they use a single MP.
126 Environment variables
127 ~~~~~~~~~~~~~~~~~~~~~
129 - ``MLX4_INLINE_RECV_SIZE``
131 A nonzero value enables inline receive for packets up to that size. May
132 significantly improve performance in some cases but lower it in
133 others. Requires careful testing.
135 Run-time configuration
136 ~~~~~~~~~~~~~~~~~~~~~~
138 - librte_pmd_mlx4 brings kernel network interfaces up during initialization
139 because it is affected by their state. Forcing them down prevents packets
142 - **ethtool** operations on related kernel interfaces also affect the PMD.
144 - ``port`` parameter [int]
146 This parameter provides a physical port to probe and can be specified multiple
147 times for additional ports. All ports are probed by default if left
150 Kernel module parameters
151 ~~~~~~~~~~~~~~~~~~~~~~~~
153 The **mlx4_core** kernel module has several parameters that affect the
154 behavior and/or the performance of librte_pmd_mlx4. Some of them are described
157 - **num_vfs** (integer or triplet, optionally prefixed by device address
160 Create the given number of VFs on the specified devices.
162 - **log_num_mgm_entry_size** (integer)
164 Device-managed flow steering (DMFS) is required by DPDK applications. It is
165 enabled by using a negative value, the last four bits of which have a
168 - **-1**: force device-managed flow steering (DMFS).
169 - **-7**: configure optimized steering mode to improve performance with the
170 following limitation: VLAN filtering is not supported with this mode.
171 This is the recommended mode in case VLAN filter is not needed.
176 This driver relies on external libraries and kernel drivers for resources
177 allocations and initialization. The following dependencies are not part of
178 DPDK and must be installed separately:
182 User space verbs framework used by librte_pmd_mlx4. This library provides
183 a generic interface between the kernel and low-level user space drivers
186 It allows slow and privileged operations (context initialization, hardware
187 resources allocations) to be managed by the kernel and fast operations to
188 never leave user space.
192 Low-level user space driver library for Mellanox ConnectX-3 devices,
193 it is automatically loaded by libibverbs.
195 This library basically implements send/receive calls to the hardware
198 - **Kernel modules** (mlnx-ofed-kernel)
200 They provide the kernel-side verbs API and low level device drivers that
201 manage actual hardware initialization and resources sharing with user
204 Unlike most other PMDs, these modules must remain loaded and bound to
207 - mlx4_core: hardware driver managing Mellanox ConnectX-3 devices.
208 - mlx4_en: Ethernet device driver that provides kernel network interfaces.
209 - mlx4_ib: InifiniBand device driver.
210 - ib_uverbs: user space driver for verbs (entry point for libibverbs).
212 - **Firmware update**
214 Mellanox OFED releases include firmware updates for ConnectX-3 adapters.
216 Because each release provides new features, these updates must be applied to
217 match the kernel modules and libraries they come with.
221 Both libraries are BSD and GPL licensed. Linux kernel modules are GPL
224 Currently supported by DPDK:
226 - Mellanox OFED **4.1**.
227 - Firmware version **2.36.5000** and above.
229 Getting Mellanox OFED
230 ~~~~~~~~~~~~~~~~~~~~~
232 While these libraries and kernel modules are available on OpenFabrics
233 Alliance's `website <https://www.openfabrics.org/>`_ and provided by package
234 managers on most distributions, this PMD requires Ethernet extensions that
235 may not be supported at the moment (this is a work in progress).
238 <http://www.mellanox.com/page/products_dyn?product_family=26&mtag=linux_sw_drivers>`_
239 includes the necessary support and should be used in the meantime. For DPDK,
240 only libibverbs, libmlx4, mlnx-ofed-kernel packages and firmware updates are
241 required from that distribution.
245 Several versions of Mellanox OFED are available. Installing the version
246 this DPDK release was developed and tested against is strongly
247 recommended. Please check the `prerequisites`_.
252 * Mellanox(R) ConnectX(R)-3 Pro 40G MCX354A-FCC_Ax (2*40G)
257 1. Download latest Mellanox OFED. For more info check the `prerequisites`_.
259 2. Install the required libraries and kernel modules either by installing
260 only the required set, or by installing the entire Mellanox OFED:
264 .. code-block:: console
268 For SR-IOV hypervisors use:
270 .. code-block:: console
272 ./mlnxofedinstall --enable-sriov -hypervisor
274 For SR-IOV virtual machine use:
276 .. code-block:: console
278 ./mlnxofedinstall --guest
280 3. Verify the firmware is the correct one:
282 .. code-block:: console
286 4. Set all ports links to Ethernet, follow instructions on the screen:
288 .. code-block:: console
292 Or in the manual way:
294 .. code-block:: console
296 PCI=<NIC PCI address>
297 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port0"
298 echo eth > "/sys/bus/pci/devices/$PCI/mlx4_port1"
300 5. In case of bare metal or hypervisor, configure optimized steering mode
301 by adding the following line to ``/etc/modprobe.d/mlx4_core.conf``:
303 .. code-block:: console
305 options mlx4_core log_num_mgm_entry_size=-7
309 If VLAN filtering is used, set log_num_mgm_entry_size=-1.
310 Performance degradation can occur on this case.
312 6. Restart the driver:
314 .. code-block:: console
316 /etc/init.d/openibd restart
320 .. code-block:: console
322 service openibd restart
324 7. Compile DPDK and you are ready to go. See instructions on
325 :ref:`Development Kit Build System <Development_Kit_Build_System>`
330 1. Verify the optimized steering mode is configured:
332 .. code-block:: console
334 cat /sys/module/mlx4_core/parameters/log_num_mgm_entry_size
336 2. Use environment variable MLX4_INLINE_RECV_SIZE=64 to get maximum
337 performance for 64B messages.
339 3. Use the CPU near local NUMA node to which the PCIe adapter is connected,
340 for better performance. For VMs, verify that the right CPU
341 and NUMA node are pinned according to the above. Run:
343 .. code-block:: console
347 to identify the NUMA node to which the PCIe adapter is connected.
349 4. If more than one adapter is used, and root complex capabilities allow
350 to put both adapters on the same NUMA node without PCI bandwidth degradation,
351 it is recommended to locate both adapters on the same NUMA node.
352 This in order to forward packets from one to the other without
353 NUMA performance penalty.
355 5. Disable pause frames:
357 .. code-block:: console
359 ethtool -A <netdev> rx off tx off
361 6. Verify IO non-posted prefetch is disabled by default. This can be checked
362 via the BIOS configuration. Please contact you server provider for more
363 information about the settings.
367 On some machines, depends on the machine integrator, it is beneficial
368 to set the PCI max read request parameter to 1K. This can be
369 done in the following way:
371 To query the read request size use:
373 .. code-block:: console
375 setpci -s <NIC PCI address> 68.w
377 If the output is different than 3XXX, set it by:
379 .. code-block:: console
381 setpci -s <NIC PCI address> 68.w=3XXX
383 The XXX can be different on different systems. Make sure to configure
384 according to the setpci output.
389 This section demonstrates how to launch **testpmd** with Mellanox ConnectX-3
390 devices managed by librte_pmd_mlx4.
392 #. Load the kernel modules:
394 .. code-block:: console
396 modprobe -a ib_uverbs mlx4_en mlx4_core mlx4_ib
398 Alternatively if MLNX_OFED is fully installed, the following script can
401 .. code-block:: console
403 /etc/init.d/openibd restart
407 User space I/O kernel modules (uio and igb_uio) are not used and do
408 not have to be loaded.
410 #. Make sure Ethernet interfaces are in working order and linked to kernel
411 verbs. Related sysfs entries should be present:
413 .. code-block:: console
415 ls -d /sys/class/net/*/device/infiniband_verbs/uverbs* | cut -d / -f 5
419 .. code-block:: console
426 #. Optionally, retrieve their PCI bus addresses for whitelisting:
428 .. code-block:: console
431 for intf in eth2 eth3 eth4 eth5;
433 (cd "/sys/class/net/${intf}/device/" && pwd -P);
436 sed -n 's,.*/\(.*\),-w \1,p'
440 .. code-block:: console
449 There are only two distinct PCI bus addresses because the Mellanox
450 ConnectX-3 adapters installed on this system are dual port.
452 #. Request huge pages:
454 .. code-block:: console
456 echo 1024 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages/nr_hugepages
458 #. Start testpmd with basic parameters:
460 .. code-block:: console
462 testpmd -l 8-15 -n 4 -w 0000:83:00.0 -w 0000:84:00.0 -- --rxq=2 --txq=2 -i
466 .. code-block:: console
469 EAL: PCI device 0000:83:00.0 on NUMA socket 1
470 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
471 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_0" (VF: false)
472 PMD: librte_pmd_mlx4: 2 port(s) detected
473 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:b7:50
474 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:b7:51
475 EAL: PCI device 0000:84:00.0 on NUMA socket 1
476 EAL: probe driver: 15b3:1007 librte_pmd_mlx4
477 PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_1" (VF: false)
478 PMD: librte_pmd_mlx4: 2 port(s) detected
479 PMD: librte_pmd_mlx4: port 1 MAC address is 00:02:c9:b5:ba:b0
480 PMD: librte_pmd_mlx4: port 2 MAC address is 00:02:c9:b5:ba:b1
481 Interactive-mode selected
482 Configuring Port 0 (socket 0)
483 PMD: librte_pmd_mlx4: 0x867d60: TX queues number update: 0 -> 2
484 PMD: librte_pmd_mlx4: 0x867d60: RX queues number update: 0 -> 2
485 Port 0: 00:02:C9:B5:B7:50
486 Configuring Port 1 (socket 0)
487 PMD: librte_pmd_mlx4: 0x867da0: TX queues number update: 0 -> 2
488 PMD: librte_pmd_mlx4: 0x867da0: RX queues number update: 0 -> 2
489 Port 1: 00:02:C9:B5:B7:51
490 Configuring Port 2 (socket 0)
491 PMD: librte_pmd_mlx4: 0x867de0: TX queues number update: 0 -> 2
492 PMD: librte_pmd_mlx4: 0x867de0: RX queues number update: 0 -> 2
493 Port 2: 00:02:C9:B5:BA:B0
494 Configuring Port 3 (socket 0)
495 PMD: librte_pmd_mlx4: 0x867e20: TX queues number update: 0 -> 2
496 PMD: librte_pmd_mlx4: 0x867e20: RX queues number update: 0 -> 2
497 Port 3: 00:02:C9:B5:BA:B1
498 Checking link statuses...
499 Port 0 Link Up - speed 10000 Mbps - full-duplex
500 Port 1 Link Up - speed 40000 Mbps - full-duplex
501 Port 2 Link Up - speed 10000 Mbps - full-duplex
502 Port 3 Link Up - speed 40000 Mbps - full-duplex