4 .. **Read this first.**
6 The text in the sections below explains how to update the release notes.
8 Use proper spelling, capitalization and punctuation in all sections.
10 Variable and config names should be quoted as fixed width text:
13 Build the docs and view the output file to ensure the changes are correct::
17 xdg-open build/doc/html/guides/rel_notes/release_17_11.html
23 .. This section should contain new features added in this release. Sample
26 * **Add a title in the past tense with a full stop.**
28 Add a short 1-2 sentence description in the past tense. The description
29 should be enough to allow someone scanning the release notes to
30 understand the new feature.
32 If the feature adds a lot of sub-features you can use a bullet list like
35 * Added feature foo to do something.
36 * Enhanced feature bar to do something else.
38 Refer to the previous release notes for examples.
40 This section is a comment. do not overwrite or remove it.
41 Also, make sure to start the actual text at the margin.
42 =========================================================
44 * **Extended port_id range from uint8_t to uint16_t.**
46 Increased port_id range from 8 bits to 16 bits in order to support more than
47 256 ports in dpdk. All ethdev APIs which have port_id as parameter are changed
50 * **Modified the return type of rte_eth_stats_reset.**
52 Changed return type of ``rte_eth_stats_reset`` from ``void`` to ``int``
53 so the caller may know whether a device supports the operation or not
54 and if the operation was carried out.
56 * **Added a new driver for Marvell Armada 7k/8k devices.**
58 Added the new mrvl net driver for Marvell Armada 7k/8k devices. See the
59 "Network Interface Controller Drivers" document for more details on this new
62 * **Updated mlx4 driver.**
64 Updated the mlx4 driver including the following changes:
66 * Isolated mode (rte_flow) can now be enabled anytime, not only during
67 initial device configuration.
68 * Flow rules now support up to 4096 priority levels usable at will by
70 * Enhanced error message to help debugging invalid/unsupported flow rules.
71 * Flow rules matching all multicast and promiscuous traffic are now allowed.
72 * No more software restrictions on flow rules with the RSS action, their
73 configuration is much more flexible.
74 * Significantly reduced memory footprint for Rx and Tx queue objects.
75 * While supported, UDP RSS is temporarily disabled due to a remaining issue
76 with its support in the Linux kernel.
77 * The new RSS implementation does not automatically spread traffic according
78 to the inner packet of VXLAN frames anymore, only the outer one (like
80 * Partial (Tx only) support for secondary processes was broken and had to be
82 * Refactored driver to get rid of dependency on the components provided by
83 Mellanox OFED and instead rely on the current and public rdma-core
84 package and Linux version from now on.
85 * Removed compile-time limitation on number of device instances the PMD
88 * **Updated mlx5 driver.**
90 Updated the mlx5 driver including the following changes:
92 * Enabled PMD to run on top of upstream linux kernel and rdma-core libs.
93 By that removed the dependency on specific Mellanox OFED libraries.
94 * Improved PMD latency performance.
95 * Improved PMD memory footprint.
96 * Supported vectorized Rx/Tx burst for ARMv8.
97 * Supported secondary process.
98 * Supported flow counters.
99 * Supported Rx hardware timestamp offload.
100 * Supported device removal event.
102 * **Added SoftNIC PMD.**
104 Added new SoftNIC PMD. This virtual device offers applications a software
105 fallback support for traffic management.
107 * **Added support for NXP DPAA Devices.**
109 Added support for NXP's DPAA devices - LS104x series. This includes:
112 * DPAA Mempool driver for supporting offloaded packet memory pool
113 * DPAA PMD for DPAA devices
115 See the "Network Interface Controller Drivers" document for more details of
118 * **Updated support for Cavium OCTEONTX Device.**
120 Updated support for Cavium's OCTEONTX device(CN83xx). This includes:
122 * OCTEONTX Mempool driver for supporting offloaded packet memory pool
123 * OCTEONTX Ethdev PMD
124 * OCTEONTX Eventdev-Ethdev Rx adapter
126 See the "Network Interface Controller Drivers" document for more details of
129 * **nfp: Added PF support.**
131 Previously Netronome's NFP PMD had just support for VFs. PF support is
132 just as a basic DPDK port and has no VF management yet.
134 PF support comes with firmware upload support which allows the PMD to
135 independently work from kernel netdev NFP drivers.
137 NFP 4000 devices are also now supported along with previous 6000 devices.
139 * **Updated bnxt PMD.**
141 Major enhancements include:
143 * Support for Flow API
144 * Support for Tx and Rx descriptor status functions
146 * **Add bus agnostic functions to cryptodev for PMD initialisation**
148 Adds new PMD assist functions ``rte_cryptodev_pmd_parse_input_args()``,
149 ``rte_cryptodev_pmd_create()`` and ``rte_cryptodev_pmd_destroy()`` which
150 are bus independent for driver to manage creation and destruction of new
153 * **Updated QAT crypto PMD.**
155 Performance enhancements:
157 * Removed atomics from the internal queue pair structure.
158 * Coalesce writes to HEAD CSR on response processing.
159 * Coalesce writes to TAIL CSR on request processing.
161 Additional support for:
165 * **Updated the AESNI MB PMD.**
167 The AESNI MB PMD has been updated with additional support for:
170 * DES DOCSIS BPI algorithm.
172 This requires the IPSec Multi-buffer library 0.47. For more details,
173 check out the AESNI MB PMD documenation.
175 * **Updated the OpenSSL PMD.**
177 The OpenSSL PMD has been updated with additional support for:
182 * **Added NXP DPAA SEC crypto PMD.**
184 A new "dpaa_sec" hardware based crypto PMD for NXP DPAA devices has been
185 added. See the "Crypto Device Drivers" document for more details on this
188 * **Added MRVL crypto PMD.**
190 A new crypto PMD has been added, which provides several ciphering and hashing
191 algorithms. All cryptography operations use the MUSDK library crypto API.
193 * **Add new benchmarking mode to dpdk-test-crypto-perf application.**
195 Added new "PMD cyclecount" benchmark mode to dpdk-test-crypto-perf application
196 that displays more detailed breakdown of CPU cycles used by hardware
199 * **Added the Security Offload Library.**
201 Added an experimental library - rte_security. It provide security APIs for
202 protocols like IPSec using inline ipsec offload to ethernet device or full
203 protocol offload with lookaside crypto device.
205 See the "Security_Library" section of the DPDK Programmers Guide document,
206 for more information.
208 * **Updated DPAA2_SEC crypto driver.**
210 Updated dpaa2_sec crypto PMD to support rte_security lookaside protocol
213 * **Updated IXGBE ethernet driver.**
215 Updated ixgbe ethernet PMD to support rte_security inline IPSec offload.
217 * **Updated ipsec-secgw application**
219 Updated ipsec-secgw sample application to support rte_security actions for
220 ipsec inline and full protocol offload using lookaside crypto offload.
222 * **Added IOMMU support to libvhost-user**
224 Implemented device IOTLB in Vhost-user backend, and enabled Virtio's IOMMU
225 feature. The feature is disabled by default, and can be enabled by setting
226 RTE_VHOST_USER_IOMMU_SUPPORT flag at vhost device registration time.
228 * **Added the Event Ethernet Adapter Library.**
230 Added the Event Ethernet Adapter library. It provices APIs for
231 eventdev applications to configure the ethdev to eventdev packet flow.
233 * **Updated DPAA2 Event PMD.**
235 Added support for eventdev ethernet adapter for DPAA2.
237 * **Added Membership library (rte_member).**
239 Added membership library. It provides an API for DPDK applications to insert a
240 new member, delete an existing member, or query the existence of a member in a
241 given set, or a group of sets. For the case of a group of sets the library
242 will return not only whether the element has been inserted before in one of
243 the sets but also which set it belongs to.
245 The Membership Library is an extension and generalization of a traditional
246 filter (for example Bloom Filter) structure that has multiple usages in a wide
247 variety of workloads and applications. In general, the Membership Library is a
248 data structure that provides a “set-summary” and responds to set-membership
249 queries whether a certain member belongs to a set(s).
251 See the :ref:`Membership Library <Member_Library>` documentation in
252 the Programmers Guide document, for more information.
254 * **Added the Generic Segmentation Offload Library.**
256 Added the Generic Segmentation Offload (GSO) library to enable
257 applications to split large packets (e.g. MTU is 64KB) into small
258 ones (e.g. MTU is 1500B). Supported packet types are:
261 * VxLAN packets, which must have an outer IPv4 header, and contain
262 an inner TCP/IPv4 packet.
263 * GRE packets, which must contain an outer IPv4 header, and inner
266 The GSO library doesn't check if the input packets have correct
267 checksums, and doesn't update checksums for output packets.
268 Additionally, the GSO library doesn't process IP fragmented packets.
270 * **Added the Flow Classification Library.**
272 Added the Flow Classification library, it provides an API for DPDK
273 applications to classify an input packet by matching it against a set of flow
274 rules. It uses the librte_table API to manage the flow rules.
280 .. This section should contain bug fixes added to the relevant
281 sections. Sample format:
283 * **code/section Fixed issue in the past tense with a full stop.**
285 Add a short 1-2 sentence description of the resolved issue in the past
288 The title should contain the code/lib section like a commit message.
290 Add the entries in alphabetic order in the relevant sections below.
292 This section is a comment. do not overwrite or remove it.
293 Also, make sure to start the actual text at the margin.
294 =========================================================
300 * **Service core fails to call service callback due to atomic lock**
302 In a specific configuration of multi-thread unsafe services and service
303 cores, a service core previously did not correctly release the atomic lock
304 on the service. This would result in the cores polling the service, but it
305 looked like another thread was executing the service callback. The logic for
306 atomic locking of the services has been fixed and refactored for readability.
327 .. This section should contain new known issues in this release. Sample format:
329 * **Add title in present tense with full stop.**
331 Add a short 1-2 sentence description of the known issue in the present
332 tense. Add information on any known workarounds.
334 This section is a comment. do not overwrite or remove it.
335 Also, make sure to start the actual text at the margin.
336 =========================================================
342 .. This section should contain API changes. Sample format:
344 * Add a short 1-2 sentence description of the API change. Use fixed width
345 quotes for ``rte_function_names`` or ``rte_struct_names``. Use the past
348 This section is a comment. do not overwrite or remove it.
349 Also, make sure to start the actual text at the margin.
350 =========================================================
352 * **Ethdev device name length increased**
354 The size of internal device name is increased to 64 characters
355 to allow for storing longer bus specific name.
357 * **Ethdev flag RTE_ETH_DEV_DETACHABLE was removed**
359 This flag is not necessary anymore, with the new hotplug implementation.
360 It is now removed from the ether library. Its semantic is expressed at the bus
363 * **Service cores API updated for usability**
365 The service cores API has been changed, removing pointers from the API
366 where possible, instead using integer IDs to identify each service. This
367 simplifed application code, aids debugging, and provides better
368 encapsulation. A summary of the main changes made is as follows:
370 * Services identified by ID not by ``rte_service_spec`` pointer
371 * Reduced API surface by using ``set`` functions instead of enable/disable
372 * Reworked ``rte_service_register`` to provide the service ID to registrar
373 * Rework start and stop APIs into ``rte_service_runstate_set``
374 * Added API to set runstate of service implementation to indicate readyness
376 * **The following changes made in mempool library**
378 * Moved ``flags`` datatype from int to unsigned int for ``rte_mempool``.
379 * Removed ``__rte_unused int flag`` param from ``rte_mempool_generic_put``
380 and ``rte_mempool_generic_get`` API.
381 * Added ``flags`` param in ``rte_mempool_xmem_size`` and
382 ``rte_mempool_xmem_usage``.
384 * ``rte_mem_phy2mch`` was used in Xen dom0 to obtain the physical address;
385 remove this API as Xen dom0 support was removed.
387 * **Some data type, structure members and functions related to physical address
388 are deprecated and have new alias with IOVA wording.**
390 * ``phys_addr_t`` can be often replaced by ``rte_iova_t`` of same size.
391 * ``RTE_BAD_PHYS_ADDR`` is often replaced by ``RTE_BAD_IOVA`` of same value.
392 * ``rte_memseg.phys_addr`` is aliased with ``rte_memseg.iova_addr``.
393 * ``rte_mem_virt2phy()`` can often be replaced by ``rte_mem_virt2iova``.
394 * ``rte_malloc_virt2phy`` is aliased with ``rte_malloc_virt2iova``.
395 * ``rte_memzone.phys_addr`` is aliased with ``rte_memzone.iova``.
396 * ``rte_mempool_objhdr.physaddr`` is aliased with ``rte_mempool_objhdr.iova``.
397 * ``rte_mempool_memhdr.phys_addr`` is aliased with ``rte_mempool_memhdr.iova``.
398 * ``rte_mempool_virt2phy()`` can be replaced by ``rte_mempool_virt2iova()``.
399 * ``rte_mempool_populate_phys*()`` are aliased with ``rte_mempool_populate_iova*()``
400 * ``rte_mbuf.buf_physaddr`` is aliased with ``rte_mbuf.buf_iova``.
401 * ``rte_mbuf_data_dma_addr*()`` are aliased with ``rte_mbuf_data_iova*()``.
402 * ``rte_pktmbuf_mtophys*`` are aliased with ``rte_pktmbuf_iova*()``.
404 * **PCI bus API moved outside of the EAL**
406 The PCI bus previously implemented within the EAL has been moved.
407 A first part has been added as an RTE library providing PCI helpers to
408 parse device locations or other such utilities.
409 A second part consisting in the actual bus driver has been moved to its
410 proper subdirectory, without changing its functionalities.
412 As such, several PCI-related functions are not proposed by the EAL anymore:
417 * rte_pci_ioport_read
418 * rte_pci_ioport_unmap
419 * rte_pci_ioport_write
423 * rte_pci_read_config
426 * rte_pci_unmap_device
428 * rte_pci_write_config
430 These functions are made available either as part of ``librte_pci`` or
433 * **Moved vdev bus APIs outside of the EAL**
435 Moved the following APIs from ``librte_eal`` to ``librte_bus_vdev``:
437 * ``rte_vdev_register``
438 * ``rte_vdev_uninit``
439 * ``rte_vdev_unregister``
441 * **Add return value to stats_get dev op API**
443 The ``stats_get`` dev op API return value has been changed to be int.
444 By this way PMDs can return an error value in case of failure at stats
445 getting process time.
447 * **Modified the rte_cryptodev_allocate_driver function in the cryptodev library.**
449 The function ``rte_cryptodev_allocate_driver()`` has been modified.
450 An extra parameter ``struct cryptodev_driver *crypto_drv`` has been added.
452 * **Removed virtual device bus specific functions from librte_cryptodev.**
454 The functions ``rte_cryptodev_vdev_parse_init_params()`` and
455 ``rte_cryptodev_vdev_pmd_init()`` have been removed from librte_cryptodev
456 and have been replaced by non bus specific functions
457 ``rte_cryptodev_pmd_parse_input_args()`` and ``rte_cryptodev_pmd_create()``.
459 * ``rte_cryptodev_create_vdev`` was removed to avoid the dependency on vdev
460 in librte_cryptodev; instead, users can call rte_vdev_init() directly.
462 * **Removed PCI device bus specific functions from librte_cryptodev.**
464 The functions ``rte_cryptodev_pci_generic_probe()`` and
465 ``rte_cryptodev_pci_generic_remove()`` have been removed from librte_cryptodev
466 and have been replaced by non bus specific functions
467 ``rte_cryptodev_pmd_create()`` and ``rte_cryptodev_pmd_destroy()``.
469 * **Removed deprecated functions to manage log level or type.**
471 The functions ``rte_set_log_level()``, ``rte_get_log_level()``,
472 ``rte_set_log_type()`` and ``rte_get_log_type()`` have been removed.
473 They are respectively replaced by ``rte_log_set_global_level()``,
474 ``rte_log_get_global_level()``, ``rte_log_set_level()`` and
475 ``rte_log_get_level()``.
477 * **Removed ``mbuf`` flags ``PKT_RX_VLAN_PKT`` and ``PKT_RX_QINQ_PKT``.**
479 The ``mbuf`` flags ``PKT_RX_VLAN_PKT`` and ``PKT_RX_QINQ_PKT`` have
480 been removed since their behavior were not properly described.
482 * **Added ``mbuf`` flags ``PKT_RX_VLAN`` and ``PKT_RX_QINQ``.**
484 Two ``mbuf`` flags have been added to indicate that the VLAN
485 identifier has been saved in in the ``mbuf`` structure. For instance:
487 - if VLAN is not stripped and TCI is saved: ``PKT_RX_VLAN``
488 - if VLAN is stripped and TCI is saved: ``PKT_RX_VLAN | PKT_RX_VLAN_STRIPPED``
490 * **Modified the vlan_offload_set_t function prototype in the ethdev library.**
492 Changed the function prototype of ``vlan_offload_set_t``. The return value
493 has been changed from ``void`` to ``int`` so the caller to knows whether
494 the backing device supports the operation or if the operation was
495 successfully performed.
501 .. This section should contain ABI changes. Sample format:
503 * Add a short 1-2 sentence description of the ABI change that was announced
504 in the previous releases and made in this release. Use fixed width quotes
505 for ``rte_function_names`` or ``rte_struct_names``. Use the past tense.
507 This section is a comment. do not overwrite or remove it.
508 Also, make sure to start the actual text at the margin.
509 =========================================================
511 * **Extended port_id range.**
513 The size of the field ``port_id`` in the ``rte_eth_dev_data`` structure
514 changed, as described in the `New Features` section.
516 * **New parameter added to rte_eth_dev.**
518 New parameter ``security_ctx`` added to ``rte_eth_dev`` to support security
519 operations like IPSec inline.
521 * **New parameter added to rte_cryptodev.**
523 New parameter ``security_ctx`` added to ``rte_cryptodev`` to support security
524 operations like lookaside crypto.
530 .. This section should contain removed items in this release. Sample format:
532 * Add a short 1-2 sentence description of the removed item in the past
535 This section is a comment. do not overwrite or remove it.
536 Also, make sure to start the actual text at the margin.
537 =========================================================
539 * Xen dom0 in EAL was removed, as well as xenvirt PMD and vhost_xen.
541 * The crypto performance unit tests have been removed,
542 replaced by the dpdk-test-crypto-perf application.
545 Shared Library Versions
546 -----------------------
548 .. Update any library version updated in this release and prepend with a ``+``
552 + librte_cfgfile.so.2
555 This section is a comment. do not overwrite or remove it.
556 =========================================================
559 The libraries prepended with a plus sign were incremented in this version.
564 + librte_bitratestats.so.2
565 + librte_bus_vdev.so.1
568 + librte_cryptodev.so.4
569 librte_distributor.so.1
572 + librte_eventdev.so.3
573 + librte_flow_classify.so.1
581 librte_latencystats.so.1
584 + librte_mempool.so.3
591 + librte_pmd_bnxt.so.2
592 + librte_pmd_bond.so.2
593 + librte_pmd_i40e.so.2
594 + librte_pmd_ixgbe.so.2
596 + librte_pmd_softnic.so.1
597 + librte_pmd_vhost.so.2
603 + librte_security.so.1
612 .. This section should contain a list of platforms that were tested with this
617 * <vendor> platform with <vendor> <type of devices> combinations
622 * Other relevant details...
624 This section is a comment. do not overwrite or remove it.
625 Also, make sure to start the actual text at the margin.
626 =========================================================
628 * Intel(R) platforms with Intel(R) NICs combinations
632 * Intel(R) Atom(TM) CPU C2758 @ 2.40GHz
633 * Intel(R) Xeon(R) CPU D-1540 @ 2.00GHz
634 * Intel(R) Xeon(R) CPU D-1541 @ 2.10GHz
635 * Intel(R) Xeon(R) CPU E5-4667 v3 @ 2.00GHz
636 * Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz
637 * Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz
638 * Intel(R) Xeon(R) CPU E5-2695 v4 @ 2.10GHz
639 * Intel(R) Xeon(R) CPU E5-2658 v2 @ 2.40GHz
640 * Intel(R) Xeon(R) CPU E5-2658 v3 @ 2.20GHz
648 * Red Hat Enterprise Linux Server release 7.3
649 * SUSE Enterprise Linux 12
656 * Intel(R) 82599ES 10 Gigabit Ethernet Controller
658 * Firmware version: 0x61bf0001
659 * Device id (pf/vf): 8086:10fb / 8086:10ed
660 * Driver version: 5.2.3 (ixgbe)
662 * Intel(R) Corporation Ethernet Connection X552/X557-AT 10GBASE-T
664 * Firmware version: 0x800003e7
665 * Device id (pf/vf): 8086:15ad / 8086:15a8
666 * Driver version: 4.4.6 (ixgbe)
668 * Intel(R) Ethernet Converged Network Adapter X710-DA4 (4x10G)
670 * Firmware version: 6.01 0x80003205
671 * Device id (pf/vf): 8086:1572 / 8086:154c
672 * Driver version: 2.1.26 (i40e)
674 * Intel(R) Ethernet Converged Network Adapter X710-DA2 (2x10G)
676 * Firmware version: 6.01 0x80003204
677 * Device id (pf/vf): 8086:1572 / 8086:154c
678 * Driver version: 2.1.26 (i40e)
680 * Intel(R) Ethernet Converged Network Adapter XXV710-DA2 (2x25G)
682 * Firmware version: 6.01 0x80003221
683 * Device id (pf/vf): 8086:158b
684 * Driver version: 2.1.26 (i40e)
686 * Intel(R) Ethernet Converged Network Adapter XL710-QDA2 (2X40G)
688 * Firmware version: 6.01 0x8000321c
689 * Device id (pf/vf): 8086:1583 / 8086:154c
690 * Driver version: 2.1.26 (i40e)
692 * Intel(R) Corporation I350 Gigabit Network Connection
694 * Firmware version: 1.63, 0x80000dda
695 * Device id (pf/vf): 8086:1521 / 8086:1520
696 * Driver version: 5.3.0-k (igb)