1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright 2020 The DPDK contributors
4 .. include:: <isonum.txt>
9 .. **Read this first.**
11 The text in the sections below explains how to update the release notes.
13 Use proper spelling, capitalization and punctuation in all sections.
15 Variable and config names should be quoted as fixed width text:
18 Build the docs and view the output file to ensure the changes are correct::
21 xdg-open build/doc/html/guides/rel_notes/release_20_11.html
27 .. This section should contain new features added in this release.
30 * **Add a title in the past tense with a full stop.**
32 Add a short 1-2 sentence description in the past tense.
33 The description should be enough to allow someone scanning
34 the release notes to understand the new feature.
36 If the feature adds a lot of sub-features you can use a bullet list
39 * Added feature foo to do something.
40 * Enhanced feature bar to do something else.
42 Refer to the previous release notes for examples.
44 Suggested order in release notes items:
45 * Core libs (EAL, mempool, ring, mbuf, buses)
46 * Device abstraction libs and PMDs
48 - cryptodev (lib, PMDs)
49 - eventdev (lib, PMDs)
52 * Apps, Examples, Tools (if significant)
54 This section is a comment. Do not overwrite or remove it.
55 Also, make sure to start the actual text at the margin.
56 =======================================================
58 * **Added write combining store APIs.**
60 Added ``rte_write32_wc`` and ``rte_write32_wc_relaxed`` APIs
61 that enable write combining stores (depending on architecture).
62 The functions are provided as a generic stub and
63 x86 specific implementation.
65 * **Added prefetch with intention to write APIs.**
67 Added new prefetch function variants e.g. ``rte_prefetch0_write``,
68 which allow the programmer to prefetch a cache line and also indicate
69 the intention to write.
71 * **Added the rte_cldemote API.**
73 Added a hardware hint CLDEMOTE, which is similar to prefetch in reverse.
74 CLDEMOTE moves the cache line to the more remote cache, where it expects
75 sharing to be efficient. Moving the cache line to a level more distant from
76 the processor helps to accelerate core-to-core communication.
77 This API is specific to x86 and implemented as a stub for other
80 * **Added support for limiting maximum SIMD bitwidth.**
82 Added a new EAL config setting ``max_simd_bitwidth`` to limit the vector
83 path selection at runtime. This value can be set by apps using the
84 ``rte_vect_set_max_simd_bitwidth`` function, or by the user with EAL flag
85 ``--force-max-simd-bitwidth``.
87 * **Added zero copy APIs for rte_ring.**
89 For rings with producer/consumer in ``RTE_RING_SYNC_ST``, ``RTE_RING_SYNC_MT_HTS``
90 modes, these APIs split enqueue/dequeue operation into three phases
91 (enqueue/dequeue start, copy data to/from ring, enqueue/dequeue finish).
92 Along with the advantages of the peek APIs, these provide the ability to
93 copy the data to the ring memory directly without the need for temporary
96 * **Updated CRC modules of the net library.**
98 * Added runtime selection of the optimal architecture-specific CRC path.
99 * Added optimized implementations of CRC32-Ethernet and CRC16-CCITT
100 using the AVX512 and VPCLMULQDQ instruction sets.
102 * **Introduced extended buffer description for receiving.**
104 Added the extended Rx buffer description for Rx queue setup routine
105 providing the individual settings for each Rx segment with maximal size,
106 buffer offset and memory pool to allocate data buffers from.
108 * **Added the FEC API, for a generic FEC query and config.**
110 Added the FEC API which provides functions for query FEC capabilities and
111 current FEC mode from device. An API for configuring FEC mode is also provided.
113 * **Added thread safety to rte_flow functions.**
115 Added the ``RTE_ETH_DEV_FLOW_OPS_THREAD_SAFE`` device flag to indicate
116 whether a PMD supports thread safe operations. If the PMD doesn't set the flag,
117 the rte_flow API level functions will protect the flow operations with a mutex.
119 * **Added flow-based traffic sampling support.**
121 Added a new action ``RTE_FLOW_ACTION_TYPE_SAMPLE`` that will sample the
122 incoming traffic and send a duplicated traffic with the specified ratio to
123 the application, while the original packet will continue to the target
126 The packets sampling is '1/ratio'. A ratio value set to 1 means that the
127 packets will be completely mirrored. The sample packet can be assigned with
128 a different set of actions than the original packet.
130 * **Added support of shared action in flow API.**
132 Added shared action support to use single flow actions in multiple flow
133 rules. An update to the shared action configuration alters the behavior of all
136 * Added a new action: ``RTE_FLOW_ACTION_TYPE_SHARED`` to use shared action
138 * Added new flow APIs to create/update/destroy/query shared actions.
140 * **Added support to flow rules to allow private PMD items/actions.**
142 * Flow rule verification has been updated to accept private PMD
145 * **Added a generic API to offload tunneled traffic and restore missed packets.**
147 * Added a new hardware independent helper to the flow API that
148 offloads tunneled traffic and restores missed packets.
150 * **Updated the ethdev library to support hairpin between two ports.**
152 New APIs have been introduced to support binding / unbinding of 2 ports in a
153 hairpin configuration. The hairpin Tx part flow rules can be inserted
154 explicitly. A new API has been added to get the hairpin peer ports list.
156 * **Updated the Amazon ena driver.**
158 Updated the ena PMD with new features and improvements, including:
160 * Added network interface metrics which can be read using xstats.
162 * **Updated Broadcom bnxt driver.**
164 Updated the Broadcom bnxt driver with new features and improvements, including:
166 * Added support for 200G PAM4 link speed.
167 * Added support for RSS hash level selection.
168 * Updated HWRM structures to 1.10.1.70 version.
169 * Added TRUFLOW support for Stingray devices.
170 * Added support for representors on MAIA cores of SR.
171 * Added support for VXLAN decap offload using rte_flow.
172 * Added support to indicate native rte_flow API thread safety.
174 * **Updated Cisco enic driver.**
176 * Added support for VF representors with single-queue Tx/Rx and flow API
177 * Added support for egress PORT_ID action
178 * Added support for non-zero priorities for group 0 flows
179 * Added support for VXLAN decap combined with VLAN pop
181 * **Added hns3 FEC PMD, for supporting query and config FEC mode.**
183 Added the FEC PMD which provides functions for querying FEC capabilities and
184 current FEC mode from a device. A PMD for configuring FEC mode is also provided.
186 * **Updated the Intel iavf driver.**
188 Updated the iavf PMD with new features and improvements, including:
190 * Added support for flexible descriptor metadata extraction.
191 * Added support for outer IP hash of GTPC and GTPU.
192 * Added support of AVX512 instructions in Rx and Tx path.
193 * Added support for up to 256 queue pairs per VF.
195 * **Updated Intel ixgbe driver.**
197 Updated the Intel ixgbe driver to use write combining stores.
199 * **Updated Intel i40e driver.**
201 Updated the Intel i40e driver to use write combining stores.
203 * **Updated Intel ice driver.**
205 * Added support for write combining stores.
206 * Added ACL filter support for the Intel DCF.
208 * **Updated Mellanox mlx5 driver.**
210 Updated the Mellanox mlx5 driver with new features and improvements, including:
212 * Added vectorized Multi-Packet Rx Queue burst.
213 * Added support for 2 new miniCQE formats: Flow Tag and L3/L4 header.
214 * Added support for PMD level multiple-thread flow insertion.
215 * Added support for matching on fragmented/non-fragmented IPv4/IPv6 packets.
216 * Added support for QinQ packets matching.
217 * Added support for the new VLAN fields ``has_vlan`` in the Ethernet item
218 and ``has_more_vlan`` in the VLAN item.
219 * Updated the supported timeout for Age action to the maximal value supported
221 * Added support for Age action query.
222 * Added support for multi-ports hairpin.
223 * Allow unknown link speed.
225 Updated Mellanox mlx5 vDPA driver:
227 * Added support of vDPA VirtQ error handling.
229 * **Updated Solarflare network PMD.**
231 Updated the Solarflare ``sfc_efx`` driver with changes including:
233 * Added SR-IOV PF support
234 * Added Alveo SN1000 SmartNICs (EF100 architecture) support including
235 flow API transfer rules for switch HW offload
236 * Added ARMv8 support
237 * Claimed flow API native thread safety
239 * **Added Wangxun txgbe PMD.**
241 Added a new PMD driver for Wangxun 10 Gigabit Ethernet NICs.
243 See the :doc:`../nics/txgbe` for more details.
245 * **Updated Virtio driver.**
247 * Added support for Vhost-vDPA backend to the Virtio-user PMD.
248 * Changed default link speed to unknown.
249 * Added support for the 200G link speed.
251 * **Updated Memif PMD.**
253 * Added support for abstract socket addresses.
254 * Changed default socket address type to abstract.
256 * **Added UDP/IPv4 GRO support for VxLAN and non-VxLAN packets.**
258 For VxLAN packets, added inner UDP/IPv4 support.
259 For non-VxLAN packets, added UDP/IPv4 support.
261 * **Extended the flow-perf application.**
263 * Added support for user order instead of bit mask.
264 Now the user can create any structure of rte_flow
265 using the flow performance application with any order.
266 Moreover the app also now starts to support inner
267 items matching as well.
268 * Added header modify actions.
270 * Added raw encap/decap actions.
271 * Added VXLAN encap/decap actions.
272 * Added ICMP (code/type/identifier/sequence number) and ICMP6 (code/type) matching items.
273 * Added option to set port mask for insertion/deletion:
275 where N represents the hexadecimal bitmask of the ports used.
277 * **Added raw data-path APIs for cryptodev library.**
279 Added raw data-path APIs to Cryptodev to help accelerate external libraries
280 or applications which need to avail of fast cryptodev enqueue/dequeue
281 operations but which do not necessarily need to depend on mbufs and
282 cryptodev operation mempools.
284 * **Updated the aesni_mb crypto PMD.**
286 * Added support for intel-ipsec-mb version 0.55.
287 * Added support for AES-ECB 128, 192 and 256.
288 * Added support for ZUC-EEA3/EIA3 algorithms.
289 * Added support for SNOW3G-UEA2/UIA2 algorithms.
290 * Added support for KASUMI-F8/F9 algorithms.
291 * Added support for Chacha20-Poly1305.
292 * Added support for AES-256 CCM algorithm.
294 * **Updated the aesni_gcm crypto PMD.**
296 * Added SGL support for AES-GMAC.
298 * **Added Broadcom BCMFS symmetric crypto PMD.**
300 Added a symmetric crypto PMD for Broadcom FlexSparc crypto units.
301 See :doc:`../cryptodevs/bcmfs` guide for more details on this new PMD.
303 * **Updated NXP DPAA2_SEC crypto PMD.**
305 * Added DES-CBC support for cipher_only, chain and ipsec protocol.
306 * Added support for non-HMAC auth algorithms
307 (MD5, SHA1, SHA224, SHA256, SHA384, SHA512).
309 * **Updated Marvell NITROX symmetric crypto PMD.**
311 * Added AES-GCM support.
312 * Added cipher only offload support.
314 * **Updated Marvell OCTEON TX2 crypto PMD.**
316 * Updated the OCTEON TX2 crypto PMD lookaside protocol offload for IPsec with
319 * **Updated Intel QAT PMD.**
321 * Added Raw Data-path APIs support.
322 * Added support for write combining stores.
324 * **Added Intel ACC100 bbdev PMD.**
326 Added a new ``acc100`` bbdev driver for the Intel\ |reg| ACC100 accelerator
327 also known as Mount Bryce. See the
328 :doc:`../bbdevs/acc100` BBDEV guide for more details on this new driver.
330 * **Updated rte_security library to support SDAP.**
332 ``rte_security_pdcp_xform`` in ``rte_security`` lib is updated to enable
333 5G NR processing of SDAP headers in PMDs.
335 * **Added Marvell OCTEON TX2 regex PMD.**
337 Added a new PMD driver for the hardware regex offload block for OCTEON TX2 SoC.
339 See the :doc:`../regexdevs/octeontx2` for more details.
341 * **Updated Software Eventdev driver.**
343 Added performance tuning arguments to allow tuning the scheduler for
344 better throughput in high core count use cases.
346 * **Added a new driver for the Intel Dynamic Load Balancer v1.0 device.**
348 Added the new ``dlb`` eventdev driver for the Intel DLB V1.0 device. See the
349 :doc:`../eventdevs/dlb` eventdev guide for more details on this new driver.
351 * **Added a new driver for the Intel Dynamic Load Balancer v2.0 device.**
353 Added the new ``dlb2`` eventdev driver for the Intel DLB V2.0 device. See the
354 :doc:`../eventdevs/dlb2` eventdev guide for more details on this new driver.
356 * **Added Ice Lake (Gen4) support for Intel NTB.**
358 Added NTB device support (4th generation) for the Intel Ice Lake platform.
360 * **Updated ioat rawdev driver.**
362 The ioat rawdev driver has been updated and enhanced. Changes include:
364 * Added support for Intel\ |reg| Data Streaming Accelerator hardware. For
365 more information, see `Introducing the Intel Data Streaming Accelerator
367 <https://01.org/blogs/2019/introducing-intel-data-streaming-accelerator>`_.
368 * Added support for the fill operation via the API ``rte_ioat_enqueue_fill()``,
369 where the hardware fills an area of memory with a repeating pattern.
370 * Added a per-device configuration flag to disable management
371 of user-provided completion handles.
372 * Renamed the ``rte_ioat_do_copies()`` API to ``rte_ioat_perform_ops()``,
373 and renamed the ``rte_ioat_completed_copies()`` API to ``rte_ioat_completed_ops()``
374 to better reflect the APIs' purposes, and remove the implication that
375 they are limited to copy operations only.
376 Note: The old API is still provided but marked as deprecated in the code.
377 * Added a new API ``rte_ioat_fence()`` to add a fence between operations.
378 This API replaces the ``fence`` flag parameter in the ``rte_ioat_enqueue_copies()`` function,
379 and is clearer as there is no ambiguity as to whether the flag should be
380 set on the last operation before the fence or the first operation after it.
382 * **Updated the pipeline library for alignment with the P4 language.**
384 Added a new Software Switch (SWX) pipeline type that provides more
385 flexibility through APIs and feature alignment with the P4 language.
386 Some enhancements are:
388 * The packet headers, meta-data, actions, tables and pipelines are
389 dynamically defined instead of selected from a pre-defined set.
390 * The actions and the pipeline are defined with instructions.
391 * Extern objects and functions can be plugged into the pipeline.
392 * Transaction-oriented table updates.
394 * **Added new AVX512 specific classify algorithms for ACL library.**
396 * Added new ``RTE_ACL_CLASSIFY_AVX512X16`` vector implementation,
397 which can process up to 16 flows in parallel. Requires AVX512 support.
399 * Added new ``RTE_ACL_CLASSIFY_AVX512X32`` vector implementation,
400 which can process up to 32 flows in parallel. Requires AVX512 support.
402 * **Added AVX512 lookup implementation for FIB.**
404 Added a AVX512 lookup functions implementation into FIB and FIB6 libraries.
406 * **Added support to update subport bandwidth dynamically.**
408 * Added new API ``rte_sched_port_subport_profile_add`` to add new
409 subport bandwidth profiles to the subport profile table at runtime.
411 * Added support to update the subport rate dynamically.
413 * **Updated FIPS validation sample application.**
415 * Added scatter gather support.
416 * Added NIST GCMVS complaint GMAC test method support.
418 * **Updated l3wfd-acl sample application.**
420 * Added new optional parameter ``--eth-dest`` for the ``l3fwd-acl`` to allow
421 the user to specify the destination mac address for each ethernet port
423 * Replaced ``--scalar`` command-line option with ``--alg=<value>``, to allow
424 the user to select the desired classify method.
426 * **Updated vhost sample application.**
428 Added vhost asynchronous APIs support, which demonstrates how the application
429 can leverage IOAT DMA channels with vhost asynchronous APIs.
430 See the :doc:`../sample_app_ug/vhost` for more details.
436 .. This section should contain removed items in this release. Sample format:
438 * Add a short 1-2 sentence description of the removed item
441 This section is a comment. Do not overwrite or remove it.
442 Also, make sure to start the actual text at the margin.
443 =======================================================
445 * build: Support for the Make build system has been removed from DPDK.
446 Meson is now the primary build system.
447 Sample applications can still be built with Make standalone, using pkg-config.
449 * vhost: Dequeue zero-copy support has been removed.
451 * kernel: The module ``igb_uio`` has been moved to the git repository
452 `dpdk-kmods <https://git.dpdk.org/dpdk-kmods/>`_ in a new directory
455 * Removed Python 2 support since it was sunsetted in January 2020. See
456 `Sunsetting Python 2 <https://www.python.org/doc/sunset-python-2/>`_
458 * Removed TEP termination sample application.
460 * Removed the deprecated ``dpdk-setup.sh`` script.
466 .. This section should contain API changes. Sample format:
468 * sample: Add a short 1-2 sentence description of the API change
469 which was announced in the previous releases and made in this release.
470 Start with a scope label like "ethdev:".
471 Use fixed width quotes for ``function_names`` or ``struct_names``.
474 This section is a comment. Do not overwrite or remove it.
475 Also, make sure to start the actual text at the margin.
476 =======================================================
478 * build macros: The macros defining ``RTE_MACHINE_CPUFLAG_*`` have been removed.
479 The information provided by these macros is now available through standard
482 * eal: Replaced the function ``rte_get_master_lcore()`` with
483 ``rte_get_main_lcore()``. The old function is deprecated.
485 The iterator for worker lcores is also changed:
486 ``RTE_LCORE_FOREACH_SLAVE`` is replaced with
487 ``RTE_LCORE_FOREACH_WORKER``.
489 * eal: The definitions related to including and excluding devices
490 have been changed from blacklist/whitelist to block/allow list.
491 There are compatibility macros and command line mapping to accept
492 the old values but applications and scripts are strongly encouraged
493 to migrate to the new names.
495 * eal: The ``rte_logs`` struct and global symbol was made private
496 and is no longer part of the API.
498 * eal: Made the ``rte_dev_event`` structure private to the EAL as no public API
501 * eal: ``rte_cio_rmb()`` and ``rte_cio_wmb()`` were deprecated since 20.08
502 and are removed in this release.
504 * mem: Removed the unioned field ``phys_addr`` from
505 the structures ``rte_memseg`` and ``rte_memzone``.
506 The field ``iova`` remains from the old unions.
508 * mempool: Removed the unioned fields ``phys_addr`` and ``physaddr`` from
509 the structures ``rte_mempool_memhdr`` and ``rte_mempool_objhdr``.
510 The field ``iova`` remains from the old unions.
511 The flag name ``MEMPOOL_F_NO_PHYS_CONTIG`` is removed,
512 while the aliased flag ``MEMPOOL_F_NO_IOVA_CONTIG`` is kept.
514 * mbuf: Removed the functions ``rte_mbuf_data_dma_addr*``
515 and the macros ``rte_pktmbuf_mtophys*``.
516 The same functionality is still available with the functions and macros
517 having ``iova`` in their names instead of ``dma_addr`` or ``mtophys``.
519 * mbuf: Removed the unioned field ``buf_physaddr`` from ``rte_mbuf``.
520 The field ``buf_iova`` remains from the old union.
522 * mbuf: Removed the unioned field ``refcnt_atomic`` from
523 the structures ``rte_mbuf`` and ``rte_mbuf_ext_shared_info``.
524 The field ``refcnt`` remains from the old unions.
526 * mbuf: Removed the unioned fields ``userdata`` and ``udata64``
527 from the structure ``rte_mbuf``. It is replaced with dynamic fields.
529 * mbuf: Removed the field ``seqn`` from the structure ``rte_mbuf``.
530 It is replaced with dynamic fields.
532 * mbuf: Removed the field ``timestamp`` from the structure ``rte_mbuf``.
533 It is replaced with the dynamic field RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
534 which was previously used only for Tx.
536 * pci: Removed the ``rte_kernel_driver`` enum defined in rte_dev.h and
537 replaced with a private enum in the PCI subsystem.
539 * pci: Removed the PCI resources map API from the public API
540 (``pci_map_resource`` and ``pci_unmap_resource``) and moved it to the
541 PCI bus driver along with the PCI resources lists and associated structures
542 (``pci_map``, ``pci_msix_table``, ``mapped_pci_resource`` and
543 ``mapped_pci_res_list``).
545 * ethdev: Removed the ``kdrv`` field in the ethdev ``rte_eth_dev_data``
546 structure as it gave no useful abstracted information to the applications.
548 * ethdev: ``rte_eth_rx_descriptor_done()`` API has been deprecated.
550 * ethdev: Renamed basic statistics per queue. An underscore is inserted
551 between the queue number and the rest of the xstat name:
553 * ``rx_qN*`` -> ``rx_qN_*``
554 * ``tx_qN*`` -> ``tx_qN_*``
556 * ethdev: Added capability to query age flow action.
558 * ethdev: Changed ``rte_eth_dev_stop`` return value from ``void`` to
559 ``int`` to provide a way to report various error conditions.
561 * ethdev: Added ``int`` return type to ``rte_eth_dev_close()``.
563 * ethdev: Renamed internal functions:
565 * ``_rte_eth_dev_callback_process()`` -> ``rte_eth_dev_callback_process()``
566 * ``_rte_eth_dev_reset`` -> ``rte_eth_dev_internal_reset()``
568 * ethdev: Modified field type of ``base`` and ``nb_queue`` in struct
569 ``rte_eth_dcb_tc_queue_mapping`` from ``uint8_t`` to ``uint16_t``.
570 As the data of ``uint8_t`` will be truncated when queue number in
571 a TC is greater than 256.
573 * ethdev: Removed the legacy filter API, including
574 ``rte_eth_dev_filter_supported()`` and ``rte_eth_dev_filter_ctrl()``.
576 * ethdev: Removed the legacy L2 tunnel configuration API, including
577 ``rte_eth_dev_l2_tunnel_eth_type_conf()`` and
578 ``rte_eth_dev_l2_tunnel_offload_set()``..
580 * vhost: Moved vDPA APIs from experimental to stable.
582 * vhost: Add a new function ``rte_vhost_crypto_driver_start`` to be called
583 instead of ``rte_vhost_driver_start`` by crypto applications.
585 * cryptodev: The structure ``rte_crypto_sym_vec`` is updated to support both
586 cpu_crypto synchronous operations and asynchronous raw data-path APIs.
588 * cryptodev: ``RTE_CRYPTO_AEAD_LIST_END`` from ``enum rte_crypto_aead_algorithm``,
589 ``RTE_CRYPTO_CIPHER_LIST_END`` from ``enum rte_crypto_cipher_algorithm`` and
590 ``RTE_CRYPTO_AUTH_LIST_END`` from ``enum rte_crypto_auth_algorithm``
591 are removed to avoid future ABI breakage while adding new algorithms.
593 * scheduler: Renamed functions ``rte_cryptodev_scheduler_slave_attach``,
594 ``rte_cryptodev_scheduler_slave_detach`` and
595 ``rte_cryptodev_scheduler_slaves_get`` to
596 ``rte_cryptodev_scheduler_worker_attach``,
597 ``rte_cryptodev_scheduler_worker_detach`` and
598 ``rte_cryptodev_scheduler_workers_get`` accordingly.
600 * scheduler: Renamed the configuration value
601 ``RTE_CRYPTODEV_SCHEDULER_MAX_NB_SLAVES`` to
602 ``RTE_CRYPTODEV_SCHEDULER_MAX_NB_WORKERS``.
604 * security: The ``hfn_ovrd`` field in ``rte_security_pdcp_xform`` is changed from
605 ``uint32_t`` to ``uint8_t`` so that a new field ``sdap_enabled`` can be added
608 * security: The API ``rte_security_session_create`` is updated to take two
609 mempool objects: one for session and other for session private data.
610 So the application need to create two mempools and get the size of session
611 private data using API ``rte_security_session_get_size`` for private session
614 * ipsec: ``RTE_SATP_LOG2_NUM`` has been dropped from ``enum`` and
615 subsequently moved ``rte_ipsec`` lib from experimental to stable.
617 * baseband/fpga_lte_fec: Renamed function ``fpga_lte_fec_configure`` to
618 ``rte_fpga_lte_fec_configure`` and structure ``fpga_lte_fec_conf`` to
619 ``rte_fpga_lte_fec_conf``.
621 * baseband/fpga_5gnr_fec: Renamed function ``fpga_5gnr_fec_configure`` to
622 ``rte_fpga_5gnr_fec_configure`` and structure ``fpga_5gnr_fec_conf`` to
623 ``rte_fpga_5gnr_fec_conf``.
625 * rawdev: Added a structure size parameter to the functions
626 ``rte_rawdev_queue_setup()``, ``rte_rawdev_queue_conf_get()``,
627 ``rte_rawdev_info_get()`` and ``rte_rawdev_configure()``,
628 allowing limited driver type-checking and ABI compatibility.
630 * rawdev: Changed the return type of the function ``rte_dev_info_get()``
631 and the function ``rte_rawdev_queue_conf_get()``
632 from ``void`` to ``int`` allowing the return of error codes from drivers.
634 * rawdev: The running of a drivers ``selftest()`` function can now be done
635 using the ``rawdev_autotest`` command in the ``dpdk-test`` binary. This
636 command now calls the self-test function for each rawdev found on the
637 system, and does not require a specific command per device type.
638 Following this change, the ``ioat_rawdev_autotest`` command has been
639 removed as no longer needed.
641 * raw/ioat: As noted above, the ``rte_ioat_do_copies()`` and
642 ``rte_ioat_completed_copies()`` functions have been renamed to
643 ``rte_ioat_perform_ops()`` and ``rte_ioat_completed_ops()`` respectively.
645 * stack: the experimental tag has been dropped from the stack library, and its
646 interfaces are considered stable as of DPDK 20.11.
648 * bpf: ``RTE_BPF_XTYPE_NUM`` has been dropped from ``rte_bpf_xtype``.
650 * gso: Changed ``rte_gso_segment`` behaviour and return value:
652 * ``pkt`` is not saved to ``pkts_out[0]`` if not GSOed.
653 * Return 0 instead of 1 for the above case.
654 * ``pkt`` is not freed, no matter whether it is GSOed, leaving to the caller.
656 * acl: ``RTE_ACL_CLASSIFY_NUM`` enum value has been removed.
657 This enum value was not used inside DPDK, while it prevented the addition of new
658 classify algorithms without causing an ABI breakage.
660 * sched: Added ``subport_profile_id`` as an argument
661 to function ``rte_sched_subport_config``.
663 * sched: Removed ``tb_rate``, ``tc_rate``, ``tc_period`` and ``tb_size``
664 from ``struct rte_sched_subport_params``.
670 .. This section should contain ABI changes. Sample format:
672 * sample: Add a short 1-2 sentence description of the ABI change
673 which was announced in the previous releases and made in this release.
674 Start with a scope label like "ethdev:".
675 Use fixed width quotes for ``function_names`` or ``struct_names``.
678 This section is a comment. Do not overwrite or remove it.
679 Also, make sure to start the actual text at the margin.
680 =======================================================
682 * eal: Removed the unimplemented function ``rte_dump_registers()``.
686 * The following device operation function pointers moved
687 from ``struct eth_dev_ops`` to ``struct rte_eth_dev``:
689 * ``eth_rx_queue_count_t rx_queue_count;``
690 * ``eth_rx_descriptor_done_t rx_descriptor_done;``
691 * ``eth_rx_descriptor_status_t rx_descriptor_status;``
692 * ``eth_tx_descriptor_status_t tx_descriptor_status;``
694 * ``struct eth_dev_ops`` is no longer accessible by applications,
695 which was already an internal data structure.
697 * ``ethdev`` internal functions are marked with ``__rte_internal`` tag.
699 * Added extensions' attributes to struct ``rte_flow_item_ipv6``.
700 A set of additional values added to struct, indicating the existence of
701 every defined extension header type.
702 Applications should use the new values for identification of existing
703 extensions in the packet header.
705 * Added fields ``rx_seg`` and ``rx_nseg`` to ``rte_eth_rxconf`` structure
706 to provide extended description of the receiving buffer.
708 * ``struct rte_eth_hairpin_conf`` has two new members:
710 * ``uint32_t tx_explicit:1;``
711 * ``uint32_t manual_bind:1;``
713 * Added new field ``has_vlan`` to structure ``rte_flow_item_eth``,
714 indicating that packet header contains at least one VLAN.
716 * Added new field ``has_more_vlan`` to the structure
717 ``rte_flow_item_vlan``, indicating that packet header contains
718 at least one more VLAN, after this VLAN.
720 * eventdev: The following structures are modified to support DLB/DLB2 PMDs
721 and future extensions:
723 * ``rte_event_dev_info``
724 * ``rte_event_dev_config``
725 * ``rte_event_port_conf``
727 * sched: Added new fields to ``struct rte_sched_subport_port_params``.
729 * lpm: Removed fields other than ``tbl24`` and ``tbl8`` from the struct
730 ``rte_lpm``. The removed fields were made internal.
736 .. This section should contain a list of platforms that were tested
741 * <vendor> platform with <vendor> <type of devices> combinations
746 * Other relevant details...
748 This section is a comment. Do not overwrite or remove it.
749 Also, make sure to start the actual text at the margin.
750 =======================================================
752 * Intel\ |reg| platforms with Intel\ |reg| NICs combinations
756 * Intel\ |reg| Atom\ |trade| CPU C3758 @ 2.20GHz
757 * Intel\ |reg| Atom\ |trade| CPU C3858 @ 2.00GHz
758 * Intel\ |reg| Atom\ |trade| CPU C3958 @ 2.00GHz
759 * Intel\ |reg| Xeon\ |reg| CPU D-1541 @ 2.10GHz
760 * Intel\ |reg| Xeon\ |reg| CPU D-1553N @ 2.30GHz
761 * Intel\ |reg| Xeon\ |reg| CPU E5-2680 0 @ 2.70GHz
762 * Intel\ |reg| Xeon\ |reg| CPU E5-2680 v2 @ 2.80GHz
763 * Intel\ |reg| Xeon\ |reg| CPU E5-2699 v3 @ 2.30GHz
764 * Intel\ |reg| Xeon\ |reg| CPU E5-2699 v4 @ 2.20GHz
765 * Intel\ |reg| Xeon\ |reg| Gold 5218N CPU @ 2.30GHz
766 * Intel\ |reg| Xeon\ |reg| Gold 6139 CPU @ 2.30GHz
767 * Intel\ |reg| Xeon\ |reg| Gold 6252N CPU @ 2.30GHz
768 * Intel\ |reg| Xeon\ |reg| Platinum 8180 CPU @ 2.50GHz
769 * Intel\ |reg| Xeon\ |reg| Platinum 8280M CPU @ 2.70GHz
777 * Red Hat Enterprise Linux Server release 8.2
785 * Intel\ |reg| Ethernet Controller E810-C for SFP (4x25G)
787 * Firmware version: 2.30 0x80004dcf 1.2839.0
788 * Device id (pf/vf): 8086:1593 / 8086:1889
789 * Driver version: 1.3.0 (ice)
790 * OS Default DDP: 1.3.20.0
791 * COMMS DDP: 1.3.24.0
793 * Intel\ |reg| Ethernet Controller E810-C for QSFP (2x100G)
795 * Firmware version: 2.30 0x80004dd0 1.2839.0
796 * Device id (pf/vf): 8086:1592 / 8086:1889
797 * Driver version: 1.3.0 (ice)
798 * OS Default DDP: 1.3.20.0
799 * COMMS DDP: 1.3.24.0
801 * Intel\ |reg| 82599ES 10 Gigabit Ethernet Controller
803 * Firmware version: 0x61bf0001
804 * Device id (pf/vf): 8086:10fb / 8086:10ed
805 * Driver version: 5.7.1 (ixgbe)
807 * Intel\ |reg| Corporation Ethernet Connection X552/X557-AT 10GBASE-T
809 * Firmware version: 0x800003e7
810 * Device id (pf/vf): 8086:15ad / 8086:15a8
811 * Driver version: 5.1.0-k (ixgbe)
813 * Intel\ |reg| Corporation Ethernet Controller 10G X550T
815 * Firmware version: 0x80000482
816 * Device id (pf): 8086:1563
817 * Driver version: 5.7.1 (ixgbe)
819 * Intel\ |reg| Ethernet Converged Network Adapter X710-DA4 (4x10G)
821 * Firmware version: 8.00 0x80008b82 1.2766.0
822 * Device id (pf/vf): 8086:1572 / 8086:154c
823 * Driver version: 2.13.10 (i40e)
825 * Intel\ |reg| Corporation Ethernet Connection X722 for 10GbE SFP+ (4x10G)
827 * Firmware version: 5.00 0x800023c3 1.2766.0
828 * Device id (pf/vf): 8086:37d0 / 8086:37cd
829 * Driver version: 2.13.10 (i40e)
831 * Intel\ |reg| Corporation Ethernet Connection X722 for 10GBASE-T (2x10G)
833 * Firmware version: 4.10 0x80001a7a
834 * Device id (pf/vf): 8086:37d2 / 8086:37cd
835 * Driver version: 2.13.10 (i40e)
837 * Intel\ |reg| Ethernet Converged Network Adapter XXV710-DA2 (2x25G)
839 * Firmware version: 8.00 0x80008c1a 1.2766.0
840 * Device id (pf/vf): 8086:158b / 8086:154c
841 * Driver version: 2.13.10 (i40e)
843 * Intel\ |reg| Ethernet Converged Network Adapter XL710-QDA2 (2X40G)
845 * Firmware version: 8.00 0x80008b82 1.2766.0
846 * Device id (pf/vf): 8086:1583 / 8086:154c
847 * Driver version: 2.13.10 (i40e)
849 * Intel\ |reg| Corporation I350 Gigabit Network Connection
851 * Firmware version: 1.63, 0x80000cbc
852 * Device id (pf/vf): 8086:1521 / 8086:1520
853 * Driver version: 5.4.0-k (igb)
855 * Intel\ |reg| Corporation I210 Gigabit Network Connection
857 * Firmware version: 3.25, 0x800006eb
858 * Device id (pf): 8086:1533
859 * Driver version: 5.4.0-k (igb)
861 * Intel\ |reg| Ethernet Controller 10-Gigabit X540-AT2
863 * Firmware version: 0x800005f9
864 * Device id (pf): 8086:1528
865 * Driver version: 5.1.0-k (ixgbe)
867 * Intel\ |reg| Ethernet Converged Network Adapter X710-T2L
869 * Firmware version: 8.00 0x80008d10 1.2766.0
870 * Device id (pf): 8086:15ff
871 * Driver version: 2.13.10 (i40e)
873 * Intel\ |reg| platforms with Mellanox\ |reg| NICs combinations
877 * Intel\ |reg| Xeon\ |reg| Gold 6154 CPU @ 3.00GHz
878 * Intel\ |reg| Xeon\ |reg| CPU E5-2697A v4 @ 2.60GHz
879 * Intel\ |reg| Xeon\ |reg| CPU E5-2697 v3 @ 2.60GHz
880 * Intel\ |reg| Xeon\ |reg| CPU E5-2680 v2 @ 2.80GHz
881 * Intel\ |reg| Xeon\ |reg| CPU E5-2670 0 @ 2.60GHz
882 * Intel\ |reg| Xeon\ |reg| CPU E5-2650 v4 @ 2.20GHz
883 * Intel\ |reg| Xeon\ |reg| CPU E5-2640 @ 2.50GHz
884 * Intel\ |reg| Xeon\ |reg| CPU E5-2650 0 @ 2.00GHz
885 * Intel\ |reg| Xeon\ |reg| CPU E5-2620 v4 @ 2.10GHz
889 * Red Hat Enterprise Linux release 8.2 (Ootpa)
890 * Red Hat Enterprise Linux Server release 7.8 (Maipo)
891 * Red Hat Enterprise Linux Server release 7.6 (Maipo)
892 * Red Hat Enterprise Linux Server release 7.5 (Maipo)
893 * Red Hat Enterprise Linux Server release 7.4 (Maipo)
894 * Red Hat Enterprise Linux Server release 7.3 (Maipo)
895 * Red Hat Enterprise Linux Server release 7.2 (Maipo)
899 * SUSE Enterprise Linux 15 SP2
900 * SUSE Enterprise Linux 12 SP4
904 * MLNX_OFED 5.2-0.3.3.0 and above
905 * MLNX_OFED 5.1-2.5.8.0
909 * Linux 5.10.0-rc2 and above
913 * rdma-core-31.0-1 and above
917 * Mellanox\ |reg| ConnectX\ |reg|-3 Pro 40G MCX354A-FCC_Ax (2x40G)
919 * Host interface: PCI Express 3.0 x8
920 * Device ID: 15b3:1007
921 * Firmware version: 2.42.5000
923 * Mellanox\ |reg| ConnectX\ |reg|-3 Pro 40G MCX354A-FCCT (2x40G)
925 * Host interface: PCI Express 3.0 x8
926 * Device ID: 15b3:1007
927 * Firmware version: 2.42.5000
929 * Mellanox\ |reg| ConnectX\ |reg|-4 Lx 25G MCX4121A-ACAT (2x25G)
931 * Host interface: PCI Express 3.0 x8
932 * Device ID: 15b3:1015
933 * Firmware version: 14.29.0476 and above
935 * Mellanox\ |reg| ConnectX\ |reg|-4 Lx 50G MCX4131A-GCAT (1x50G)
937 * Host interface: PCI Express 3.0 x8
938 * Device ID: 15b3:1015
939 * Firmware version: 14.29.0476 and above
941 * Mellanox\ |reg| ConnectX\ |reg|-5 100G MCX516A-CCAT (2x100G)
943 * Host interface: PCI Express 3.0 x16
944 * Device ID: 15b3:1017
945 * Firmware version: 16.29.0476 and above
947 * Mellanox\ |reg| ConnectX\ |reg|-5 100G MCX556A-ECAT (2x100G)
949 * Host interface: PCI Express 3.0 x16
950 * Device ID: 15b3:1017
951 * Firmware version: 16.29.0476 and above
953 * Mellanox\ |reg| ConnectX\ |reg|-5 100G MCX556A-EDAT (2x100G)
955 * Host interface: PCI Express 3.0 x16
956 * Device ID: 15b3:1017
957 * Firmware version: 16.29.0476 and above
959 * Mellanox\ |reg| ConnectX\ |reg|-5 Ex EN 100G MCX516A-CDAT (2x100G)
961 * Host interface: PCI Express 4.0 x16
962 * Device ID: 15b3:1019
963 * Firmware version: 16.29.0476 and above
965 * Mellanox\ |reg| ConnectX\ |reg|-6 Dx EN 100G MCX623106AN-CDAT (2x100G)
967 * Host interface: PCI Express 4.0 x16
968 * Device ID: 15b3:101d
969 * Firmware version: 22.29.0476 and above
971 * Mellanox\ |reg| ConnectX\ |reg|-6 Lx EN 25G MCX631102AN-ADAT (2x25G)
973 * Host interface: PCI Express 4.0 x8
974 * Device ID: 15b3:101f
975 * Firmware version: 26.29.0470 and above
977 * Mellanox\ |reg| BlueField\ |reg| SmartNIC
979 * Mellanox\ |reg| BlueField\ |reg| 2 SmartNIC MT41686 - MBF2H332A-AEEOT (2x25G)
981 * Host interface: PCI Express 3.0 x16
982 * Device ID: 15b3:a2d2
983 * Firmware version: 24.29.0476 and above
987 * CentOS Linux release 7.6.1810 (AltArch)
988 * MLNX_OFED 5.2-0.3.2 and above
989 * DPDK application running on Arm cores
991 * Intel\ |reg| platforms with Broadcom\ |reg| NICs combinations
995 * Intel\ |reg| Xeon\ |reg| Gold 6154 CPU @ 3.00GHz
996 * Intel\ |reg| Xeon\ |reg| CPU E5-2667 v3 @ 3.20GHz
997 * Intel\ |reg| Xeon\ |reg| CPU E5-2650 v2 @ 2.60GHz
998 * Intel\ |reg| Xeon\ |reg| Gold 6142 CPU @ 2.60GHz
999 * Intel\ |reg| Xeon\ |reg| Silver 4110 CPU @ 2.10GHz
1000 * Intel\ |reg| Xeon\ |reg| CPU E3-1270 v3 @ 3.50GHz
1001 * Intel\ |reg| Xeon\ |reg| Gold 6134M CPU @ 3.20GHz
1005 * Red Hat Enterprise Linux Server release 8.1
1006 * Red Hat Enterprise Linux Server release 7.6
1007 * Red Hat Enterprise Linux Server release 7.5
1019 * Broadcom\ |reg| NetXtreme-E\ |reg| Series P225p (2x25G)
1021 * Host interface: PCI Express 3.0 x8
1022 * Firmware version: 214.4.114.0 and above
1024 * Broadcom\ |reg| NetXtreme-E\ |reg| Series P425p (4x25G)
1026 * Host interface: PCI Express 3.0 x16
1027 * Firmware version: 216.4.259.0 and above
1029 * Broadcom\ |reg| NetXtreme-E\ |reg| Series P2100G (2x100G)
1031 * Host interface: PCI Express 3.0 x16
1032 * Firmware version: 216.1.259.0 and above
1034 * Broadcom\ |reg| NetXtreme-E\ |reg| Series P425p (4x25G)
1036 * Host interface: PCI Express 4.0 x16
1037 * Firmware version: 216.1.259.0 and above
1039 * Broadcom\ |reg| NetXtreme-E\ |reg| Series P2100G (2x100G)
1041 * Host interface: PCI Express 4.0 x16
1042 * Firmware version: 216.1.259.0 and above
1044 * Broadcom\ |reg| NetXtreme-S\ |reg| Series SmartNIC
1046 * Broadcom\ |reg| NetXtreme-S\ |reg| Series PS225-H16 SmartNIC (2x25G)
1048 * Host interface: PCI Express 3.0 x8
1049 * Firmware version: 217.0.59.0, 218.1.63.0
1051 * Embedded software:
1053 * Broadcom Yocto Linux
1054 * Kernel version: 4.14.174
1055 * DPDK application running on 8 Arm Cortex-A72 cores