7 The text below explains how to update the release notes.
9 Use proper spelling, capitalization and punctuation in all sections.
11 Variable and config names should be quoted as fixed width text: ``LIKE_THIS``.
13 Build the docs and view the output file to ensure the changes are correct::
17 firefox build/doc/html/guides/rel_notes/release_16_04.html
23 This section should contain new features added in this release. Sample format:
25 * **Add a title in the past tense with a full stop.**
27 Add a short 1-2 sentence description in the past tense. The description
28 should be enough to allow someone scanning the release notes to understand
31 If the feature adds a lot of sub-features you can use a bullet list like this.
33 * Added feature foo to do something.
34 * Enhanced feature bar to do something else.
36 Refer to the previous release notes for examples.
38 * **Added function to check primary process state.**
40 A new function ``rte_eal_primary_proc_alive()`` has been added
41 to allow the user to detect if a primary process is running.
42 Use cases for this feature include fault detection, and monitoring
43 using secondary processes.
45 * **Enabled bulk allocation of mbufs.**
47 A new function ``rte_pktmbuf_alloc_bulk()`` has been added to allow the user
48 to allocate a bulk of mbufs.
50 * **Added device link speed capabilities.**
52 The structure ``rte_eth_dev_info`` has now a ``speed_capa`` bitmap, which
53 allows the application to know the supported speeds of each device.
55 * **Added bitmap of link speeds to advertise.**
57 Allow defining a set of advertised speeds for auto-negotiation,
58 explicitly disabling link auto-negotiation (single speed)
59 and full auto-negotiation.
61 * **Added new poll-mode driver for Amazon Elastic Network Adapters (ENA).**
63 The driver operates variety of ENA adapters through feature negotiation
64 with the adapter and upgradable commands set.
65 ENA driver handles PCI Physical and Virtual ENA functions.
67 * **Restored vmxnet3 Tx data ring.**
69 Tx data ring has been shown to improve small pkt forwarding performance
70 on vSphere environment.
72 * **Added vmxnet3 Tx L4 checksum offload.**
74 Support TCP/UDP checksum offload.
76 * **Added vmxnet3 TSO support.**
78 * **Added vmxnet3 support for jumbo frames.**
80 Added support for linking multi-segment buffers together to
85 Enabled virtio 1.0 support for virtio pmd driver.
87 * **Supported virtio for ARM.**
89 Enabled virtio support for armv7/v8. Tested for arm64.
90 Virtio for arm support VFIO-noiommu mode only.
91 Virtio can work with other non-x86 arch too like powerpc.
93 * **Supported virtio offload in vhost-user.**
95 Add the offload and negotiation of checksum and TSO between vhost-user and
96 vanilla Linux virtio guest.
98 * **Added vhost-user live migration support.**
100 * **Added vhost driver.**
102 Added virtual PMD that wraps librte_vhost.
104 * **Added multicast promiscuous mode support on VF for ixgbe.**
106 Added multicast promiscuous mode support on ixgbe VF driver. So all the VFs
107 can receive the multicast packets.
109 Please note if we want to use this promiscuous mode, we need both PF and VF
110 driver to support it. The reason is this VF feature is configured on PF.
111 If use kernel PF driver + dpdk VF driver, make sure kernel PF driver support
112 VF multicast promiscuous mode. If use dpdk PF + dpdk VF, better make sure PF
113 driver is the same version as VF.
115 * **Added support for E-tag on X550.**
117 E-tag is defined in 802.1br. Please reference
118 http://www.ieee802.org/1/pages/802.1br.html.
120 This feature is for VF, but the settings are on PF. It means
121 the CLIs should be used on PF, but some of their effects will be shown on VF.
122 The forwarding of E-tag packets based on GRP and E-CID_base will have effect
123 on PF. Theoretically the E-tag packets can be forwarded to any pool/queue.
124 But normally we'd like to forward the packets to the pools/queues belonging
125 to the VFs. And E-tag insertion and stripping will have effect on VFs. When
126 VF receives E-tag packets, it should strip the E-tag. When VF transmits
127 packets, it should insert the E-tag. Both can be offloaded.
129 When we want to use this E-tag support feature, the forwarding should be
130 enabled to forward the packets received by PF to indicated VFs. And insertion
131 and stripping should be enabled for VFs to offload the effort to HW.
133 * Support E-tag offloading of insertion and stripping.
134 * Support Forwarding E-tag packets to pools based on
137 * **Added support for VxLAN & NVGRE checksum off-load on X550.**
139 * Added support for VxLAN & NVGRE RX/TX checksum off-load on
140 X550. RX/TX checksum off-load is provided on both inner and
141 outer IP header and TCP header.
142 * Added functions to support VxLAN port configuration. The
143 default VxLAN port number is 4789 but this can be updated
146 * **Added new X550EM_a devices.**
148 Added new X550EM_a devices and their mac types, X550EM_a and X550EM_a_vf.
149 Updated the code to use the new devices and mac types.
151 * **Added x550em_x V2 device support.**
153 Only x550em_x V1 was supported before. Now V2 is supported.
154 A mask for V1 and V2 is defined and used to support both.
156 * **Supported link speed auto-negotiation on X550EM_X**
158 Normally the auto-negotiation is supported by FW. SW need not care about
159 that. But on x550em_x, FW doesn't support auto-neg. As the ports of x550em_x
160 are 10G, if we connect the port with a peer which is 1G, the link will always
162 We added the support of auto-neg by SW to avoid this link down issue.
164 * **Added sw-firmware sync on X550EM_a.**
166 Added support for sw-firmware sync for resource sharing.
167 Use the PHY token, shared between sw-fw for PHY access on X550EM_a.
169 * **Updated the i40e base driver.**
171 The i40e base driver was updated with changes including the
174 * Use Rx control AQ commands to read/write Rx control registers.
175 * Add new X722 device IDs, and removed X710 one was never used.
176 * Expose registers for HASH/FD input set configuring.
178 * **Enabled PCI extended tag for i40e.**
180 It enabled extended tag by checking and writing corresponding PCI config
181 space bytes, to boost the performance. In the meanwhile, it deprecated the
182 legacy way via reading/writing sysfile supported by kernel module igb_uio.
184 * **Added i40e support for setting mac addresses.**
186 * **Added dump of i40e registers and EEPROM.**
188 * **Supported ether type setting of single and double VLAN for i40e**
190 * **Added VMDQ DCB mode in i40e.**
192 Added support for DCB in VMDQ mode to i40e driver.
194 * **Added i40e VEB switching support.**
196 * **Added Flow director enhancements in i40e.**
198 * **Added PF reset event reporting in i40e VF driver.**
200 * **Added fm10k Rx interrupt support.**
202 * **Optimized fm10k Tx.**
204 * Free multiple mbufs at a time to reduce freeing mbuf cycles.
206 * **Handled error flags in fm10k vector Rx.**
208 Parse err flags in Rx desc and set error bits in mbuf with vector instructions.
210 * **Added fm10k FTAG based forwarding support.**
212 * **Added mlx5 flow director support.**
214 Added flow director support (RTE_FDIR_MODE_PERFECT and
215 RTE_FDIR_MODE_PERFECT_MAC_VLAN).
217 Only available with Mellanox OFED >= 3.2.
219 * **Added mlx5 RX VLAN stripping support.**
221 Added support for RX VLAN stripping.
223 Only available with Mellanox OFED >= 3.2.
225 * **Added mlx5 link up/down callbacks.**
227 Implemented callbacks to bring link up and down.
229 * **Added mlx5 support for operation in secondary processes.**
231 Implemented TX support in secondary processes (like mlx4).
233 * **Added mlx5 RX CRC stripping configuration.**
235 Until now, CRC was always stripped. It can now be configured.
237 Only available with Mellanox OFED >= 3.2.
239 * **Added mlx5 optional packet padding by HW.**
241 Added an option to make PCI bus transactions rounded to multiple of a
242 cache line size for better alignment.
244 Only available with Mellanox OFED >= 3.2.
246 * **Added mlx5 TX VLAN insertion support.**
248 Added support for TX VLAN insertion.
250 Only available with Mellanox OFED >= 3.2.
252 * **Changed szedata2 type of driver from vdev to pdev.**
254 Previously szedata2 device had to be added by ``--vdev`` option.
255 Now szedata2 PMD recognises the device automatically during EAL
258 * **Added szedata2 functions for setting link up/down.**
260 * **Added szedata2 promiscuous and allmulticast modes.**
262 * **Added af_packet dynamic removal function.**
264 Af_packet device can now be detached using API, like other PMD devices.
266 * **Increased number of next hops for LPM IPv4 to 2^24.**
268 The next_hop field is extended from 8 bits to 24 bits for IPv4.
270 * **Added support of SNOW 3G (UEA2 and UIA2) for Intel Quick Assist devices.**
272 Enabled support for SNOW 3G wireless algorithm for Intel Quick Assist devices.
273 Support for cipher only, hash only is also provided
274 along with alg-chaining operations.
276 * **Added SNOW3G SW PMD.**
278 A new Crypto PMD has been added, which provides SNOW 3G UEA2 ciphering
279 and SNOW3G UIA2 hashing.
281 * **Added AES GCM PMD.**
283 Added new Crypto PMD to support AES-GCM authenticated encryption and
284 authenticated decryption in SW.
286 * **Added NULL Crypto PMD**
288 Added new Crypto PMD to support null crypto operations in SW.
290 * **Improved IP Pipeline Application.**
292 The following features have been added to ip_pipeline application;
294 * Added CPU utilization measurement and idle cycle rate computation.
295 * Added link idenfication support through existing port-mask option or by
296 specifying PCI device in every LINK section in the configuration file.
297 * Added load balancing support in passthrough pipeline.
299 * **Added IPsec security gateway example.**
301 New application implementing an IPsec Security Gateway.
307 This section should contain bug fixes added to the relevant sections. Sample format:
309 * **code/section: Fixed issue in the past tense with a full stop.**
311 Add a short 1-2 sentence description of the resolved issue in the past tense.
312 The title should contain the code/lib section like a commit message.
313 Add the entries in alphabetic order in the relevant sections below.
323 * **ethdev: Fixed overflow for 100Gbps.**
325 100Gbps in Mbps (100000) was exceeding 16-bit max value of ``link_speed``
328 * **ethdev: Fixed byte order consistency between fdir flow and mask.**
330 Fixed issue in ethdev library that the structure for setting
331 fdir's mask and flow entry was not consistent in byte ordering.
333 * **cxgbe: Fixed crash due to incorrect size allocated for RSS table.**
335 Fixed a segfault that occurs when accessing part of port 0's RSS
336 table that gets overwritten by subsequent port 1's part of the RSS
337 table due to incorrect size allocated for each entry in the table.
339 * **cxgbe: Fixed setting wrong device MTU.**
341 Fixed an incorrect device MTU being set due to ethernet header and
342 CRC lengths being added twice.
344 * **ixgbe: Fixed zeroed VF mac address.**
346 Resolved an issue where VF mac address is zeroed out in cases where the VF
347 driver is loaded while the PF interface is down.
348 The solution is to only set it when we get an ACK from the PF.
350 * **ixgbe: Fixed setting flow director flag twice.**
352 Resolved an issue where packets were being dropped when switching to perfect
355 * **ixgbe: Set MDIO speed after MAC reset.**
357 The MDIO clock speed must be reconfigured after the MAC reset. The MDIO clock
358 speed becomes invalid, therefore the driver reads invalid PHY register values.
359 The driver now set the MDIO clock speed prior to initializing PHY ops and
360 again after the MAC reset.
362 * **ixgbe: Fixed maximum number of available TX queues.**
364 In IXGBE, the maximum number of TX queues varies depending on the NIC operating
365 mode. This was not being updated in the device information, providing
366 an incorrect number in some cases.
368 * **i40e: Generated MAC address for each VFs.**
370 It generates a MAC address for each VFs during PF host initialization,
371 and keeps the VF MAC address the same among different VF launch.
373 * **i40e: Fixed failure of reading/writing Rx control registers.**
375 Fixed i40e issue of failing to read/write rx control registers when
376 under stress with traffic, which might result in application launch
379 * **i40e: Enabled vector driver by default.**
381 Previously, vector driver is disabled by default as it cannot fill packet type
382 info for l3fwd to work well. Now there is an option for l3fwd to analysis
383 packet type softly, so enable vector driver by default.
385 * **i40e: Fixed link info of VF.**
387 Previously, the VF's link speed kept as 10G and status always was up.
388 It did not change even the physical link's status changed.
389 Now this issue is fixed to make VF's link info consistent with physical link.
391 * **mlx5: Fixed possible crash during initialization.**
393 A crash could occur when failing to allocate private device context.
395 * **mlx5: Added port type check.**
397 Done to prevent port initialization on non-Ethernet link layers and
400 * **mlx5: Applied VLAN filtering to broadcast and IPv6 multicast flows.**
402 Prevented reception of multicast frames outside of configured VLANs.
404 * **mlx5: Fixed RX checksum offload in non L3/L4 packets.**
406 Fixed report of bad checksum for packets of unknown type.
408 * **aesni_mb: Fixed wrong return value when creating a device.**
410 cryptodev_aesni_mb_init() was returning the device id of the device created,
411 instead of 0 (when success), that rte_eal_vdev_init() expects.
412 This made impossible the creation of more than one aesni_mb device
415 * **qat: Fixed AES GCM decryption.**
417 Allowed AES GCM on the cryptodev API, but in some cases gave invalid results
418 due to incorrect IV setting.
424 * **hash: Fixed CRC32c hash computation for non multiple of 4 bytes sizes.**
426 Fix crc32c hash functions to return a valid crc32c value for data lengths
427 not multiple of 4 bytes.
429 * **hash: Fixed hash library to support multi-process mode.**
431 Fix hash library to support multi-process mode, using a jump table,
432 instead of storing a function pointer to the key compare function.
433 Multi-process mode only works with the built-in compare functions,
434 however a custom compare function (not in the jump table) can only
435 be used in single-process mode.
437 * **lpm: Fixed return value when allocating an existing object.**
439 Changed the ``rte_lpm*_create()`` functions to return ``NULL`` and set
440 ``rte_errno`` to ``EEXIST`` when the object name already exists. This is
441 the behavior described in the API documentation in the header file.
442 The previous behavior was to return a pointer to the existing object in
443 that case, preventing the caller to know if the object had to be freed
446 * **librte_port: Fixed segmentation fault for ring and ethdev writer nodrop.**
448 Fixed core dump issue on txq and swq when dropless is set to yes.
454 * **l3fwd-power: Fixed memory leak for non-IP packet.**
456 Fixed issue in l3fwd-power where, on receiving packets of types
457 other than IPv4 or IPv6, the mbuf was not released, and caused
460 * **l3fwd: Fixed using packet type blindly.**
462 l3fwd makes use of packet type information without even query if devices or PMDs
463 really set it. For those don't set ptypes, add an option to parse it softly.
465 * **examples/vhost: Fixed frequent mbuf allocation failure.**
467 vhost-switch often fails to allocate mbuf when dequeue from vring because it
468 wrongly calculates the number of mbufs needed.
478 This section should contain new known issues in this release. Sample format:
480 * **Add title in present tense with full stop.**
482 Add a short 1-2 sentence description of the known issue in the present
483 tense. Add information on any known workarounds.
489 This section should contain API changes. Sample format:
491 * Add a short 1-2 sentence description of the API change. Use fixed width
492 quotes for ``rte_function_names`` or ``rte_struct_names``. Use the past tense.
494 * The ethdev statistics counter imissed is considered to be independent of ierrors.
495 All drivers are now counting the missed packets only once, i.e. drivers will
496 not increment ierrors anymore for missed packets.
498 * The ethdev structure ``rte_eth_dev_info`` was changed to support device
501 * The ethdev structures ``rte_eth_link`` and ``rte_eth_conf`` were changed to
502 support the new link API.
504 * The functions ``rte_eth_dev_udp_tunnel_add`` and ``rte_eth_dev_udp_tunnel_delete``
505 have been renamed into ``rte_eth_dev_udp_tunnel_port_add`` and
506 ``rte_eth_dev_udp_tunnel_port_delete``.
508 * The ``outer_mac`` and ``inner_mac`` fields in structure
509 ``rte_eth_tunnel_filter_conf`` are changed from pointer to struct in order
510 to keep code's readability.
512 * The fields in ethdev structure ``rte_eth_fdir_masks`` were changed
515 * A parameter ``vlan_type`` has been added to the function
516 ``rte_eth_dev_set_vlan_ether_type``.
518 * AF_packet device init function is no longer public. Device should be attached
521 * The LPM ``next_hop`` field is extended from 8 bits to 24 bits for IPv4
522 while keeping ABI compatibility.
524 * A new ``rte_lpm_config`` structure is used so LPM library will allocate
525 exactly the amount of memory which is necessary to hold application’s rules.
526 The previous ABI is kept for compatibility.
528 * The prototype for the pipeline input port, output port and table action
529 handlers are updated: the pipeline parameter is added,
530 the packets mask parameter has been either removed or made input-only.
536 * Add a short 1-2 sentence description of the ABI change that was announced in
537 the previous releases and made in this release. Use fixed width quotes for
538 ``rte_function_names`` or ``rte_struct_names``. Use the past tense.
540 * The RETA entry size in ``rte_eth_rss_reta_entry64`` has been increased
541 from 8-bit to 16-bit.
543 * The ethdev flow director structure ``rte_eth_fdir_flow`` structure was
544 changed. New fields were added to extend flow director's input set.
546 * The cmdline buffer size has been increase from 256 to 512.
549 Shared Library Versions
550 -----------------------
552 Update any library version updated in this release and prepend with a ``+`` sign.
554 The libraries prepended with a plus sign were incremented in this version.
561 + librte_cmdline.so.2
562 librte_distributor.so.1
574 + librte_pipeline.so.3