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_18_05.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 * **Added bucket mempool driver.**
46 Added bucket mempool driver which provides a way to allocate contiguous
48 Number of objects in the block depends on how many objects fit in
49 RTE_DRIVER_MEMPOOL_BUCKET_SIZE_KB memory chunk which is build time option.
50 The number may be obtained using rte_mempool_ops_get_info() API.
51 Contiguous blocks may be allocated using rte_mempool_get_contig_blocks() API.
53 * **Added PMD-recommended Tx and Rx parameters**
55 Applications can now query drivers for device-tuned values of
56 ring sizes, burst sizes, and number of queues.
58 * **Added RSS hash and key update to CXGBE PMD.**
60 Support to update RSS hash and key has been added to CXGBE PMD.
62 * **Added CXGBE VF PMD.**
64 CXGBE VF Poll Mode Driver has been added to run DPDK over Chelsio
65 T5/T6 NIC VF instances.
67 * **Updated Solarflare network PMD.**
69 Updated the sfc_efx driver including the following changes:
71 * Added support for Solarflare XtremeScale X2xxx family adapters.
72 * Added support for NVGRE, VXLAN and GENEVE filters in flow API.
73 * Added support for DROP action in flow API.
74 * Added support for equal stride super-buffer Rx mode (X2xxx only).
75 * Added support for MARK and FLAG actions in flow API (X2xxx only).
77 * **Added Ethernet poll mode driver for AMD XGBE devices.**
79 Added the new ``axgbe`` ethernet poll mode driver for AMD XGBE devices.
80 See the :doc:`../nics/axgbe` nic driver guide for more details on this
83 * **Updated szedata2 PMD.**
85 Added support for new NFB-200G2QL card.
86 New API was introduced in the libsze2 library which the szedata2 PMD depends
87 on thus the new version of the library was needed.
88 New versions of the packages are available and the minimum required version
91 * **Added support for Broadcom NetXtreme-S (BCM58800) family of controllers (aka Stingray)**
93 The BCM58800 devices feature a NetXtreme E-Series advanced network controller, a high-performance
94 ARM CPU block, PCI Express (PCIe) Gen3 interfaces, key accelerators for compute offload and a high-
95 speed memory subsystem including L3 cache and DDR4 interfaces, all interconnected by a coherent
96 Network-on-chip (NOC) fabric.
98 The ARM CPU subsystem features eight ARMv8 Cortex-A72 CPUs at 3.0 GHz, arranged in a multi-cluster
101 * **Added IFCVF vDPA driver.**
103 Added IFCVF vDPA driver to support Intel FPGA 100G VF device. IFCVF works
104 as a HW vhost data path accelerator, it supports live migration and is
105 compatible with virtio 0.95 and 1.0. This driver registers ifcvf vDPA driver
106 to vhost lib, when virtio connected, with the help of the registered vDPA
107 driver the assigned VF gets configured to Rx/Tx directly to VM's virtio
110 * **Added support for virtio-user server mode.**
111 In a container environment if the vhost-user backend restarts, there's no way
112 for it to reconnect to virtio-user. To address this, support for server mode
113 is added. In this mode the socket file is created by virtio-user, which the
114 backend connects to. This means that if the backend restarts, it can reconnect
115 to virtio-user and continue communications.
117 * **Added crypto workload support to vhost library.**
119 New APIs are introduced in vhost library to enable virtio crypto support
120 including session creation/deletion handling and translating virtio-crypto
121 request into DPDK crypto operations. A sample application is also introduced.
123 * **Added virtio crypto PMD.**
125 Added a new poll mode driver for virtio crypto devices, which provides
126 AES-CBC ciphering and AES-CBC with HMAC-SHA1 algorithm-chaining. See the
127 :doc:`../cryptodevs/virtio` crypto driver guide for more details on
130 * **Added AMD CCP Crypto PMD.**
132 Added the new ``ccp`` crypto driver for AMD CCP devices. See the
133 :doc:`../cryptodevs/ccp` crypto driver guide for more details on
136 * **Updated AESNI MB PMD.**
138 The AESNI MB PMD has been updated with additional support for:
140 * AES-CMAC (128-bit key).
142 * **Added the Event Timer Adapter Library.**
144 The Event Timer Adapter Library extends the event-based model by introducing
145 APIs that allow applications to arm/cancel event timers that generate
146 timer expiry events. This new type of event is scheduled by an event device
147 along with existing types of events.
149 * **Added DPAA2 QDMA Driver (in rawdev).**
151 The DPAA2 QDMA is an implementation of the rawdev API, that provide means
152 to initiate a DMA transaction from CPU. The initiated DMA is performed
153 without CPU being involved in the actual DMA transaction.
155 See the :doc:`../rawdevs/dpaa2_qdma` guide for more details.
157 * **Added device event monitor framework.**
159 Added a general device event monitor framework at EAL, for device dynamic management.
160 Such as device hotplug awareness and actions adopted accordingly. The list of new APIs:
162 * ``rte_dev_event_monitor_start`` and ``rte_dev_event_monitor_stop`` are for
163 the event monitor enable and disable.
164 * ``rte_dev_event_callback_register`` and ``rte_dev_event_callback_unregister``
165 are for the user's callbacks register and unregister.
167 Linux uevent is supported as backend of this device event notification framework.
169 * **Added support for procinfo and pdump on eth vdev.**
171 For ethernet virtual devices (like tap, pcap, etc), with this feature, we can get
172 stats/xstats on shared memory from secondary process, and also pdump packets on
173 those virtual devices.
179 .. This section should contain API changes. Sample format:
181 * Add a short 1-2 sentence description of the API change. Use fixed width
182 quotes for ``rte_function_names`` or ``rte_struct_names``. Use the past
185 This section is a comment. Do not overwrite or remove it.
186 Also, make sure to start the actual text at the margin.
187 =========================================================
189 * service cores: no longer marked as experimental.
191 The service cores functions are no longer marked as experimental, and have
192 become part of the normal DPDK API and ABI. Any future ABI changes will be
193 announced at least one release before the ABI change is made. There are no
194 ABI breaking changes planned.
196 * eal: ``rte_lcore_has_role()`` return value changed.
198 This function now returns true or false, respectively,
199 rather than 0 or <0 for success or failure.
200 It makes use of the function more intuitive.
202 * mempool: capability flags and related functions have been removed.
204 Flags ``MEMPOOL_F_CAPA_PHYS_CONTIG`` and
205 ``MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS`` were used by octeontx mempool
206 driver to customize generic mempool library behaviour.
207 Now the new driver callbacks ``calc_mem_size`` and ``populate`` may be
208 used to achieve it without specific knowledge in the generic code.
210 * mempool: xmem functions have been deprecated:
212 - ``rte_mempool_xmem_create``
213 - ``rte_mempool_xmem_size``
214 - ``rte_mempool_xmem_usage``
215 - ``rte_mempool_populate_iova_tab``
217 * mbuf: The control mbuf API has been removed in v18.05. The impacted
218 functions and macros are:
220 - ``rte_ctrlmbuf_init()``
221 - ``rte_ctrlmbuf_alloc()``
222 - ``rte_ctrlmbuf_free()``
223 - ``rte_ctrlmbuf_data()``
224 - ``rte_ctrlmbuf_len()``
225 - ``rte_is_ctrlmbuf()``
228 The packet mbuf API should be used as a replacement.
230 * meter: updated to accommodate configuration profiles.
232 The meter API is changed to support meter configuration profiles. The
233 configuration profile represents the set of configuration parameters
234 for a given meter object, such as the rates and sizes for the token
235 buckets. These configuration parameters were previously the part of meter
236 object internal data strcuture. The separation of the configuration
237 parameters from meter object data structure results in reducing its
238 memory footprint which helps in better cache utilization when large number
239 of meter objects are used.
241 * ethdev: The function ``rte_eth_dev_count``, often mis-used to iterate
242 over ports, is deprecated and replaced by ``rte_eth_dev_count_avail``.
243 There is also a new function ``rte_eth_dev_count_total`` to get the
244 total number of allocated ports, available or not.
245 The hotplug-proof applications should use ``RTE_ETH_FOREACH_DEV`` or
246 ``RTE_ETH_FOREACH_DEV_OWNED_BY`` as port iterators.
248 * ethdev, in struct ``struct rte_eth_dev_info``, field ``rte_pci_device *pci_dev``
249 replaced with field ``struct rte_device *device``.
251 * **Changes to semantics of rte_eth_dev_configure() parameters.**
253 If both the ``nb_rx_q`` and ``nb_tx_q`` parameters are zero,
254 ``rte_eth_dev_configure`` will now use PMD-recommended queue sizes, or if
255 recommendations are not provided by the PMD the function will use ethdev
256 fall-back values. Previously setting both of the parameters to zero would
257 have resulted in ``-EINVAL`` being returned.
259 * **Changes to semantics of rte_eth_rx_queue_setup() parameters.**
261 If the ``nb_rx_desc`` parameter is zero, ``rte_eth_rx_queue_setup`` will
262 now use the PMD-recommended Rx ring size, or in the case where the PMD
263 does not provide a recommendation, will use an ethdev-provided
264 fall-back value. Previously, setting ``nb_rx_desc`` to zero would have
265 resulted in an error.
267 * **Changes to semantics of rte_eth_tx_queue_setup() parameters.**
269 If the ``nb_tx_desc`` parameter is zero, ``rte_eth_tx_queue_setup`` will
270 now use the PMD-recommended Tx ring size, or in the case where the PMD
271 does not provide a recoomendation, will use an ethdev-provided
272 fall-back value. Previously, setting ``nb_tx_desc`` to zero would have
273 resulted in an error.
275 * ethdev: several changes were made to the flow API.
277 * Unused DUP action was removed.
278 * Actions semantics in flow rules: list order now matters ("first
279 to last" instead of "all simultaneously"), repeated actions are now
280 all performed, and they do not individually have (non-)terminating
282 * Flow rules are now always terminating unless a PASSTHRU action is
284 * C99-style flexible arrays were replaced with standard pointers in RSS
285 action and in RAW pattern item structures due to compatibility issues.
286 * The RSS action was modified to not rely on external
287 ``struct rte_eth_rss_conf`` anymore to instead expose its own and more
288 appropriately named configuration fields directly
289 (``rss_conf->rss_key`` => ``key``,
290 ``rss_conf->rss_key_len`` => ``key_len``,
291 ``rss_conf->rss_hf`` => ``types``,
292 ``num`` => ``queue_num``), and the addition of missing RSS parameters
293 (``func`` for RSS hash function to apply and ``level`` for the
294 encapsulation level).
295 * The VLAN pattern item (``struct rte_flow_item_vlan``) was modified to
296 include inner EtherType instead of outer TPID. Its default mask was also
297 modified to cover the VID part (lower 12 bits) of TCI only.
298 * A new transfer attribute was added to ``struct rte_flow_attr`` in order
299 to clarify the behavior of some pattern items.
300 * PF and VF pattern items are now only accepted by PMDs that implement
301 them (bnxt and i40e) when the transfer attribute is also present for
303 * Pattern item PORT was renamed PHY_PORT to avoid confusion with DPDK port
305 * An action counterpart to the PHY_PORT pattern item was added in order to
306 redirect matching traffic to a specific physical port.
307 * PORT_ID pattern item and actions were added to match and target DPDK
308 port IDs at a higher level than PHY_PORT.
310 * ethdev: change flow APIs regarding count action:
311 * ``rte_flow_create()`` API count action now requires the ``struct rte_flow_action_count``.
312 * ``rte_flow_query()`` API parameter changed from action type to action structure.
318 .. This section should contain ABI changes. Sample format:
320 * Add a short 1-2 sentence description of the ABI change that was announced
321 in the previous releases and made in this release. Use fixed width quotes
322 for ``rte_function_names`` or ``rte_struct_names``. Use the past tense.
324 This section is a comment. Do not overwrite or remove it.
325 Also, make sure to start the actual text at the margin.
326 =========================================================
328 * ring: the alignment constraints on the ring structure has been relaxed
329 to one cache line instead of two, and an empty cache line padding is
330 added between the producer and consumer structures. The size of the
331 structure and the offset of the fields remains the same on platforms
332 with 64B cache line, but change on other platforms.
334 * mempool: ops have changed.
336 A new callback ``calc_mem_size`` has been added to ``rte_mempool_ops``
337 to allow to customize required memory size calculation.
338 A new callback ``populate`` has been added to ``rte_mempool_ops``
339 to allow to customize objects population.
340 Callback ``get_capabilities`` has been removed from ``rte_mempool_ops``
341 since its features are covered by ``calc_mem_size`` and ``populate``
343 Callback ``register_memory_area`` has been removed from ``rte_mempool_ops``
344 since the new callback ``populate`` may be used instead of it.
346 * **Additional fields in rte_eth_dev_info.**
348 The ``rte_eth_dev_info`` structure has had two extra entries appended to the
349 end of it: ``default_rxportconf`` and ``default_txportconf``. Each of these
350 in turn are ``rte_eth_dev_portconf`` structures containing three fields of
351 type ``uint16_t``: ``burst_size``, ``ring_size``, and ``nb_queues``. These
352 are parameter values recommended for use by the PMD.
354 * ethdev: ABI for all flow API functions was updated.
356 This includes functions ``rte_flow_copy``, ``rte_flow_create``,
357 ``rte_flow_destroy``, ``rte_flow_error_set``, ``rte_flow_flush``,
358 ``rte_flow_isolate``, ``rte_flow_query`` and ``rte_flow_validate``, due to
359 changes in error type definitions (``enum rte_flow_error_type``), removal
360 of the unused DUP action (``enum rte_flow_action_type``), modified
361 behavior for flow rule actions (see API changes), removal of C99 flexible
362 array from RAW pattern item (``struct rte_flow_item_raw``), complete
363 rework of the RSS action definition (``struct rte_flow_action_rss``),
364 sanity fix in the VLAN pattern item (``struct rte_flow_item_vlan``) and
365 new transfer attribute (``struct rte_flow_attr``).
371 .. This section should contain removed items in this release. Sample format:
373 * Add a short 1-2 sentence description of the removed item in the past
376 This section is a comment. Do not overwrite or remove it.
377 Also, make sure to start the actual text at the margin.
378 =========================================================
384 .. This section should contain new known issues in this release. Sample format:
386 * **Add title in present tense with full stop.**
388 Add a short 1-2 sentence description of the known issue in the present
389 tense. Add information on any known workarounds.
391 This section is a comment. Do not overwrite or remove it.
392 Also, make sure to start the actual text at the margin.
393 =========================================================
395 * **pdump is not compatible with old applications.**
397 As we changed to use generic multi-process communication for pdump negotiation
398 instead of previous dedicated unix socket way, pdump applications, including
399 dpdk-pdump example and any other applications using librte_pdump, cannot work
400 with older version DPDK primary applications.
403 Shared Library Versions
404 -----------------------
406 .. Update any library version updated in this release and prepend with a ``+``
410 + librte_cfgfile.so.2
413 This section is a comment. Do not overwrite or remove it.
414 =========================================================
417 The libraries prepended with a plus sign were incremented in this version.
423 librte_bitratestats.so.2
425 librte_bus_fslmc.so.1
430 + librte_common_octeontx.so.1
431 librte_cryptodev.so.4
432 librte_distributor.so.1
436 librte_flow_classify.so.1
444 librte_latencystats.so.1
447 + librte_mempool.so.4
457 librte_pmd_ixgbe.so.2
458 + librte_pmd_dpaa2_qdma.so.1
460 librte_pmd_softnic.so.1
461 librte_pmd_vhost.so.2
477 .. This section should contain a list of platforms that were tested with this
482 * <vendor> platform with <vendor> <type of devices> combinations
487 * Other relevant details...
489 This section is a comment. Do not overwrite or remove it.
490 Also, make sure to start the actual text at the margin.
491 =========================================================