1.. SPDX-License-Identifier: BSD-3-Clause 2 Copyright 2016 The DPDK contributors 3 4DPDK Release 16.04 5================== 6 7New Features 8------------ 9 10* **Added function to check primary process state.** 11 12 A new function ``rte_eal_primary_proc_alive()`` has been added 13 to allow the user to detect if a primary process is running. 14 Use cases for this feature include fault detection, and monitoring 15 using secondary processes. 16 17* **Enabled bulk allocation of mbufs.** 18 19 A new function ``rte_pktmbuf_alloc_bulk()`` has been added to allow the user 20 to bulk allocate mbufs. 21 22* **Added device link speed capabilities.** 23 24 The structure ``rte_eth_dev_info`` now has a ``speed_capa`` bitmap, which 25 allows the application to determine the supported speeds of each device. 26 27* **Added bitmap of link speeds to advertise.** 28 29 Added a feature to allow the definition of a set of advertised speeds for auto-negotiation, 30 explicitly disabling link auto-negotiation (single speed) 31 and full auto-negotiation. 32 33* **Added new poll-mode driver for Amazon Elastic Network Adapters (ENA).** 34 35 The driver operates for a variety of ENA adapters through feature negotiation 36 with the adapter and upgradable commands set. 37 The ENA driver handles PCI Physical and Virtual ENA functions. 38 39* **Restored vmxnet3 TX data ring.** 40 41 TX data ring has been shown to improve small packet forwarding performance 42 on the vSphere environment. 43 44* **Added vmxnet3 TX L4 checksum offload.** 45 46 Added support for TCP/UDP checksum offload to vmxnet3. 47 48* **Added vmxnet3 TSO support.** 49 50 Added support for TSO to vmxnet3. 51 52* **Added vmxnet3 support for jumbo frames.** 53 54 Added support for linking multi-segment buffers together to 55 handle Jumbo packets. 56 57* **Enabled Virtio 1.0 support.** 58 59 Enabled Virtio 1.0 support for Virtio PMD. 60 61* **Supported Virtio for ARM.** 62 63 Enabled Virtio support for ARMv7/v8. Tested for ARM64. 64 Virtio for ARM supports VFIO-noiommu mode only. 65 Virtio can work with other non-x86 architectures as well, like PowerPC. 66 67* **Supported Virtio offload in vhost-user.** 68 69 Added the offload and negotiation of checksum and TSO between vhost-user and 70 vanilla Linux Virtio guest. 71 72* **Added vhost-user live migration support.** 73 74* **Added vhost driver.** 75 76 Added a virtual PMD that wraps ``librte_vhost``. 77 78* **Added multicast promiscuous mode support on VF for ixgbe.** 79 80 Added multicast promiscuous mode support for the ixgbe VF driver so all VFs 81 can receive the multicast packets. 82 83 Please note if you want to use this promiscuous mode, you need both PF and VF 84 driver to support it. The reason is that this VF feature is configured in the PF. 85 If you use kernel PF driver and the dpdk VF driver, make sure the kernel PF driver supports 86 VF multicast promiscuous mode. If you use dpdk PF and dpdk VF ensure the PF 87 driver is the same version as the VF. 88 89* **Added support for E-tag on X550.** 90 91 E-tag is defined in `802.1BR - Bridge Port Extension <http://www.ieee802.org/1/pages/802.1br.html>`_. 92 93 This feature is for the VF, but the settings are on the PF. It means 94 the CLIs should be used on the PF, but some of their effects will be shown on the VF. 95 The forwarding of E-tag packets based on GRP and E-CID_base will have an effect 96 on the PF. Theoretically, the E-tag packets can be forwarded to any pool/queue 97 but normally we'd like to forward the packets to the pools/queues belonging 98 to the VFs. And E-tag insertion and stripping will have an effect on VFs. When 99 a VF receives E-tag packets it should strip the E-tag. When the VF transmits 100 packets, it should insert the E-tag. Both actions can be offloaded. 101 102 When we want to use this E-tag support feature, the forwarding should be 103 enabled to forward the packets received by the PF to the indicated VFs. And insertion 104 and stripping should be enabled for VFs to offload the effort to hardware. 105 106 Features added: 107 108 * Support E-tag offloading of insertion and stripping. 109 * Support Forwarding E-tag packets to pools based on 110 GRP and E-CID_base. 111 112* **Added support for VxLAN and NVGRE checksum off-load on X550.** 113 114 * Added support for VxLAN and NVGRE RX/TX checksum off-load on 115 X550. RX/TX checksum off-load is provided on both inner and 116 outer IP header and TCP header. 117 * Added functions to support VxLAN port configuration. The 118 default VxLAN port number is 4789 but this can be updated 119 programmatically. 120 121* **Added support for new X550EM_a devices.** 122 123 Added support for new X550EM_a devices and their MAC types, X550EM_a and X550EM_a_vf. 124 Updated the relevant PMD to use the new devices and MAC types. 125 126* **Added x550em_x V2 device support.** 127 128 Added support for x550em_x V2 device. Only x550em_x V1 was supported before. 129 A mask for V1 and V2 is defined and used to support both. 130 131* **Supported link speed auto-negotiation on X550EM_X** 132 133 Normally the auto-negotiation is supported by firmware and software doesn't care about 134 it. But on x550em_x, firmware doesn't support auto-negotiation. As the ports of x550em_x 135 are 10GbE, if we connect the port with a peer which is 1GbE, the link will always 136 be down. 137 We added the support for auto-negotiation by software to avoid this link down issue. 138 139* **Added software-firmware sync on X550EM_a.** 140 141 Added support for software-firmware sync for resource sharing. 142 Use the PHY token, shared between software-firmware for PHY access on X550EM_a. 143 144* **Updated the i40e base driver.** 145 146 The i40e base driver was updated with changes including the 147 following: 148 149 * Use RX control AQ commands to read/write RX control registers. 150 * Add new X722 device IDs, and removed X710 one was never used. 151 * Expose registers for HASH/FD input set configuring. 152 153* **Enabled PCI extended tag for i40e.** 154 155 Enabled extended tag for i40e by checking and writing corresponding PCI config 156 space bytes, to boost the performance. 157 The legacy method of reading/writing sysfile supported by kernel module igb_uio 158 is now deprecated. 159 160* **Added i40e support for setting mac addresses.** 161 162* **Added dump of i40e registers and EEPROM.** 163 164* **Supported ether type setting of single and double VLAN for i40e** 165 166* **Added VMDQ DCB mode in i40e.** 167 168 Added support for DCB in VMDQ mode to i40e driver. 169 170* **Added i40e VEB switching support.** 171 172* **Added Flow director enhancements in i40e.** 173 174* **Added PF reset event reporting in i40e VF driver.** 175 176* **Added fm10k RX interrupt support.** 177 178* **Optimized fm10k TX.** 179 180 Optimized fm10k TX by freeing multiple mbufs at a time. 181 182* **Handled error flags in fm10k vector RX.** 183 184 Parse error flags in RX descriptor and set error bits in mbuf with vector instructions. 185 186* **Added fm10k FTAG based forwarding support.** 187 188* **Added mlx5 flow director support.** 189 190 Added flow director support (``RTE_FDIR_MODE_PERFECT`` and 191 ``RTE_FDIR_MODE_PERFECT_MAC_VLAN``). 192 193 Only available with Mellanox OFED >= 3.2. 194 195* **Added mlx5 RX VLAN stripping support.** 196 197 Added support for RX VLAN stripping. 198 199 Only available with Mellanox OFED >= 3.2. 200 201* **Added mlx5 link up/down callbacks.** 202 203 Implemented callbacks to bring link up and down. 204 205* **Added mlx5 support for operation in secondary processes.** 206 207 Implemented TX support in secondary processes (like mlx4). 208 209* **Added mlx5 RX CRC stripping configuration.** 210 211 Until now, CRC was always stripped. It can now be configured. 212 213 Only available with Mellanox OFED >= 3.2. 214 215* **Added mlx5 optional packet padding by HW.** 216 217 Added an option to make PCI bus transactions rounded to a multiple of a 218 cache line size for better alignment. 219 220 Only available with Mellanox OFED >= 3.2. 221 222* **Added mlx5 TX VLAN insertion support.** 223 224 Added support for TX VLAN insertion. 225 226 Only available with Mellanox OFED >= 3.2. 227 228* **Changed szedata2 driver type from vdev to pdev.** 229 230 Previously szedata2 device had to be added by ``--vdev`` option. 231 Now szedata2 PMD recognizes the device automatically during EAL 232 initialization. 233 234* **Added szedata2 functions for setting link up/down.** 235 236* **Added szedata2 promiscuous and allmulticast modes.** 237 238* **Added af_packet dynamic removal function.** 239 240 An af_packet device can now be detached using the API, like other PMD devices. 241 242* **Increased number of next hops for LPM IPv4 to 2^24.** 243 244 The ``next_hop`` field has been extended from 8 bits to 24 bits for IPv4. 245 246* **Added support of SNOW 3G (UEA2 and UIA2) for Intel Quick Assist devices.** 247 248 Enabled support for the SNOW 3G wireless algorithm for Intel Quick Assist devices. 249 Support for cipher-only and hash-only is also provided 250 along with algorithm-chaining operations. 251 252* **Added SNOW3G SW PMD.** 253 254 A new Crypto PMD has been added, which provides SNOW 3G UEA2 ciphering 255 and SNOW3G UIA2 hashing. 256 257* **Added AES GCM PMD.** 258 259 Added new Crypto PMD to support AES-GCM authenticated encryption and 260 authenticated decryption in software. 261 262* **Added NULL Crypto PMD** 263 264 Added new Crypto PMD to support null crypto operations in software. 265 266* **Improved IP Pipeline Application.** 267 268 The following features have been added to ip_pipeline application; 269 270 * Added CPU utilization measurement and idle cycle rate computation. 271 * Added link identification support through existing port-mask option or by 272 specifying PCI device in every LINK section in the configuration file. 273 * Added load balancing support in passthrough pipeline. 274 275* **Added IPsec security gateway example.** 276 277 Added a new application implementing an IPsec Security Gateway. 278 279 280Resolved Issues 281--------------- 282 283Drivers 284~~~~~~~ 285 286* **ethdev: Fixed overflow for 100Gbps.** 287 288 100Gbps in Mbps (100000) was exceeding the 16-bit max value of ``link_speed`` 289 in ``rte_eth_link``. 290 291* **ethdev: Fixed byte order consistency between fdir flow and mask.** 292 293 Fixed issue in ethdev library where the structure for setting 294 fdir's mask and flow entry was not consistent in byte ordering. 295 296* **cxgbe: Fixed crash due to incorrect size allocated for RSS table.** 297 298 Fixed a segfault that occurs when accessing part of port 0's RSS 299 table that gets overwritten by subsequent port 1's part of the RSS 300 table due to incorrect size allocated for each entry in the table. 301 302* **cxgbe: Fixed setting wrong device MTU.** 303 304 Fixed an incorrect device MTU being set due to the Ethernet header and 305 CRC lengths being added twice. 306 307* **ixgbe: Fixed zeroed VF mac address.** 308 309 Resolved an issue where the VF MAC address is zeroed out in cases where the VF 310 driver is loaded while the PF interface is down. 311 The solution is to only set it when we get an ACK from the PF. 312 313* **ixgbe: Fixed setting flow director flag twice.** 314 315 Resolved an issue where packets were being dropped when switching to perfect 316 filters mode. 317 318* **ixgbe: Set MDIO speed after MAC reset.** 319 320 The MDIO clock speed must be reconfigured after the MAC reset. The MDIO clock 321 speed becomes invalid, therefore the driver reads invalid PHY register values. 322 The driver now set the MDIO clock speed prior to initializing PHY ops and 323 again after the MAC reset. 324 325* **ixgbe: Fixed maximum number of available TX queues.** 326 327 In IXGBE, the maximum number of TX queues varies depending on the NIC operating 328 mode. This was not being updated in the device information, providing 329 an incorrect number in some cases. 330 331* **i40e: Generated MAC address for each VFs.** 332 333 It generates a MAC address for each VFs during PF host initialization, 334 and keeps the VF MAC address the same among different VF launch. 335 336* **i40e: Fixed failure of reading/writing RX control registers.** 337 338 Fixed i40e issue of failing to read/write rx control registers when 339 under stress with traffic, which might result in application launch 340 failure. 341 342* **i40e: Enabled vector driver by default.** 343 344 Previously, vector driver was disabled by default as it couldn't fill packet type 345 info for l3fwd to work well. Now there is an option for l3fwd to analyze 346 the packet type so the vector driver is enabled by default. 347 348* **i40e: Fixed link info of VF.** 349 350 Previously, the VF's link speed stayed at 10GbE and status always was up. 351 It did not change even when the physical link's status changed. 352 Now this issue is fixed to make VF's link info consistent with physical link. 353 354* **mlx5: Fixed possible crash during initialization.** 355 356 A crash could occur when failing to allocate private device context. 357 358* **mlx5: Added port type check.** 359 360 Added port type check to prevent port initialization on non-Ethernet link layers and 361 to report an error. 362 363* **mlx5: Applied VLAN filtering to broadcast and IPv6 multicast flows.** 364 365 Prevented reception of multicast frames outside of configured VLANs. 366 367* **mlx5: Fixed RX checksum offload in non L3/L4 packets.** 368 369 Fixed report of bad checksum for packets of unknown type. 370 371* **aesni_mb: Fixed wrong return value when creating a device.** 372 373 The ``cryptodev_aesni_mb_init()`` function was returning the device id of the device created, 374 instead of 0 (on success) that ``rte_eal_vdev_init()`` expects. 375 This made it impossible to create more than one aesni_mb device 376 from the command line. 377 378* **qat: Fixed AES GCM decryption.** 379 380 Allowed AES GCM on the cryptodev API, but in some cases gave invalid results 381 due to incorrect IV setting. 382 383 384Libraries 385~~~~~~~~~ 386 387* **hash: Fixed CRC32c hash computation for non multiple of 4 bytes sizes.** 388 389 Fix crc32c hash functions to return a valid crc32c value for data lengths 390 not a multiple of 4 bytes. 391 392* **hash: Fixed hash library to support multi-process mode.** 393 394 Fix hash library to support multi-process mode, using a jump table, 395 instead of storing a function pointer to the key compare function. 396 Multi-process mode only works with the built-in compare functions, 397 however a custom compare function (not in the jump table) can only 398 be used in single-process mode. 399 400* **hash: Fixed return value when allocating an existing hash table.** 401 402 Changed the ``rte_hash*_create()`` functions to return ``NULL`` and set 403 ``rte_errno`` to ``EEXIST`` when the object name already exists. This is 404 the behavior described in the API documentation in the header file. 405 The previous behavior was to return a pointer to the existing object in 406 that case, preventing the caller from knowing if the object had to be freed 407 or not. 408 409* **lpm: Fixed return value when allocating an existing object.** 410 411 Changed the ``rte_lpm*_create()`` functions to return ``NULL`` and set 412 ``rte_errno`` to ``EEXIST`` when the object name already exists. This is 413 the behavior described in the API documentation in the header file. 414 The previous behavior was to return a pointer to the existing object in 415 that case, preventing the caller from knowing if the object had to be freed 416 or not. 417 418* **librte_port: Fixed segmentation fault for ring and ethdev writer nodrop.** 419 420 Fixed core dump issue on txq and swq when dropless is set to yes. 421 422 423Examples 424~~~~~~~~ 425 426* **l3fwd-power: Fixed memory leak for non-IP packet.** 427 428 Fixed issue in l3fwd-power where, on receiving packets of types 429 other than IPv4 or IPv6, the mbuf was not released, and caused 430 a memory leak. 431 432* **l3fwd: Fixed using packet type blindly.** 433 434 l3fwd makes use of packet type information without querying if devices or PMDs 435 really set it. For those devices that don't set ptypes, add an option to parse it. 436 437* **examples/vhost: Fixed frequent mbuf allocation failure.** 438 439 The vhost-switch often fails to allocate mbuf when dequeue from vring because it 440 wrongly calculates the number of mbufs needed. 441 442 443API Changes 444----------- 445 446* The ethdev statistics counter ``imissed`` is considered to be independent of ``ierrors``. 447 All drivers are now counting the missed packets only once, i.e. drivers will 448 not increment ierrors anymore for missed packets. 449 450* The ethdev structure ``rte_eth_dev_info`` was changed to support device 451 speed capabilities. 452 453* The ethdev structures ``rte_eth_link`` and ``rte_eth_conf`` were changed to 454 support the new link API. 455 456* The functions ``rte_eth_dev_udp_tunnel_add`` and ``rte_eth_dev_udp_tunnel_delete`` 457 have been renamed into ``rte_eth_dev_udp_tunnel_port_add`` and 458 ``rte_eth_dev_udp_tunnel_port_delete``. 459 460* The ``outer_mac`` and ``inner_mac`` fields in structure 461 ``rte_eth_tunnel_filter_conf`` are changed from pointer to struct in order 462 to keep code's readability. 463 464* The fields in ethdev structure ``rte_eth_fdir_masks`` were changed 465 to be in big endian. 466 467* A parameter ``vlan_type`` has been added to the function 468 ``rte_eth_dev_set_vlan_ether_type``. 469 470* The af_packet device init function is no longer public. The device should be attached 471 via the API. 472 473* The LPM ``next_hop`` field is extended from 8 bits to 24 bits for IPv4 474 while keeping ABI compatibility. 475 476* A new ``rte_lpm_config`` structure is used so the LPM library will allocate 477 exactly the amount of memory which is necessary to hold application’s rules. 478 The previous ABI is kept for compatibility. 479 480* The prototype for the pipeline input port, output port and table action 481 handlers are updated: the pipeline parameter is added, 482 the packets mask parameter has been either removed or made input-only. 483 484 485ABI Changes 486----------- 487 488* The RETA entry size in ``rte_eth_rss_reta_entry64`` has been increased 489 from 8-bit to 16-bit. 490 491* The ethdev flow director structure ``rte_eth_fdir_flow`` structure was 492 changed. New fields were added to extend flow director's input set. 493 494* The cmdline buffer size has been increase from 256 to 512. 495 496 497Shared Library Versions 498----------------------- 499 500The libraries prepended with a plus sign were incremented in this version. 501 502.. code-block:: diff 503 504 + libethdev.so.3 505 librte_acl.so.2 506 librte_cfgfile.so.2 507 + librte_cmdline.so.2 508 librte_distributor.so.1 509 librte_eal.so.2 510 librte_hash.so.2 511 librte_ip_frag.so.1 512 librte_ivshmem.so.1 513 librte_jobstats.so.1 514 librte_kni.so.2 515 librte_kvargs.so.1 516 librte_lpm.so.2 517 librte_mbuf.so.2 518 librte_mempool.so.1 519 librte_meter.so.1 520 + librte_pipeline.so.3 521 librte_pmd_bond.so.1 522 librte_pmd_ring.so.2 523 librte_port.so.2 524 librte_power.so.1 525 librte_reorder.so.1 526 librte_ring.so.1 527 librte_sched.so.1 528 librte_table.so.2 529 librte_timer.so.1 530 librte_vhost.so.2 531 532 533Tested Platforms 534---------------- 535 536#. SuperMicro 1U 537 538 - BIOS: 1.0c 539 - Processor: Intel(R) Atom(TM) CPU C2758 @ 2.40GHz 540 541#. SuperMicro 1U 542 543 - BIOS: 1.0a 544 - Processor: Intel(R) Xeon(R) CPU D-1540 @ 2.00GHz 545 - Onboard NIC: Intel(R) X552/X557-AT (2x10G) 546 547 - Firmware-version: 0x800001cf 548 - Device ID (PF/VF): 8086:15ad /8086:15a8 549 550 - kernel driver version: 4.2.5 (ixgbe) 551 552#. SuperMicro 1U 553 554 - BIOS: 1.0a 555 - Processor: Intel(R) Xeon(R) CPU E5-4667 v3 @ 2.00GHz 556 557#. Intel(R) Server board S2600GZ 558 559 - BIOS: SE5C600.86B.02.02.0002.122320131210 560 - Processor: Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz 561 562#. Intel(R) Server board W2600CR 563 564 - BIOS: SE5C600.86B.02.01.0002.082220131453 565 - Processor: Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz 566 567#. Intel(R) Server board S2600CWT 568 569 - BIOS: SE5C610.86B.01.01.0009.060120151350 570 - Processor: Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz 571 572#. Intel(R) Server board S2600WTT 573 574 - BIOS: SE5C610.86B.01.01.0005.101720141054 575 - Processor: Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz 576 577#. Intel(R) Server board S2600WTT 578 579 - BIOS: SE5C610.86B.11.01.0044.090120151156 580 - Processor: Intel(R) Xeon(R) CPU E5-2695 v4 @ 2.10GHz 581 582 583Tested NICs 584----------- 585 586#. Intel(R) Ethernet Controller X540-AT2 587 588 - Firmware version: 0x80000389 589 - Device id (pf): 8086:1528 590 - Driver version: 3.23.2 (ixgbe) 591 592#. Intel(R) 82599ES 10 Gigabit Ethernet Controller 593 594 - Firmware version: 0x61bf0001 595 - Device id (pf/vf): 8086:10fb / 8086:10ed 596 - Driver version: 4.0.1-k (ixgbe) 597 598#. Intel(R) Corporation Ethernet Connection X552/X557-AT 10GBASE-T 599 600 - Firmware version: 0x800001cf 601 - Device id (pf/vf): 8086:15ad / 8086:15a8 602 - Driver version: 4.2.5 (ixgbe) 603 604#. Intel(R) Ethernet Converged Network Adapter X710-DA4 (4x10G) 605 606 - Firmware version: 5.02 0x80002284 607 - Device id (pf/vf): 8086:1572 / 8086:154c 608 - Driver version: 1.4.26 (i40e) 609 610#. Intel(R) Ethernet Converged Network Adapter X710-DA2 (2x10G) 611 612 - Firmware version: 5.02 0x80002282 613 - Device id (pf/vf): 8086:1572 / 8086:154c 614 - Driver version: 1.4.25 (i40e) 615 616#. Intel(R) Ethernet Converged Network Adapter XL710-QDA1 (1x40G) 617 618 - Firmware version: 5.02 0x80002281 619 - Device id (pf/vf): 8086:1584 / 8086:154c 620 - Driver version: 1.4.25 (i40e) 621 622#. Intel(R) Ethernet Converged Network Adapter XL710-QDA2 (2X40G) 623 624 - Firmware version: 5.02 0x80002285 625 - Device id (pf/vf): 8086:1583 / 8086:154c 626 - Driver version: 1.4.25 (i40e) 627 628#. Intel(R) 82576EB Gigabit Ethernet Controller 629 630 - Firmware version: 1.2.1 631 - Device id (pf): 8086:1526 632 - Driver version: 5.2.13-k (igb) 633 634#. Intel(R) Ethernet Controller I210 635 636 - Firmware version: 3.16, 0x80000500, 1.304.0 637 - Device id (pf): 8086:1533 638 - Driver version: 5.2.13-k (igb) 639 640#. Intel(R) Corporation I350 Gigabit Network Connection 641 642 - Firmware version: 1.48, 0x800006e7 643 - Device id (pf/vf): 8086:1521 / 8086:1520 644 - Driver version: 5.2.13-k (igb) 645 646 647#. Intel(R) Ethernet Multi-host Controller FM10000 648 649 - Firmware version: N/A 650 - Device id (pf/vf): 8086:15d0 651 - Driver version: 0.17.0.9 (fm10k) 652