1.. SPDX-License-Identifier: BSD-3-Clause 2 Copyright 2016 The DPDK contributors 3 4DPDK Release 16.04 5================== 6 7 8New Features 9------------ 10 11* **Added function to check primary process state.** 12 13 A new function ``rte_eal_primary_proc_alive()`` has been added 14 to allow the user to detect if a primary process is running. 15 Use cases for this feature include fault detection, and monitoring 16 using secondary processes. 17 18* **Enabled bulk allocation of mbufs.** 19 20 A new function ``rte_pktmbuf_alloc_bulk()`` has been added to allow the user 21 to bulk allocate mbufs. 22 23* **Added device link speed capabilities.** 24 25 The structure ``rte_eth_dev_info`` now has a ``speed_capa`` bitmap, which 26 allows the application to determine the supported speeds of each device. 27 28* **Added bitmap of link speeds to advertise.** 29 30 Added a feature to allow the definition of a set of advertised speeds for auto-negotiation, 31 explicitly disabling link auto-negotiation (single speed) 32 and full auto-negotiation. 33 34* **Added new poll-mode driver for Amazon Elastic Network Adapters (ENA).** 35 36 The driver operates for a variety of ENA adapters through feature negotiation 37 with the adapter and upgradable commands set. 38 The ENA driver handles PCI Physical and Virtual ENA functions. 39 40* **Restored vmxnet3 TX data ring.** 41 42 TX data ring has been shown to improve small packet forwarding performance 43 on the vSphere environment. 44 45* **Added vmxnet3 TX L4 checksum offload.** 46 47 Added support for TCP/UDP checksum offload to vmxnet3. 48 49* **Added vmxnet3 TSO support.** 50 51 Added support for TSO to vmxnet3. 52 53* **Added vmxnet3 support for jumbo frames.** 54 55 Added support for linking multi-segment buffers together to 56 handle Jumbo packets. 57 58* **Enabled Virtio 1.0 support.** 59 60 Enabled Virtio 1.0 support for Virtio pmd driver. 61 62* **Supported Virtio for ARM.** 63 64 Enabled Virtio support for ARMv7/v8. Tested for ARM64. 65 Virtio for ARM supports VFIO-noiommu mode only. 66 Virtio can work with other non-x86 architectures as well, like PowerPC. 67 68* **Supported Virtio offload in vhost-user.** 69 70 Added the offload and negotiation of checksum and TSO between vhost-user and 71 vanilla Linux Virtio guest. 72 73* **Added vhost-user live migration support.** 74 75* **Added vhost driver.** 76 77 Added a virtual PMD that wraps ``librte_vhost``. 78 79* **Added multicast promiscuous mode support on VF for ixgbe.** 80 81 Added multicast promiscuous mode support for the ixgbe VF driver so all VFs 82 can receive the multicast packets. 83 84 Please note if you want to use this promiscuous mode, you need both PF and VF 85 driver to support it. The reason is that this VF feature is configured in the PF. 86 If you use kernel PF driver and the dpdk VF driver, make sure the kernel PF driver supports 87 VF multicast promiscuous mode. If you use dpdk PF and dpdk VF ensure the PF 88 driver is the same version as the VF. 89 90* **Added support for E-tag on X550.** 91 92 E-tag is defined in `802.1BR - Bridge Port Extension <http://www.ieee802.org/1/pages/802.1br.html>`_. 93 94 This feature is for the VF, but the settings are on the PF. It means 95 the CLIs should be used on the PF, but some of their effects will be shown on the VF. 96 The forwarding of E-tag packets based on GRP and E-CID_base will have an effect 97 on the PF. Theoretically, the E-tag packets can be forwarded to any pool/queue 98 but normally we'd like to forward the packets to the pools/queues belonging 99 to the VFs. And E-tag insertion and stripping will have an effect on VFs. When 100 a VF receives E-tag packets it should strip the E-tag. When the VF transmits 101 packets, it should insert the E-tag. Both actions can be offloaded. 102 103 When we want to use this E-tag support feature, the forwarding should be 104 enabled to forward the packets received by the PF to the indicated VFs. And insertion 105 and stripping should be enabled for VFs to offload the effort to hardware. 106 107 Features added: 108 109 * Support E-tag offloading of insertion and stripping. 110 * Support Forwarding E-tag packets to pools based on 111 GRP and E-CID_base. 112 113* **Added support for VxLAN and NVGRE checksum off-load on X550.** 114 115 * Added support for VxLAN and NVGRE RX/TX checksum off-load on 116 X550. RX/TX checksum off-load is provided on both inner and 117 outer IP header and TCP header. 118 * Added functions to support VxLAN port configuration. The 119 default VxLAN port number is 4789 but this can be updated 120 programmatically. 121 122* **Added support for new X550EM_a devices.** 123 124 Added support for new X550EM_a devices and their MAC types, X550EM_a and X550EM_a_vf. 125 Updated the relevant PMD to use the new devices and MAC types. 126 127* **Added x550em_x V2 device support.** 128 129 Added support for x550em_x V2 device. Only x550em_x V1 was supported before. 130 A mask for V1 and V2 is defined and used to support both. 131 132* **Supported link speed auto-negotiation on X550EM_X** 133 134 Normally the auto-negotiation is supported by firmware and software doesn't care about 135 it. But on x550em_x, firmware doesn't support auto-negotiation. As the ports of x550em_x 136 are 10GbE, if we connect the port with a peer which is 1GbE, the link will always 137 be down. 138 We added the support for auto-negotiation by software to avoid this link down issue. 139 140* **Added software-firmware sync on X550EM_a.** 141 142 Added support for software-firmware sync for resource sharing. 143 Use the PHY token, shared between software-firmware for PHY access on X550EM_a. 144 145* **Updated the i40e base driver.** 146 147 The i40e base driver was updated with changes including the 148 following: 149 150 * Use RX control AQ commands to read/write RX control registers. 151 * Add new X722 device IDs, and removed X710 one was never used. 152 * Expose registers for HASH/FD input set configuring. 153 154* **Enabled PCI extended tag for i40e.** 155 156 Enabled extended tag for i40e by checking and writing corresponding PCI config 157 space bytes, to boost the performance. 158 The legacy method of reading/writing sysfile supported by kernel module igb_uio 159 is now deprecated. 160 161* **Added i40e support for setting mac addresses.** 162 163* **Added dump of i40e registers and EEPROM.** 164 165* **Supported ether type setting of single and double VLAN for i40e** 166 167* **Added VMDQ DCB mode in i40e.** 168 169 Added support for DCB in VMDQ mode to i40e driver. 170 171* **Added i40e VEB switching support.** 172 173* **Added Flow director enhancements in i40e.** 174 175* **Added PF reset event reporting in i40e VF driver.** 176 177* **Added fm10k RX interrupt support.** 178 179* **Optimized fm10k TX.** 180 181 Optimized fm10k TX by freeing multiple mbufs at a time. 182 183* **Handled error flags in fm10k vector RX.** 184 185 Parse error flags in RX descriptor and set error bits in mbuf with vector instructions. 186 187* **Added fm10k FTAG based forwarding support.** 188 189* **Added mlx5 flow director support.** 190 191 Added flow director support (``RTE_FDIR_MODE_PERFECT`` and 192 ``RTE_FDIR_MODE_PERFECT_MAC_VLAN``). 193 194 Only available with Mellanox OFED >= 3.2. 195 196* **Added mlx5 RX VLAN stripping support.** 197 198 Added support for RX VLAN stripping. 199 200 Only available with Mellanox OFED >= 3.2. 201 202* **Added mlx5 link up/down callbacks.** 203 204 Implemented callbacks to bring link up and down. 205 206* **Added mlx5 support for operation in secondary processes.** 207 208 Implemented TX support in secondary processes (like mlx4). 209 210* **Added mlx5 RX CRC stripping configuration.** 211 212 Until now, CRC was always stripped. It can now be configured. 213 214 Only available with Mellanox OFED >= 3.2. 215 216* **Added mlx5 optional packet padding by HW.** 217 218 Added an option to make PCI bus transactions rounded to a multiple of a 219 cache line size for better alignment. 220 221 Only available with Mellanox OFED >= 3.2. 222 223* **Added mlx5 TX VLAN insertion support.** 224 225 Added support for TX VLAN insertion. 226 227 Only available with Mellanox OFED >= 3.2. 228 229* **Changed szedata2 driver type from vdev to pdev.** 230 231 Previously szedata2 device had to be added by ``--vdev`` option. 232 Now szedata2 PMD recognizes the device automatically during EAL 233 initialization. 234 235* **Added szedata2 functions for setting link up/down.** 236 237* **Added szedata2 promiscuous and allmulticast modes.** 238 239* **Added af_packet dynamic removal function.** 240 241 An af_packet device can now be detached using the API, like other PMD devices. 242 243* **Increased number of next hops for LPM IPv4 to 2^24.** 244 245 The ``next_hop`` field has been extended from 8 bits to 24 bits for IPv4. 246 247* **Added support of SNOW 3G (UEA2 and UIA2) for Intel Quick Assist devices.** 248 249 Enabled support for the SNOW 3G wireless algorithm for Intel Quick Assist devices. 250 Support for cipher-only and hash-only is also provided 251 along with algorithm-chaining operations. 252 253* **Added SNOW3G SW PMD.** 254 255 A new Crypto PMD has been added, which provides SNOW 3G UEA2 ciphering 256 and SNOW3G UIA2 hashing. 257 258* **Added AES GCM PMD.** 259 260 Added new Crypto PMD to support AES-GCM authenticated encryption and 261 authenticated decryption in software. 262 263* **Added NULL Crypto PMD** 264 265 Added new Crypto PMD to support null crypto operations in software. 266 267* **Improved IP Pipeline Application.** 268 269 The following features have been added to ip_pipeline application; 270 271 * Added CPU utilization measurement and idle cycle rate computation. 272 * Added link identification support through existing port-mask option or by 273 specifying PCI device in every LINK section in the configuration file. 274 * Added load balancing support in passthrough pipeline. 275 276* **Added IPsec security gateway example.** 277 278 Added a new application implementing an IPsec Security Gateway. 279 280 281Resolved Issues 282--------------- 283 284Drivers 285~~~~~~~ 286 287* **ethdev: Fixed overflow for 100Gbps.** 288 289 100Gbps in Mbps (100000) was exceeding the 16-bit max value of ``link_speed`` 290 in ``rte_eth_link``. 291 292* **ethdev: Fixed byte order consistency between fdir flow and mask.** 293 294 Fixed issue in ethdev library where the structure for setting 295 fdir's mask and flow entry was not consistent in byte ordering. 296 297* **cxgbe: Fixed crash due to incorrect size allocated for RSS table.** 298 299 Fixed a segfault that occurs when accessing part of port 0's RSS 300 table that gets overwritten by subsequent port 1's part of the RSS 301 table due to incorrect size allocated for each entry in the table. 302 303* **cxgbe: Fixed setting wrong device MTU.** 304 305 Fixed an incorrect device MTU being set due to the Ethernet header and 306 CRC lengths being added twice. 307 308* **ixgbe: Fixed zeroed VF mac address.** 309 310 Resolved an issue where the VF MAC address is zeroed out in cases where the VF 311 driver is loaded while the PF interface is down. 312 The solution is to only set it when we get an ACK from the PF. 313 314* **ixgbe: Fixed setting flow director flag twice.** 315 316 Resolved an issue where packets were being dropped when switching to perfect 317 filters mode. 318 319* **ixgbe: Set MDIO speed after MAC reset.** 320 321 The MDIO clock speed must be reconfigured after the MAC reset. The MDIO clock 322 speed becomes invalid, therefore the driver reads invalid PHY register values. 323 The driver now set the MDIO clock speed prior to initializing PHY ops and 324 again after the MAC reset. 325 326* **ixgbe: Fixed maximum number of available TX queues.** 327 328 In IXGBE, the maximum number of TX queues varies depending on the NIC operating 329 mode. This was not being updated in the device information, providing 330 an incorrect number in some cases. 331 332* **i40e: Generated MAC address for each VFs.** 333 334 It generates a MAC address for each VFs during PF host initialization, 335 and keeps the VF MAC address the same among different VF launch. 336 337* **i40e: Fixed failure of reading/writing RX control registers.** 338 339 Fixed i40e issue of failing to read/write rx control registers when 340 under stress with traffic, which might result in application launch 341 failure. 342 343* **i40e: Enabled vector driver by default.** 344 345 Previously, vector driver was disabled by default as it couldn't fill packet type 346 info for l3fwd to work well. Now there is an option for l3fwd to analyze 347 the packet type so the vector driver is enabled by default. 348 349* **i40e: Fixed link info of VF.** 350 351 Previously, the VF's link speed stayed at 10GbE and status always was up. 352 It did not change even when the physical link's status changed. 353 Now this issue is fixed to make VF's link info consistent with physical link. 354 355* **mlx5: Fixed possible crash during initialization.** 356 357 A crash could occur when failing to allocate private device context. 358 359* **mlx5: Added port type check.** 360 361 Added port type check to prevent port initialization on non-Ethernet link layers and 362 to report an error. 363 364* **mlx5: Applied VLAN filtering to broadcast and IPv6 multicast flows.** 365 366 Prevented reception of multicast frames outside of configured VLANs. 367 368* **mlx5: Fixed RX checksum offload in non L3/L4 packets.** 369 370 Fixed report of bad checksum for packets of unknown type. 371 372* **aesni_mb: Fixed wrong return value when creating a device.** 373 374 The ``cryptodev_aesni_mb_init()`` function was returning the device id of the device created, 375 instead of 0 (on success) that ``rte_eal_vdev_init()`` expects. 376 This made it impossible to create more than one aesni_mb device 377 from the command line. 378 379* **qat: Fixed AES GCM decryption.** 380 381 Allowed AES GCM on the cryptodev API, but in some cases gave invalid results 382 due to incorrect IV setting. 383 384 385Libraries 386~~~~~~~~~ 387 388* **hash: Fixed CRC32c hash computation for non multiple of 4 bytes sizes.** 389 390 Fix crc32c hash functions to return a valid crc32c value for data lengths 391 not a multiple of 4 bytes. 392 393* **hash: Fixed hash library to support multi-process mode.** 394 395 Fix hash library to support multi-process mode, using a jump table, 396 instead of storing a function pointer to the key compare function. 397 Multi-process mode only works with the built-in compare functions, 398 however a custom compare function (not in the jump table) can only 399 be used in single-process mode. 400 401* **hash: Fixed return value when allocating an existing hash table.** 402 403 Changed the ``rte_hash*_create()`` functions to return ``NULL`` and set 404 ``rte_errno`` to ``EEXIST`` when the object name already exists. This is 405 the behavior described in the API documentation in the header file. 406 The previous behavior was to return a pointer to the existing object in 407 that case, preventing the caller from knowing if the object had to be freed 408 or not. 409 410* **lpm: Fixed return value when allocating an existing object.** 411 412 Changed the ``rte_lpm*_create()`` functions to return ``NULL`` and set 413 ``rte_errno`` to ``EEXIST`` when the object name already exists. This is 414 the behavior described in the API documentation in the header file. 415 The previous behavior was to return a pointer to the existing object in 416 that case, preventing the caller from knowing if the object had to be freed 417 or not. 418 419* **librte_port: Fixed segmentation fault for ring and ethdev writer nodrop.** 420 421 Fixed core dump issue on txq and swq when dropless is set to yes. 422 423 424Examples 425~~~~~~~~ 426 427* **l3fwd-power: Fixed memory leak for non-IP packet.** 428 429 Fixed issue in l3fwd-power where, on receiving packets of types 430 other than IPv4 or IPv6, the mbuf was not released, and caused 431 a memory leak. 432 433* **l3fwd: Fixed using packet type blindly.** 434 435 l3fwd makes use of packet type information without querying if devices or PMDs 436 really set it. For those devices that don't set ptypes, add an option to parse it. 437 438* **examples/vhost: Fixed frequent mbuf allocation failure.** 439 440 The vhost-switch often fails to allocate mbuf when dequeue from vring because it 441 wrongly calculates the number of mbufs needed. 442 443 444API Changes 445----------- 446 447* The ethdev statistics counter ``imissed`` is considered to be independent of ``ierrors``. 448 All drivers are now counting the missed packets only once, i.e. drivers will 449 not increment ierrors anymore for missed packets. 450 451* The ethdev structure ``rte_eth_dev_info`` was changed to support device 452 speed capabilities. 453 454* The ethdev structures ``rte_eth_link`` and ``rte_eth_conf`` were changed to 455 support the new link API. 456 457* The functions ``rte_eth_dev_udp_tunnel_add`` and ``rte_eth_dev_udp_tunnel_delete`` 458 have been renamed into ``rte_eth_dev_udp_tunnel_port_add`` and 459 ``rte_eth_dev_udp_tunnel_port_delete``. 460 461* The ``outer_mac`` and ``inner_mac`` fields in structure 462 ``rte_eth_tunnel_filter_conf`` are changed from pointer to struct in order 463 to keep code's readability. 464 465* The fields in ethdev structure ``rte_eth_fdir_masks`` were changed 466 to be in big endian. 467 468* A parameter ``vlan_type`` has been added to the function 469 ``rte_eth_dev_set_vlan_ether_type``. 470 471* The af_packet device init function is no longer public. The device should be attached 472 via the API. 473 474* The LPM ``next_hop`` field is extended from 8 bits to 24 bits for IPv4 475 while keeping ABI compatibility. 476 477* A new ``rte_lpm_config`` structure is used so the LPM library will allocate 478 exactly the amount of memory which is necessary to hold application’s rules. 479 The previous ABI is kept for compatibility. 480 481* The prototype for the pipeline input port, output port and table action 482 handlers are updated: the pipeline parameter is added, 483 the packets mask parameter has been either removed or made input-only. 484 485 486ABI Changes 487----------- 488 489* The RETA entry size in ``rte_eth_rss_reta_entry64`` has been increased 490 from 8-bit to 16-bit. 491 492* The ethdev flow director structure ``rte_eth_fdir_flow`` structure was 493 changed. New fields were added to extend flow director's input set. 494 495* The cmdline buffer size has been increase from 256 to 512. 496 497 498Shared Library Versions 499----------------------- 500 501The libraries prepended with a plus sign were incremented in this version. 502 503.. code-block:: diff 504 505 + libethdev.so.3 506 librte_acl.so.2 507 librte_cfgfile.so.2 508 + librte_cmdline.so.2 509 librte_distributor.so.1 510 librte_eal.so.2 511 librte_hash.so.2 512 librte_ip_frag.so.1 513 librte_ivshmem.so.1 514 librte_jobstats.so.1 515 librte_kni.so.2 516 librte_kvargs.so.1 517 librte_lpm.so.2 518 librte_mbuf.so.2 519 librte_mempool.so.1 520 librte_meter.so.1 521 + librte_pipeline.so.3 522 librte_pmd_bond.so.1 523 librte_pmd_ring.so.2 524 librte_port.so.2 525 librte_power.so.1 526 librte_reorder.so.1 527 librte_ring.so.1 528 librte_sched.so.1 529 librte_table.so.2 530 librte_timer.so.1 531 librte_vhost.so.2 532 533 534Tested Platforms 535---------------- 536 537#. SuperMicro 1U 538 539 - BIOS: 1.0c 540 - Processor: Intel(R) Atom(TM) CPU C2758 @ 2.40GHz 541 542#. SuperMicro 1U 543 544 - BIOS: 1.0a 545 - Processor: Intel(R) Xeon(R) CPU D-1540 @ 2.00GHz 546 - Onboard NIC: Intel(R) X552/X557-AT (2x10G) 547 548 - Firmware-version: 0x800001cf 549 - Device ID (PF/VF): 8086:15ad /8086:15a8 550 551 - kernel driver version: 4.2.5 (ixgbe) 552 553#. SuperMicro 1U 554 555 - BIOS: 1.0a 556 - Processor: Intel(R) Xeon(R) CPU E5-4667 v3 @ 2.00GHz 557 558#. Intel(R) Server board S2600GZ 559 560 - BIOS: SE5C600.86B.02.02.0002.122320131210 561 - Processor: Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz 562 563#. Intel(R) Server board W2600CR 564 565 - BIOS: SE5C600.86B.02.01.0002.082220131453 566 - Processor: Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz 567 568#. Intel(R) Server board S2600CWT 569 570 - BIOS: SE5C610.86B.01.01.0009.060120151350 571 - Processor: Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz 572 573#. Intel(R) Server board S2600WTT 574 575 - BIOS: SE5C610.86B.01.01.0005.101720141054 576 - Processor: Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz 577 578#. Intel(R) Server board S2600WTT 579 580 - BIOS: SE5C610.86B.11.01.0044.090120151156 581 - Processor: Intel(R) Xeon(R) CPU E5-2695 v4 @ 2.10GHz 582 583 584Tested NICs 585----------- 586 587#. Intel(R) Ethernet Controller X540-AT2 588 589 - Firmware version: 0x80000389 590 - Device id (pf): 8086:1528 591 - Driver version: 3.23.2 (ixgbe) 592 593#. Intel(R) 82599ES 10 Gigabit Ethernet Controller 594 595 - Firmware version: 0x61bf0001 596 - Device id (pf/vf): 8086:10fb / 8086:10ed 597 - Driver version: 4.0.1-k (ixgbe) 598 599#. Intel(R) Corporation Ethernet Connection X552/X557-AT 10GBASE-T 600 601 - Firmware version: 0x800001cf 602 - Device id (pf/vf): 8086:15ad / 8086:15a8 603 - Driver version: 4.2.5 (ixgbe) 604 605#. Intel(R) Ethernet Converged Network Adapter X710-DA4 (4x10G) 606 607 - Firmware version: 5.02 0x80002284 608 - Device id (pf/vf): 8086:1572 / 8086:154c 609 - Driver version: 1.4.26 (i40e) 610 611#. Intel(R) Ethernet Converged Network Adapter X710-DA2 (2x10G) 612 613 - Firmware version: 5.02 0x80002282 614 - Device id (pf/vf): 8086:1572 / 8086:154c 615 - Driver version: 1.4.25 (i40e) 616 617#. Intel(R) Ethernet Converged Network Adapter XL710-QDA1 (1x40G) 618 619 - Firmware version: 5.02 0x80002281 620 - Device id (pf/vf): 8086:1584 / 8086:154c 621 - Driver version: 1.4.25 (i40e) 622 623#. Intel(R) Ethernet Converged Network Adapter XL710-QDA2 (2X40G) 624 625 - Firmware version: 5.02 0x80002285 626 - Device id (pf/vf): 8086:1583 / 8086:154c 627 - Driver version: 1.4.25 (i40e) 628 629#. Intel(R) 82576EB Gigabit Ethernet Controller 630 631 - Firmware version: 1.2.1 632 - Device id (pf): 8086:1526 633 - Driver version: 5.2.13-k (igb) 634 635#. Intel(R) Ethernet Controller I210 636 637 - Firmware version: 3.16, 0x80000500, 1.304.0 638 - Device id (pf): 8086:1533 639 - Driver version: 5.2.13-k (igb) 640 641#. Intel(R) Corporation I350 Gigabit Network Connection 642 643 - Firmware version: 1.48, 0x800006e7 644 - Device id (pf/vf): 8086:1521 / 8086:1520 645 - Driver version: 5.2.13-k (igb) 646 647 648#. Intel(R) Ethernet Multi-host Controller FM10000 649 650 - Firmware version: N/A 651 - Device id (pf/vf): 8086:15d0 652 - Driver version: 0.17.0.9 (fm10k) 653