1.. SPDX-License-Identifier: BSD-3-Clause 2 Copyright 2018 The DPDK contributors 3 4ABI and API Deprecation 5======================= 6 7See the guidelines document for details of the :doc:`ABI policy 8<../contributing/abi_policy>`. API and ABI deprecation notices are to be posted 9here. 10 11Deprecation Notices 12------------------- 13 14* build: The macros defined to indicate which DPDK libraries and drivers 15 are included in the meson build are changing to a standardized format of 16 ``RTE_LIB_<NAME>`` and ``RTE_<CLASS>_<NAME>``, where ``NAME`` is the 17 upper-case component name, e.g. EAL, ETHDEV, IXGBE, and ``CLASS`` is the 18 upper-case name of the device class to which a driver belongs e.g. 19 ``NET``, ``CRYPTO``, ``VDPA``. The old macros are deprecated and will be 20 removed in a future release. 21 22* kvargs: The function ``rte_kvargs_process`` will get a new parameter 23 for returning key match count. It will ease handling of no-match case. 24 25* eal: The function ``rte_eal_remote_launch`` will return new error codes 26 after read or write error on the pipe, instead of calling ``rte_panic``. 27 28* rte_atomicNN_xxx: These APIs do not take memory order parameter. This does 29 not allow for writing optimized code for all the CPU architectures supported 30 in DPDK. DPDK will adopt C11 atomic operations semantics and provide wrappers 31 using C11 atomic built-ins. These wrappers must be used for patches that 32 need to be merged in 20.08 onwards. This change will not introduce any 33 performance degradation. 34 35* rte_smp_*mb: These APIs provide full barrier functionality. However, many 36 use cases do not require full barriers. To support such use cases, DPDK will 37 adopt C11 barrier semantics and provide wrappers using C11 atomic built-ins. 38 These wrappers must be used for patches that need to be merged in 20.08 39 onwards. This change will not introduce any performance degradation. 40 41* lib: will fix extending some enum/define breaking the ABI. There are multiple 42 samples in DPDK that enum/define terminated with a ``.*MAX.*`` value which is 43 used by iterators, and arrays holding these values are sized with this 44 ``.*MAX.*`` value. So extending this enum/define increases the ``.*MAX.*`` 45 value which increases the size of the array and depending on how/where the 46 array is used this may break the ABI. 47 ``RTE_ETH_FLOW_MAX`` is one sample of the mentioned case, adding a new flow 48 type will break the ABI because of ``flex_mask[RTE_ETH_FLOW_MAX]`` array 49 usage in following public struct hierarchy: 50 ``rte_eth_fdir_flex_conf -> rte_fdir_conf -> rte_eth_conf (in the middle)``. 51 Need to identify this kind of usages and fix in 20.11, otherwise this blocks 52 us extending existing enum/define. 53 One solution can be using a fixed size array instead of ``.*MAX.*`` value. 54 55* ethdev: The flow director API, including ``rte_eth_conf.fdir_conf`` field, 56 and the related structures (``rte_fdir_*`` and ``rte_eth_fdir_*``), 57 will be removed in DPDK 20.11. 58 59* ethdev: New offload flags ``DEV_RX_OFFLOAD_FLOW_MARK`` will be added in 19.11. 60 This will allow application to enable or disable PMDs from updating 61 ``rte_mbuf::hash::fdir``. 62 This scheme will allow PMDs to avoid writes to ``rte_mbuf`` fields on Rx and 63 thereby improve Rx performance if application wishes do so. 64 In 19.11 PMDs will still update the field even when the offload is not 65 enabled. 66 67* ethdev: ``uint32_t max_rx_pkt_len`` field of ``struct rte_eth_rxmode``, will be 68 replaced by a new ``uint32_t mtu`` field of ``struct rte_eth_conf`` in v21.11. 69 The new ``mtu`` field will be used to configure the initial device MTU via 70 ``rte_eth_dev_configure()`` API. 71 Later MTU can be changed by ``rte_eth_dev_set_mtu()`` API as done now. 72 The existing ``(struct rte_eth_dev)->data->mtu`` variable will be used to store 73 the configured ``mtu`` value, 74 and this new ``(struct rte_eth_dev)->data->dev_conf.mtu`` variable will 75 be used to store the user configuration request. 76 Unlike ``max_rx_pkt_len``, which was valid only when ``JUMBO_FRAME`` enabled, 77 ``mtu`` field will be always valid. 78 When ``mtu`` config is not provided by the application, default ``RTE_ETHER_MTU`` 79 value will be used. 80 ``(struct rte_eth_dev)->data->mtu`` should be updated after MTU set successfully, 81 either by ``rte_eth_dev_configure()`` or ``rte_eth_dev_set_mtu()``. 82 83 An application may need to configure device for a specific Rx packet size, like for 84 cases ``DEV_RX_OFFLOAD_SCATTER`` is not supported and device received packet size 85 can't be bigger than Rx buffer size. 86 To cover these cases an application needs to know the device packet overhead to be 87 able to calculate the ``mtu`` corresponding to a Rx buffer size, for this 88 ``(struct rte_eth_dev_info).max_rx_pktlen`` will be kept, 89 the device packet overhead can be calculated as: 90 ``(struct rte_eth_dev_info).max_rx_pktlen - (struct rte_eth_dev_info).max_mtu`` 91 92* ethdev: ``rx_descriptor_done`` dev_ops and ``rte_eth_rx_descriptor_done`` 93 will be removed in 21.11. 94 Existing ``rte_eth_rx_descriptor_status`` and ``rte_eth_tx_descriptor_status`` 95 APIs can be used as replacement. 96 97* ethdev: The port mirroring API can be replaced with a more fine grain flow API. 98 The structs ``rte_eth_mirror_conf``, ``rte_eth_vlan_mirror`` and the functions 99 ``rte_eth_mirror_rule_set``, ``rte_eth_mirror_rule_reset`` will be marked 100 as deprecated in DPDK 20.11, along with the associated macros ``ETH_MIRROR_*``. 101 This API will be fully removed in DPDK 21.11. 102 103* ethdev: Attribute ``shared`` of the ``struct rte_flow_action_count`` 104 is deprecated and will be removed in DPDK 21.11. Shared counters should 105 be managed using shared actions API (``rte_flow_shared_action_create`` etc). 106 107* ethdev: The flow API matching pattern structures, ``struct rte_flow_item_*``, 108 should start with relevant protocol header. 109 Some matching pattern structures implements this by duplicating protocol header 110 fields in the struct. To clarify the intention and to be sure protocol header 111 is intact, will replace those fields with relevant protocol header struct. 112 In v21.02 both individual protocol header fields and the protocol header struct 113 will be added as union, target is switch usage to the protocol header by time. 114 In v21.11 LTS, protocol header fields will be cleaned and only protocol header 115 struct will remain. 116 117* ethdev: Queue specific stats fields will be removed from ``struct rte_eth_stats``. 118 Mentioned fields are: ``q_ipackets``, ``q_opackets``, ``q_ibytes``, ``q_obytes``, 119 ``q_errors``. 120 Instead queue stats will be received via xstats API. Current method support 121 will be limited to maximum 256 queues. 122 Also compile time flag ``RTE_ETHDEV_QUEUE_STAT_CNTRS`` will be removed. 123 124* Broadcom bnxt PMD: NetXtreme devices belonging to the ``BCM573xx and 125 BCM5740x`` families will no longer be supported as of DPDK 21.02. 126 Specifically the support for the following Broadcom PCI IDs will be removed 127 from the release: ``0x16c8, 0x16c9, 0x16ca, 0x16ce, 0x16cf, 0x16df,`` 128 ``0x16d0, 0x16d1, 0x16d2, 0x16d4, 0x16d5, 0x16e7, 0x16e8, 0x16e9``. 129 130* sched: To allow more traffic classes, flexible mapping of pipe queues to 131 traffic classes, and subport level configuration of pipes and queues 132 changes will be made to macros, data structures and API functions defined 133 in "rte_sched.h". These changes are aligned to improvements suggested in the 134 RFC https://mails.dpdk.org/archives/dev/2018-November/120035.html. 135 136* metrics: The function ``rte_metrics_init`` will have a non-void return 137 in order to notify errors instead of calling ``rte_exit``. 138 139* cmdline: ``cmdline`` structure will be made opaque to hide platform-specific 140 content. On Linux and FreeBSD, supported prior to DPDK 20.11, 141 original structure will be kept until DPDK 21.11. 142