1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright 2018 The DPDK contributors
4 ABI and API Deprecation
5 =======================
7 See the guidelines document for details of the :doc:`ABI policy
8 <../contributing/abi_policy>`. API and ABI deprecation notices are to be posted
14 * kvargs: The function ``rte_kvargs_process`` will get a new parameter
15 for returning key match count. It will ease handling of no-match case.
17 * eal: The function ``rte_eal_remote_launch`` will return new error codes
18 after read or write error on the pipe, instead of calling ``rte_panic``.
20 * eal: The lcore state ``FINISHED`` will be removed from
21 the ``enum rte_lcore_state_t``.
22 The lcore state ``WAIT`` is enough to represent the same state.
24 * eal: Making ``struct rte_intr_handle`` internal to avoid any ABI breakages
27 * rte_atomicNN_xxx: These APIs do not take memory order parameter. This does
28 not allow for writing optimized code for all the CPU architectures supported
29 in DPDK. DPDK has adopted the atomic operations from
30 https://gcc.gnu.org/onlinedocs/gcc/_005f_005fatomic-Builtins.html. These
31 operations must be used for patches that need to be merged in 20.08 onwards.
32 This change will not introduce any performance degradation.
34 * rte_smp_*mb: These APIs provide full barrier functionality. However, many
35 use cases do not require full barriers. To support such use cases, DPDK has
36 adopted atomic operations from
37 https://gcc.gnu.org/onlinedocs/gcc/_005f_005fatomic-Builtins.html. These
38 operations and a new wrapper ``rte_atomic_thread_fence`` instead of
39 ``__atomic_thread_fence`` must be used for patches that need to be merged in
40 20.08 onwards. This change will not introduce any performance degradation.
42 * mempool: Helper macro ``MEMPOOL_HEADER_SIZE()`` is deprecated and will
43 be removed in DPDK 22.11. The replacement macro
44 ``RTE_MEMPOOL_HEADER_SIZE()`` is internal only.
46 * mempool: Macro to register mempool driver ``MEMPOOL_REGISTER_OPS()`` is
47 deprecated and will be removed in DPDK 22.11. Use replacement macro
48 ``RTE_MEMPOOL_REGISTER_OPS()``.
50 * mempool: The mempool API macros ``MEMPOOL_PG_*`` are deprecated and
51 will be removed in DPDK 22.11.
53 * mbuf: The mbuf offload flags ``PKT_*`` will be renamed as ``RTE_MBUF_F_*``.
54 A compatibility layer will be kept until DPDK 22.11, except for the flags
55 that are already deprecated (``PKT_RX_L4_CKSUM_BAD``, ``PKT_RX_IP_CKSUM_BAD``,
56 ``PKT_RX_EIP_CKSUM_BAD``, ``PKT_TX_QINQ_PKT``) which will be removed.
58 * pci: To reduce unnecessary ABIs exposed by DPDK bus driver, "rte_bus_pci.h"
59 will be made internal in 21.11 and macros/data structures/functions defined
60 in the header will not be considered as ABI anymore. This change is inspired
61 by the RFC https://patchwork.dpdk.org/project/dpdk/list/?series=17176.
63 * lib: will fix extending some enum/define breaking the ABI. There are multiple
64 samples in DPDK that enum/define terminated with a ``.*MAX.*`` value which is
65 used by iterators, and arrays holding these values are sized with this
66 ``.*MAX.*`` value. So extending this enum/define increases the ``.*MAX.*``
67 value which increases the size of the array and depending on how/where the
68 array is used this may break the ABI.
69 ``RTE_ETH_FLOW_MAX`` is one sample of the mentioned case, adding a new flow
70 type will break the ABI because of ``flex_mask[RTE_ETH_FLOW_MAX]`` array
71 usage in following public struct hierarchy:
72 ``rte_eth_fdir_flex_conf -> rte_fdir_conf -> rte_eth_conf (in the middle)``.
73 Need to identify this kind of usages and fix in 20.11, otherwise this blocks
74 us extending existing enum/define.
75 One solution can be using a fixed size array instead of ``.*MAX.*`` value.
77 * ethdev: Will add ``RTE_ETH_`` prefix to all ethdev macros/enums in v21.11.
78 Macros will be added for backward compatibility.
79 Backward compatibility macros will be removed on v22.11.
80 A few old backward compatibility macros from 2013 that does not have
81 proper prefix will be removed on v21.11.
83 * ethdev: The flow director API, including ``rte_eth_conf.fdir_conf`` field,
84 and the related structures (``rte_fdir_*`` and ``rte_eth_fdir_*``),
85 will be removed in DPDK 20.11.
87 * ethdev: New offload flags ``DEV_RX_OFFLOAD_FLOW_MARK`` will be added in 19.11.
88 This will allow application to enable or disable PMDs from updating
89 ``rte_mbuf::hash::fdir``.
90 This scheme will allow PMDs to avoid writes to ``rte_mbuf`` fields on Rx and
91 thereby improve Rx performance if application wishes do so.
92 In 19.11 PMDs will still update the field even when the offload is not
95 * ethdev: ``uint32_t max_rx_pkt_len`` field of ``struct rte_eth_rxmode``, will be
96 replaced by a new ``uint32_t mtu`` field of ``struct rte_eth_conf`` in v21.11.
97 The new ``mtu`` field will be used to configure the initial device MTU via
98 ``rte_eth_dev_configure()`` API.
99 Later MTU can be changed by ``rte_eth_dev_set_mtu()`` API as done now.
100 The existing ``(struct rte_eth_dev)->data->mtu`` variable will be used to store
101 the configured ``mtu`` value,
102 and this new ``(struct rte_eth_dev)->data->dev_conf.mtu`` variable will
103 be used to store the user configuration request.
104 Unlike ``max_rx_pkt_len``, which was valid only when ``JUMBO_FRAME`` enabled,
105 ``mtu`` field will be always valid.
106 When ``mtu`` config is not provided by the application, default ``RTE_ETHER_MTU``
108 ``(struct rte_eth_dev)->data->mtu`` should be updated after MTU set successfully,
109 either by ``rte_eth_dev_configure()`` or ``rte_eth_dev_set_mtu()``.
111 An application may need to configure device for a specific Rx packet size, like for
112 cases ``DEV_RX_OFFLOAD_SCATTER`` is not supported and device received packet size
113 can't be bigger than Rx buffer size.
114 To cover these cases an application needs to know the device packet overhead to be
115 able to calculate the ``mtu`` corresponding to a Rx buffer size, for this
116 ``(struct rte_eth_dev_info).max_rx_pktlen`` will be kept,
117 the device packet overhead can be calculated as:
118 ``(struct rte_eth_dev_info).max_rx_pktlen - (struct rte_eth_dev_info).max_mtu``
120 * ethdev: Announce moving from dedicated modify function for each field,
121 to using the general ``rte_flow_modify_field`` action.
123 * ethdev: The flow API matching pattern structures, ``struct rte_flow_item_*``,
124 should start with relevant protocol header.
125 Some matching pattern structures implements this by duplicating protocol header
126 fields in the struct. To clarify the intention and to be sure protocol header
127 is intact, will replace those fields with relevant protocol header struct.
128 In v21.02 both individual protocol header fields and the protocol header struct
129 will be added as union, target is switch usage to the protocol header by time.
130 In v21.11 LTS, protocol header fields will be cleaned and only protocol header
133 * ethdev: Add new Rx queue offload flag ``RTE_ETH_RX_OFFLOAD_SHARED_RXQ`` and
134 ``shared_group`` field to ``struct rte_eth_rxconf``.
136 * ethdev: Queue specific stats fields will be removed from ``struct rte_eth_stats``.
137 Mentioned fields are: ``q_ipackets``, ``q_opackets``, ``q_ibytes``, ``q_obytes``,
139 Instead queue stats will be received via xstats API. Current method support
140 will be limited to maximum 256 queues.
141 Also compile time flag ``RTE_ETHDEV_QUEUE_STAT_CNTRS`` will be removed.
143 * ethdev: Items and actions ``PF``, ``VF``, ``PHY_PORT``, ``PORT_ID`` are
144 deprecated as hard-to-use / ambiguous and will be removed in DPDK 22.11.
146 * ethdev: The use of attributes ``ingress`` / ``egress`` in "transfer" flows
147 is deprecated as ambiguous with respect to the embedded switch. The use of
148 these attributes will become invalid starting from DPDK 22.11.
150 * vhost: ``rte_vdpa_register_device``, ``rte_vdpa_unregister_device``,
151 ``rte_vhost_host_notifier_ctrl`` and ``rte_vdpa_relay_vring_used`` vDPA
152 driver interface will be marked as internal in DPDK v21.11.
154 * vhost: rename ``struct vhost_device_ops`` to ``struct rte_vhost_device_ops``
157 * vhost: The experimental tags of ``rte_vhost_driver_get_protocol_features``,
158 ``rte_vhost_driver_get_queue_num``, ``rte_vhost_crypto_create``,
159 ``rte_vhost_crypto_free``, ``rte_vhost_crypto_fetch_requests``,
160 ``rte_vhost_crypto_finalize_requests``, ``rte_vhost_crypto_set_zero_copy``,
161 ``rte_vhost_va_from_guest_pa``, ``rte_vhost_extern_callback_register``,
162 and ``rte_vhost_driver_set_protocol_features`` functions will be removed
163 and the API functions will be made stable in DPDK 21.11.
165 * cryptodev: Hide structures ``rte_cryptodev_sym_session`` and
166 ``rte_cryptodev_asym_session`` to remove unnecessary indirection between
167 session and the private data of session. An opaque pointer can be exposed
168 directly to application which can be attached to the ``rte_crypto_op``.
170 * security: Hide structure ``rte_security_session`` and expose an opaque
171 pointer for the private data to the application which can be attached
172 to the packet while enqueuing.
174 * eventdev: The file ``rte_eventdev_pmd.h`` will be renamed to ``eventdev_driver.h``
175 to make the driver interface as internal and the structures ``rte_eventdev_data``,
176 ``rte_eventdev`` and ``rte_eventdevs`` will be moved to a new file named
177 ``rte_eventdev_core.h`` in DPDK 21.11.
178 The ``rte_`` prefix for internal structures and functions will be removed across the
180 The experimental eventdev trace APIs and ``rte_event_vector_pool_create``,
181 ``rte_event_eth_rx_adapter_vector_limits_get`` will be promoted to stable.
182 An 8-byte reserved field will be added to the structure ``rte_event_timer`` to
183 support future extensions.
185 * eventdev: The structure ``rte_event_eth_rx_adapter_queue_conf`` will be
186 extended to include ``rte_event_eth_rx_adapter_event_vector_config`` elements
187 and the function ``rte_event_eth_rx_adapter_queue_event_vector_config`` will
188 be removed in DPDK 21.11.
190 An application can enable event vectorization by passing the desired vector
191 values to the function ``rte_event_eth_rx_adapter_queue_add`` using
192 the structure ``rte_event_eth_rx_adapter_queue_add``.
194 * metrics: The function ``rte_metrics_init`` will have a non-void return
195 in order to notify errors instead of calling ``rte_exit``.
197 * cmdline: ``cmdline`` structure will be made opaque to hide platform-specific
198 content. On Linux and FreeBSD, supported prior to DPDK 20.11,
199 original structure will be kept until DPDK 21.11.