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 * mbuf: The mbuf offload flags ``PKT_*`` will be renamed as ``RTE_MBUF_F_*``.
43 A compatibility layer will be kept until DPDK 22.11, except for the flags
44 that are already deprecated (``PKT_RX_L4_CKSUM_BAD``, ``PKT_RX_IP_CKSUM_BAD``,
45 ``PKT_RX_EIP_CKSUM_BAD``, ``PKT_TX_QINQ_PKT``) which will be removed.
47 * pci: To reduce unnecessary ABIs exposed by DPDK bus driver, "rte_bus_pci.h"
48 will be made internal in 21.11 and macros/data structures/functions defined
49 in the header will not be considered as ABI anymore. This change is inspired
50 by the RFC https://patchwork.dpdk.org/project/dpdk/list/?series=17176.
52 * lib: will fix extending some enum/define breaking the ABI. There are multiple
53 samples in DPDK that enum/define terminated with a ``.*MAX.*`` value which is
54 used by iterators, and arrays holding these values are sized with this
55 ``.*MAX.*`` value. So extending this enum/define increases the ``.*MAX.*``
56 value which increases the size of the array and depending on how/where the
57 array is used this may break the ABI.
58 ``RTE_ETH_FLOW_MAX`` is one sample of the mentioned case, adding a new flow
59 type will break the ABI because of ``flex_mask[RTE_ETH_FLOW_MAX]`` array
60 usage in following public struct hierarchy:
61 ``rte_eth_fdir_flex_conf -> rte_fdir_conf -> rte_eth_conf (in the middle)``.
62 Need to identify this kind of usages and fix in 20.11, otherwise this blocks
63 us extending existing enum/define.
64 One solution can be using a fixed size array instead of ``.*MAX.*`` value.
66 * ethdev: Will add ``RTE_ETH_`` prefix to all ethdev macros/enums in v21.11.
67 Macros will be added for backward compatibility.
68 Backward compatibility macros will be removed on v22.11.
69 A few old backward compatibility macros from 2013 that does not have
70 proper prefix will be removed on v21.11.
72 * ethdev: The flow director API, including ``rte_eth_conf.fdir_conf`` field,
73 and the related structures (``rte_fdir_*`` and ``rte_eth_fdir_*``),
74 will be removed in DPDK 20.11.
76 * ethdev: New offload flags ``DEV_RX_OFFLOAD_FLOW_MARK`` will be added in 19.11.
77 This will allow application to enable or disable PMDs from updating
78 ``rte_mbuf::hash::fdir``.
79 This scheme will allow PMDs to avoid writes to ``rte_mbuf`` fields on Rx and
80 thereby improve Rx performance if application wishes do so.
81 In 19.11 PMDs will still update the field even when the offload is not
84 * ethdev: ``uint32_t max_rx_pkt_len`` field of ``struct rte_eth_rxmode``, will be
85 replaced by a new ``uint32_t mtu`` field of ``struct rte_eth_conf`` in v21.11.
86 The new ``mtu`` field will be used to configure the initial device MTU via
87 ``rte_eth_dev_configure()`` API.
88 Later MTU can be changed by ``rte_eth_dev_set_mtu()`` API as done now.
89 The existing ``(struct rte_eth_dev)->data->mtu`` variable will be used to store
90 the configured ``mtu`` value,
91 and this new ``(struct rte_eth_dev)->data->dev_conf.mtu`` variable will
92 be used to store the user configuration request.
93 Unlike ``max_rx_pkt_len``, which was valid only when ``JUMBO_FRAME`` enabled,
94 ``mtu`` field will be always valid.
95 When ``mtu`` config is not provided by the application, default ``RTE_ETHER_MTU``
97 ``(struct rte_eth_dev)->data->mtu`` should be updated after MTU set successfully,
98 either by ``rte_eth_dev_configure()`` or ``rte_eth_dev_set_mtu()``.
100 An application may need to configure device for a specific Rx packet size, like for
101 cases ``DEV_RX_OFFLOAD_SCATTER`` is not supported and device received packet size
102 can't be bigger than Rx buffer size.
103 To cover these cases an application needs to know the device packet overhead to be
104 able to calculate the ``mtu`` corresponding to a Rx buffer size, for this
105 ``(struct rte_eth_dev_info).max_rx_pktlen`` will be kept,
106 the device packet overhead can be calculated as:
107 ``(struct rte_eth_dev_info).max_rx_pktlen - (struct rte_eth_dev_info).max_mtu``
109 * ethdev: ``rx_descriptor_done`` dev_ops and ``rte_eth_rx_descriptor_done``
110 will be removed in 21.11.
111 Existing ``rte_eth_rx_descriptor_status`` and ``rte_eth_tx_descriptor_status``
112 APIs can be used as replacement.
114 * ethdev: Announce moving from dedicated modify function for each field,
115 to using the general ``rte_flow_modify_field`` action.
117 * ethdev: The struct ``rte_flow_action_modify_data`` will be modified
118 to support modifying fields larger than 64 bits.
119 In addition, documentation will be updated to clarify byte order.
121 * ethdev: Attribute ``shared`` of the ``struct rte_flow_action_count``
122 is deprecated and will be removed in DPDK 21.11. Shared counters should
123 be managed using shared actions API (``rte_flow_shared_action_create`` etc).
125 * ethdev: Definition of the flow API action ``RTE_FLOW_ACTION_TYPE_PORT_ID``
126 is ambiguous and needs clarification.
127 Structure ``rte_flow_action_port_id`` will be extended to specify
128 traffic direction to the represented entity or ethdev port itself
131 * ethdev: Flow API documentation is unclear if ethdev port used to create
132 a flow rule adds any implicit match criteria in the case of transfer rules.
133 The semantics will be clarified in DPDK 21.11 and it will require fixes in
134 drivers and applications which interpret it in a different way.
136 * ethdev: The flow API matching pattern structures, ``struct rte_flow_item_*``,
137 should start with relevant protocol header.
138 Some matching pattern structures implements this by duplicating protocol header
139 fields in the struct. To clarify the intention and to be sure protocol header
140 is intact, will replace those fields with relevant protocol header struct.
141 In v21.02 both individual protocol header fields and the protocol header struct
142 will be added as union, target is switch usage to the protocol header by time.
143 In v21.11 LTS, protocol header fields will be cleaned and only protocol header
146 * ethdev: Add new Rx queue offload flag ``RTE_ETH_RX_OFFLOAD_SHARED_RXQ`` and
147 ``shared_group`` field to ``struct rte_eth_rxconf``.
149 * ethdev: Queue specific stats fields will be removed from ``struct rte_eth_stats``.
150 Mentioned fields are: ``q_ipackets``, ``q_opackets``, ``q_ibytes``, ``q_obytes``,
152 Instead queue stats will be received via xstats API. Current method support
153 will be limited to maximum 256 queues.
154 Also compile time flag ``RTE_ETHDEV_QUEUE_STAT_CNTRS`` will be removed.
156 * ethdev: The offload flag ``PKT_RX_EIP_CKSUM_BAD`` will be removed and
157 replaced by the new flag ``PKT_RX_OUTER_IP_CKSUM_BAD``. The new name is more
158 consistent with existing outer header checksum status flag naming, which
159 should help in reducing confusion about its usage.
161 * net: The structure ``rte_ipv4_hdr`` will have two unions.
162 The first union is for existing ``version_ihl`` byte
163 and new bitfield for version and IHL.
164 The second union is for existing ``fragment_offset``
165 and new bitfield for fragment flags and offset.
167 * vhost: ``rte_vdpa_register_device``, ``rte_vdpa_unregister_device``,
168 ``rte_vhost_host_notifier_ctrl`` and ``rte_vdpa_relay_vring_used`` vDPA
169 driver interface will be marked as internal in DPDK v21.11.
171 * vhost: rename ``struct vhost_device_ops`` to ``struct rte_vhost_device_ops``
174 * vhost: The experimental tags of ``rte_vhost_driver_get_protocol_features``,
175 ``rte_vhost_driver_get_queue_num``, ``rte_vhost_crypto_create``,
176 ``rte_vhost_crypto_free``, ``rte_vhost_crypto_fetch_requests``,
177 ``rte_vhost_crypto_finalize_requests``, ``rte_vhost_crypto_set_zero_copy``,
178 ``rte_vhost_va_from_guest_pa``, ``rte_vhost_extern_callback_register``,
179 and ``rte_vhost_driver_set_protocol_features`` functions will be removed
180 and the API functions will be made stable in DPDK 21.11.
182 * cryptodev: The field ``dataunit_len`` of the ``struct rte_crypto_cipher_xform``
183 has a limited size ``uint16_t``.
184 It will be moved and extended as ``uint32_t`` in DPDK 21.11.
186 * cryptodev: The structure ``rte_crypto_sym_vec`` would be updated to add
187 ``dest_sgl`` to support out of place processing.
188 This field will be null for inplace processing.
189 This change is targeted for DPDK 21.11.
191 * cryptodev: The structure ``rte_crypto_vec`` would be updated to add
192 ``tot_len`` to support total buffer length.
193 This is required for security cases like IPsec and PDCP encryption offload
194 to know how much additional memory space is available in buffer other than
195 data length so that driver/HW can write expanded size data after encryption.
196 This change is targeted for DPDK 21.11.
198 * cryptodev: Hide structures ``rte_cryptodev_sym_session`` and
199 ``rte_cryptodev_asym_session`` to remove unnecessary indirection between
200 session and the private data of session. An opaque pointer can be exposed
201 directly to application which can be attached to the ``rte_crypto_op``.
203 * security: Hide structure ``rte_security_session`` and expose an opaque
204 pointer for the private data to the application which can be attached
205 to the packet while enqueuing.
207 * security: The structure ``rte_security_ipsec_xform`` will be extended with
208 multiple fields: source and destination port of UDP encapsulation,
209 IPsec payload MSS (Maximum Segment Size), and ESN (Extended Sequence Number).
211 * security: The IPsec SA config options ``struct rte_security_ipsec_sa_options``
212 will be updated with new fields to support new features like TSO in case of
215 * ipsec: The structure ``rte_ipsec_sa_prm`` will be extended with a new field
216 ``hdr_l3_len`` to configure tunnel L3 header length.
218 * eventdev: The file ``rte_eventdev_pmd.h`` will be renamed to ``eventdev_driver.h``
219 to make the driver interface as internal and the structures ``rte_eventdev_data``,
220 ``rte_eventdev`` and ``rte_eventdevs`` will be moved to a new file named
221 ``rte_eventdev_core.h`` in DPDK 21.11.
222 The ``rte_`` prefix for internal structures and functions will be removed across the
224 The experimental eventdev trace APIs and ``rte_event_vector_pool_create``,
225 ``rte_event_eth_rx_adapter_vector_limits_get`` will be promoted to stable.
226 An 8-byte reserved field will be added to the structure ``rte_event_timer`` to
227 support future extensions.
229 * eventdev: The structure ``rte_event_eth_rx_adapter_queue_conf`` will be
230 extended to include ``rte_event_eth_rx_adapter_event_vector_config`` elements
231 and the function ``rte_event_eth_rx_adapter_queue_event_vector_config`` will
232 be removed in DPDK 21.11.
234 An application can enable event vectorization by passing the desired vector
235 values to the function ``rte_event_eth_rx_adapter_queue_add`` using
236 the structure ``rte_event_eth_rx_adapter_queue_add``.
238 * metrics: The function ``rte_metrics_init`` will have a non-void return
239 in order to notify errors instead of calling ``rte_exit``.
241 * cmdline: ``cmdline`` structure will be made opaque to hide platform-specific
242 content. On Linux and FreeBSD, supported prior to DPDK 20.11,
243 original structure will be kept until DPDK 21.11.