29. ABI and API Deprecation
See the guidelines document for details of the ABI policy. API and ABI deprecation notices are to be posted here.
29.1. Deprecation Notices
kvargs: The function
rte_kvargs_processwill get a new parameter for returning key match count. It will ease handling of no-match case.
eal: The function
rte_eal_remote_launchwill return new error codes after read or write error on the pipe, instead of calling
eal: The lcore state
FINISHEDwill be removed from the
enum rte_lcore_state_t. The lcore state
WAITis enough to represent the same state.
struct rte_intr_handleinternal to avoid any ABI breakages in future.
rte_atomicNN_xxx: These APIs do not take memory order parameter. This does not allow for writing optimized code for all the CPU architectures supported in DPDK. DPDK has adopted the atomic operations from https://gcc.gnu.org/onlinedocs/gcc/_005f_005fatomic-Builtins.html. These operations must be used for patches that need to be merged in 20.08 onwards. This change will not introduce any performance degradation.
rte_smp_*mb: These APIs provide full barrier functionality. However, many use cases do not require full barriers. To support such use cases, DPDK has adopted atomic operations from https://gcc.gnu.org/onlinedocs/gcc/_005f_005fatomic-Builtins.html. These operations and a new wrapper
__atomic_thread_fencemust be used for patches that need to be merged in 20.08 onwards. This change will not introduce any performance degradation.
mbuf: The mbuf offload flags
PKT_*will be renamed as
RTE_MBUF_F_*. A compatibility layer will be kept until DPDK 22.11, except for the flags that are already deprecated (
PKT_TX_QINQ_PKT) which will be removed.
pci: To reduce unnecessary ABIs exposed by DPDK bus driver, “rte_bus_pci.h” will be made internal in 21.11 and macros/data structures/functions defined in the header will not be considered as ABI anymore. This change is inspired by the RFC https://patchwork.dpdk.org/project/dpdk/list/?series=17176.
lib: will fix extending some enum/define breaking the ABI. There are multiple samples in DPDK that enum/define terminated with a
.*MAX.*value which is used by iterators, and arrays holding these values are sized with this
.*MAX.*value. So extending this enum/define increases the
.*MAX.*value which increases the size of the array and depending on how/where the array is used this may break the ABI.
RTE_ETH_FLOW_MAXis one sample of the mentioned case, adding a new flow type will break the ABI because of
flex_mask[RTE_ETH_FLOW_MAX]array usage in following public struct hierarchy:
rte_eth_fdir_flex_conf -> rte_fdir_conf -> rte_eth_conf (in the middle). Need to identify this kind of usages and fix in 20.11, otherwise this blocks us extending existing enum/define. One solution can be using a fixed size array instead of
ethdev: Will add
RTE_ETH_prefix to all ethdev macros/enums in v21.11. Macros will be added for backward compatibility. Backward compatibility macros will be removed on v22.11. A few old backward compatibility macros from 2013 that does not have proper prefix will be removed on v21.11.
ethdev: The flow director API, including
rte_eth_conf.fdir_conffield, and the related structures (
rte_eth_fdir_*), will be removed in DPDK 20.11.
ethdev: New offload flags
DEV_RX_OFFLOAD_FLOW_MARKwill be added in 19.11. This will allow application to enable or disable PMDs from updating
rte_mbuf::hash::fdir. This scheme will allow PMDs to avoid writes to
rte_mbuffields on Rx and thereby improve Rx performance if application wishes do so. In 19.11 PMDs will still update the field even when the offload is not enabled.
uint32_t max_rx_pkt_lenfield of
struct rte_eth_rxmode, will be replaced by a new
uint32_t mtufield of
struct rte_eth_confin v21.11. The new
mtufield will be used to configure the initial device MTU via
rte_eth_dev_configure()API. Later MTU can be changed by
rte_eth_dev_set_mtu()API as done now. The existing
(struct rte_eth_dev)->data->mtuvariable will be used to store the configured
mtuvalue, and this new
(struct rte_eth_dev)->data->dev_conf.mtuvariable will be used to store the user configuration request. Unlike
max_rx_pkt_len, which was valid only when
mtufield will be always valid. When
mtuconfig is not provided by the application, default
RTE_ETHER_MTUvalue will be used.
(struct rte_eth_dev)->data->mtushould be updated after MTU set successfully, either by
An application may need to configure device for a specific Rx packet size, like for cases
DEV_RX_OFFLOAD_SCATTERis not supported and device received packet size can’t be bigger than Rx buffer size. To cover these cases an application needs to know the device packet overhead to be able to calculate the
mtucorresponding to a Rx buffer size, for this
(struct rte_eth_dev_info).max_rx_pktlenwill be kept, the device packet overhead can be calculated as:
(struct rte_eth_dev_info).max_rx_pktlen - (struct rte_eth_dev_info).max_mtu
rte_eth_rx_descriptor_donewill be removed in 21.11. Existing
rte_eth_tx_descriptor_statusAPIs can be used as replacement.
ethdev: The port mirroring API can be replaced with a more fine grain flow API. The structs
rte_eth_vlan_mirrorand the functions
rte_eth_mirror_rule_resetwill be marked as deprecated in DPDK 20.11, along with the associated macros
ETH_MIRROR_*. This API will be fully removed in DPDK 21.11.
ethdev: Announce moving from dedicated modify function for each field, to using the general
ethdev: The struct
rte_flow_action_modify_datawill be modified to support modifying fields larger than 64 bits. In addition, documentation will be updated to clarify byte order.
struct rte_flow_action_countis deprecated and will be removed in DPDK 21.11. Shared counters should be managed using shared actions API (
ethdev: Definition of the flow API action
RTE_FLOW_ACTION_TYPE_PORT_IDis ambiguous and needs clarification. Structure
rte_flow_action_port_idwill be extended to specify traffic direction to the represented entity or ethdev port itself in DPDK 21.11.
ethdev: Flow API documentation is unclear if ethdev port used to create a flow rule adds any implicit match criteria in the case of transfer rules. The semantics will be clarified in DPDK 21.11 and it will require fixes in drivers and applications which interpret it in a different way.
ethdev: The flow API matching pattern structures,
struct rte_flow_item_*, should start with relevant protocol header. Some matching pattern structures implements this by duplicating protocol header fields in the struct. To clarify the intention and to be sure protocol header is intact, will replace those fields with relevant protocol header struct. In v21.02 both individual protocol header fields and the protocol header struct will be added as union, target is switch usage to the protocol header by time. In v21.11 LTS, protocol header fields will be cleaned and only protocol header struct will remain.
ethdev: Add new Rx queue offload flag
ethdev: Queue specific stats fields will be removed from
struct rte_eth_stats. Mentioned fields are:
q_errors. Instead queue stats will be received via xstats API. Current method support will be limited to maximum 256 queues. Also compile time flag
RTE_ETHDEV_QUEUE_STAT_CNTRSwill be removed.
ethdev: The offload flag
PKT_RX_EIP_CKSUM_BADwill be removed and replaced by the new flag
PKT_RX_OUTER_IP_CKSUM_BAD. The new name is more consistent with existing outer header checksum status flag naming, which should help in reducing confusion about its usage.
i40e: As there are both i40evf and iavf pmd, the functions of them are duplicated. And now more and more advanced features are developed on iavf. To keep consistent with kernel driver’s name (https://patchwork.ozlabs.org/patch/970154/), i40evf is no need to maintain. Starting from 21.05, the default VF driver of i40e will be iavf, but i40evf can still be used if users specify the devarg “driver=i40evf”. I40evf will be deleted in DPDK 21.11.
rte_ether_hdrstructure will be renamed in DPDK 21.11 to avoid conflict with Windows Sockets headers.
net: The structure
rte_ipv4_hdrwill have two unions. The first union is for existing
version_ihlbyte and new bitfield for version and IHL. The second union is for existing
fragment_offsetand new bitfield for fragment flags and offset.
rte_vdpa_relay_vring_usedvDPA driver interface will be marked as internal in DPDK v21.11.
struct rte_vhost_device_opsin DPDK v21.11.
vhost: The experimental tags of
rte_vhost_driver_set_protocol_featuresfunctions will be removed and the API functions will be made stable in DPDK 21.11.
rte_param_log2_rangestructure will be renamed in DPDK 21.11 to avoid conflict with Windows Sockets headers.
rte_crypto_param_rangestructure will be renamed in DPDK 21.11 to avoid conflict with Windows Sockets headers.
cryptodev: The field
struct rte_crypto_cipher_xformhas a limited size
uint16_t. It will be moved and extended as
uint32_tin DPDK 21.11.
cryptodev: The structure
rte_crypto_sym_vecwould be updated to add
dest_sglto support out of place processing. This field will be null for inplace processing. This change is targeted for DPDK 21.11.
cryptodev: The structure
rte_crypto_vecwould be updated to add
tot_lento support total buffer length. This is required for security cases like IPsec and PDCP encryption offload to know how much additional memory space is available in buffer other than data length so that driver/HW can write expanded size data after encryption. This change is targeted for DPDK 21.11.
cryptodev: Hide structures
rte_cryptodev_asym_sessionto remove unnecessary indirection between session and the private data of session. An opaque pointer can be exposed directly to application which can be attached to the
security: Hide structure
rte_security_sessionand expose an opaque pointer for the private data to the application which can be attached to the packet while enqueuing.
security: The IPsec configuration structure
struct rte_security_ipsec_xformwill be updated with new members to allow SA lifetime configuration. A new structure would be introduced to replace the current member,
security: The structure
rte_security_ipsec_xformwill be extended with multiple fields: source and destination port of UDP encapsulation, IPsec payload MSS (Maximum Segment Size), and ESN (Extended Sequence Number).
security: The IPsec SA config options
struct rte_security_ipsec_sa_optionswill be updated with new fields to support new features like IPsec inner checksum, tunnel header verification, TSO in case of protocol offload.
ipsec: The structure
rte_ipsec_sa_prmwill be extended with a new field
hdr_l3_lento configure tunnel L3 header length.
eventdev: The file
rte_eventdev_pmd.hwill be renamed to
eventdev_driver.hto make the driver interface as internal and the structures
rte_eventdevswill be moved to a new file named
rte_eventdev_core.hin DPDK 21.11. The
rte_prefix for internal structures and functions will be removed across the library. The experimental eventdev trace APIs and
rte_event_eth_rx_adapter_vector_limits_getwill be promoted to stable. An 8-byte reserved field will be added to the structure
rte_event_timerto support future extensions.
eventdev: The structure
rte_event_eth_rx_adapter_queue_confwill be extended to include
rte_event_eth_rx_adapter_event_vector_configelements and the function
rte_event_eth_rx_adapter_queue_event_vector_configwill be removed in DPDK 21.11.
An application can enable event vectorization by passing the desired vector values to the function
rte_event_eth_rx_adapter_queue_addusing the structure
eventdev: Reserved bytes of
rte_event_crypto_requestis a space holder for
response_info. Both should be decoupled for better clarity. New space for
response_infocan be made by changing
rte_event_crypto_metadatatype to structure from union. This change is targeted for DPDK 21.11.
metrics: The function
rte_metrics_initwill have a non-void return in order to notify errors instead of calling
cmdlinestructure will be made opaque to hide platform-specific content. On Linux and FreeBSD, supported prior to DPDK 20.11, original structure will be kept until DPDK 21.11.
security: The functions
rte_security_get_userdatawill be made inline functions and additional flags will be added in structure
rte_security_ctxin DPDK 21.11.
cryptodev: The structure
rte_crypto_opwould be updated to reduce reserved bytes to 2 (from 3), and use 1 byte to indicate warnings and other information from the crypto/security operation. This field will be used to communicate events such as soft expiry with IPsec in lookaside mode.