7 * **Enabled freeing of ring.**
9 New function rte_ring_free() allows the user to free a ring
10 if it was created with rte_ring_create().
12 * **Extended Statistics**
14 Define extended statistics naming scheme to store metadata in the name
15 string name of each statistic, refer to the Extended Statistics section
16 of the programmers guide. Implemented the extended stats API for these
26 * **Added API in ethdev to retrieve RX/TX queue information.**
28 * Add the ability for the upper layer to query RX/TX queue information.
29 * Add into rte_eth_dev_info new fields to represent information about
30 RX/TX descriptors min/max/align numbers per queue for the device.
32 * **Added RSS dynamic configuration to bonding.**
34 * **Added e1000 Rx interrupt support.**
36 * **Added igb TSO support for both PF and VF.**
38 * **RSS enhancement on Intel x550 NIC**
40 * Support 512 entries RSS redirection table.
41 * Support per VF RSS redirection table.
43 * **Flow director enhancement on Intel x550 NIC**
45 * Add 2 new flow director modes on x550.
46 * One is MAC VLAN mode, the other is tunnel mode.
48 * **Added i40e vector RX/TX.**
50 * **Added i40e Rx interrupt support.**
52 * **Added i40e flow control support.**
54 * **Added DCB support to i40e PF driver.**
56 * **Added RSS/FD input set granularity on Intel X710/XL710.**
58 * **Added different GRE key length for input set on Intel X710/XL710.**
60 * **Added flow director support in i40e VF.**
62 * **Added fm10k vector RX/TX.**
64 * **Added fm10k TSO support for both PF and VF.**
66 * **Added fm10k VMDQ support.**
68 * **New NIC Boulder Rapid support.**
70 Boulder Rapid is a new NIC of Intel's fm10k family.
72 * **Enhanced support for the Chelsio CXGBE driver.**
74 * Added support for Jumbo Frames.
75 * Optimize forwarding performance for Chelsio T5 40GbE cards.
77 * **Added support for Mellanox ConnectX-4 adapters (mlx5).**
79 The mlx5 poll-mode driver implements support for Mellanox ConnectX-4 EN
80 and Mellanox ConnectX-4 Lx EN families of 10/25/40/50/100 Gb/s adapters.
82 Like mlx4, this PMD is only available for Linux and is disabled by default
83 due to external dependencies (libibverbs and libmlx5).
85 * **Enhanced support for virtio driver.**
87 * Virtio ring layout optimization (fixed avail ring)
91 * **Added vhost-user multiple queue support.**
93 * **Added port hotplug support to vmxnet3.**
95 * **Added port hotplug support to xenvirt.**
104 * **eal/linux: Fixed epoll timeout.**
106 Fixed issue where the ``rte_epoll_wait()`` function didn't return when the
107 underlying call to ``epoll_wait()`` timed out.
113 * **igb: Fixed IEEE1588 frame identification in I210.**
115 Fixed issue where the flag PKT_RX_IEEE1588_PTP was not being set
116 in Intel I210 NIC, as EtherType in RX descriptor is in bits 8:10 of
117 Packet Type and not in the default bits 0:2.
119 * **igb: Fixed VF start with PF stopped.**
121 VF needs the PF interrupt support initialized even if not started.
123 * **ixgbe: Fixed issue with X550 DCB.**
125 Fixed a DCB issue with x550 where for 8 TCs (Traffic Classes), if a packet
126 with user priority 6 or 7 was injected to the NIC, then the NIC would only
127 put 3 packets into the queue. There was also a similar issue for 4 TCs.
129 * **ixgbe: Removed burst size restriction of vector RX.**
131 Fixed issue where a burst size less than 32 didn't receive anything.
133 * **ixgbe: Fixed VF start with PF stopped.**
135 VF needs the PF interrupt support initialized even if not started.
137 * **i40e: Fixed base driver allocation when not using first numa node.**
139 Fixed i40e issue that occurred when a DPDK application didn't initialize
140 ports if memory wasn't available on socket 0.
142 * **i40e: Fixed maximum of 64 queues per port.**
144 Fixed the issue in i40e of cannot supporting more than 64 queues per port,
145 though hardware actually supports that. The real number of queues may vary,
146 as long as the total number of queues used in PF, VFs, VMDq and FD does not
147 exceeds the hardware maximum.
149 * **i40e: Fixed statistics of packets.**
151 Added discarding packets on VSI to the stats and rectify the old statistics.
153 * **vhost: Fixed Qemu shutdown.**
155 Fixed issue with libvirt ``virsh destroy`` not killing the VM.
157 * **virtio: Fixed crash after changing link state.**
159 Fixed io permission in the interrupt handler.
161 * **virtio: Fixed crash when releasing queue.**
163 Fixed issue when releasing null control queue.
165 * **hash: Fixed thread scaling by reducing contention.**
167 Fixed issue in hash library where, using multiple cores with
168 hardware transactional memory support, thread scaling did not work,
169 due to the global ring that is shared by all cores.
175 * **hash: Fixed memory allocation of Cuckoo Hash key table.**
177 Fixed issue where an incorrect Cuckoo Hash key table size could be
178 calculated limiting the size to 4GB.
180 * **hash: Fixed incorrect lookup if key is all zero.**
182 Fixed issue in hash library that occurred if an all zero
183 key was not added in the table and the key was looked up,
184 resulting in an incorrect hit.
202 * The deprecated flow director API is removed.
203 It was replaced by rte_eth_dev_filter_ctrl().
205 * The dcb_queue is renamed to dcb_tc in following dcb configuration
206 structures: rte_eth_dcb_rx_conf, rte_eth_dcb_tx_conf,
207 rte_eth_vmdq_dcb_conf, rte_eth_vmdq_dcb_tx_conf.
209 * The function rte_eal_pci_close_one() is removed.
210 It was replaced by rte_eal_pci_detach().
212 * The deprecated ACL API ipv4vlan is removed.
214 * The deprecated hash function rte_jhash2() is removed.
215 It was replaced by rte_jhash_32b().
217 * The deprecated KNI functions are removed:
218 rte_kni_create(), rte_kni_get_port_id() and rte_kni_info_get().
220 * The deprecated ring PMD functions are removed:
221 rte_eth_ring_pair_create() and rte_eth_ring_pair_attach().
223 * The devargs union field virtual is renamed to virt for C++ compatibility.
229 * The EAL and ethdev structures rte_intr_handle and rte_eth_conf were changed
230 to support Rx interrupt. It was already done in 2.1 for CONFIG_RTE_NEXT_ABI.
232 * The ethdev flow director entries for SCTP were changed.
233 It was already done in 2.1 for CONFIG_RTE_NEXT_ABI.
235 * The ethdev flow director structure rte_eth_fdir_flow_ext structure is changed.
236 New fields are added to support flow director filtering in VF.
238 * The size of the ethdev structure rte_eth_hash_filter_info is changed
239 by adding a new element rte_eth_input_set_conf in an union.
241 * The new fields rx_desc_lim and tx_desc_lim are added into rte_eth_dev_info
244 * The maximum number of queues per port CONFIG_RTE_MAX_QUEUES_PER_PORT is
247 * The mbuf structure was changed to support unified packet type.
248 It was already done in 2.1 for CONFIG_RTE_NEXT_ABI.
250 * The dummy malloc library is removed. The content was moved into EAL in 2.1.
252 * The LPM structure is changed. The deprecated field mem_location is removed.
254 * librte_table LPM: A new parameter to hold the table name will be added to
255 the LPM table parameter structure.
257 * librte_port: Macros to access the packet meta-data stored within the packet
258 buffer has been adjusted to cover the packet mbuf structure.
260 * librte_cfgfile: Allow longer names and values by increasing the constants
261 CFG_NAME_LEN and CFG_VALUE_LEN to 64 and 256 respectively.
264 Shared Library Versions
265 -----------------------
267 The libraries prepended with a plus sign were incremented in this version.
273 + librte_cfgfile.so.2
275 librte_distributor.so.1
289 + librte_pmd_ring.so.2