+EAL
+~~~
+
+* **eal/linux: Fixed epoll timeout.**
+
+ Fixed issue where the ``rte_epoll_wait()`` function didn't return when the
+ underlying call to ``epoll_wait()`` timed out.
+
+
+Drivers
+~~~~~~~
+
+* **e1000/base: Synchronize PHY interface on non-ME systems.**
+
+ On power up, the MAC - PHY interface needs to be set to PCIe, even if the
+ cable is disconnected. In ME systems, the ME handles this on exit from the
+ Sx (Sticky mode) state. In non-ME, the driver handles it. Added a check for
+ non-ME system to the driver code that handles it.
+
+* **e1000/base: Increased timeout of reset check.**
+
+ Previously, in ``check_reset_block`` RSPCIPHY was polled for 100 ms before
+ determining that the ME veto was set. This was not enough and it was
+ increased to 300 ms.
+
+* **e1000/base: Disabled IPv6 extension header parsing on 82575.**
+
+ Disabled IPv6 options as per hardware limitation.
+
+* **e1000/base: Prevent ULP flow if cable connected.**
+
+ Enabling ULP on link down when the cable is connected caused an infinite
+ loop of link up/down indications in the NDIS driver.
+ The driver now enables ULP only when the cable is disconnected.
+
+* **e1000/base: Support different EEARBC for i210.**
+
+ EEARBC has changed on i210. It means EEARBC has a different address on
+ i210 than on other NICs. So, add a new entity named EEARBC_I210 to the
+ register list and make sure the right one is being used on i210.
+
+* **e1000/base: Fix K1 configuration.**
+
+ Added fix for the following updates to the K1 configurations:
+ TX idle period for entering K1 should be 128 ns.
+ Minimum TX idle period in K1 should be 256 ns.
+
+* **e1000/base: Fix link detect flow.**
+
+ Fix link detect flow in case where auto-negotiate is not enabled, by calling
+ ``e1000_setup_copper_link_generic`` instead of ``e1000_phy_setup_autoneg``.
+
+* **e1000/base: Fix link check for i354 M88E1112 PHY.**
+
+ The ``e1000_check_for_link_media_swap()`` function is supposed to check PHY
+ page 0 for copper and PHY page 1 for "other" (fiber) links. The driver
+ switched back from page 1 to page 0 too soon, before
+ ``e1000_check_for_link_82575()`` is executed and was never finding the link
+ on the fiber (other).
+
+ If the link is copper, as the M88E1112 page address is set to 1, it should be
+ set back to 0 before checking this link.
+
+* **e1000/base: Fix beacon duration for i217.**
+
+ Fix for I217 Packet Loss issue - The Management Engine sets the FEXTNVM4
+ Beacon Duration incorrectly. This fix ensures that the correct value will
+ always be set. Correct value for this field is 8 usec.
+
+* **e1000/base: Fix TIPG for non 10 half duplex mode.**
+
+ TIPG value is increased when setting speed to 10 half duplex to prevent
+ packet loss. However, it was never decreased again when speed
+ changed. This caused performance issues in the NDIS driver.
+ Fix this to restore TIPG to default value on non 10 half duplex.
+
+* **e1000/base: Fix reset of DH89XXCC SGMII.**
+
+ For DH89XXCC_SGMII, a write flush leaves registers of this device trashed
+ (0xFFFFFFFF). Add check for this device.
+
+ Also, after both Port SW Reset and Device Reset case, the platform should
+ wait at least 3ms before reading any registers. Remove this condition since
+ waiting is conditionally executed only for Device Reset.
+
+* **e1000/base: Fix redundant PHY power down for i210.**
+
+ Bit 11 of PHYREG 0 is used to power down PHY. The use of PHYREG 16 is
+ no longer necessary.
+
+* **e1000/base: fix jumbo frame CRC failures.**
+
+ Change the value of register 776.20[11:2] for jumbo mode from 0x1A to 0x1F.
+ This is to enlarge the gap between read and write pointers in the TX FIFO.
+
+* **e1000/base: Fix link flap on 82579.**
+
+ Several customers have reported a link flap issue on 82579. The symptoms
+ are random and intermittent link losses when 82579 is connected to specific
+ switches. the Issue was root caused as an inter-operability problem between
+ the NIC and at least some Broadcom PHYs in the Energy Efficient Ethernet
+ wake mechanism.
+
+ To fix the issue, we are disabling the Phase Locked Loop shutdown in 100M
+ Low Power Idle. This solution will cause an increase of power in 100M EEE
+ link. It may cost an additional 28mW in this specific mode.
+
+* **igb: Fixed IEEE1588 frame identification in I210.**
+
+ Fixed issue where the flag ``PKT_RX_IEEE1588_PTP`` was not being set
+ in the Intel I210 NIC, as the EtherType in RX descriptor is in bits 8:10 of
+ Packet Type and not in the default bits 0:2.
+
+* **igb: Fixed VF start with PF stopped.**
+
+ VF needs the PF interrupt support initialized even if not started.
+
+* **igb: Fixed VF MAC address when using with DPDK PF.**
+
+ Assign a random MAC address in VF when not assigned by PF.
+
+* **igb: Removed CRC bytes from byte counter statistics.**
+
+* **ixgbe: Fixed issue with X550 DCB.**
+
+ Fixed a DCB issue with x550 where for 8 TCs (Traffic Classes), if a packet
+ with user priority 6 or 7 was injected to the NIC, then the NIC would only
+ put 3 packets into the queue. There was also a similar issue for 4 TCs.
+
+* **ixgbe: Removed burst size restriction of vector RX.**
+
+ Fixed issue where a burst size less than 32 didn't receive anything.
+
+* **ixgbe: Fixed VF start with PF stopped.**
+
+ VF needs the PF interrupt support initialized even if not started.
+
+* **ixgbe: Fixed TX hang when RS distance exceeds HW limit.**
+
+ Fixed an issue where the TX queue can hang when a lot of highly fragmented
+ packets have to be sent. As part of that fix, ``tx_rs_thresh`` for ixgbe PMD
+ is not allowed to be greater then to 32 to comply with HW restrictions.
+
+* **ixgbe: Fixed rx error statistic counter.**
+
+ Fixed an issue that the rx error counter of ixgbe was not accurate. The
+ mac short packet discard count (mspdc) was added to the counter. Mac local
+ faults and mac remote faults are removed as they do not count packets but
+ errors, and jabber errors were removed as they are already accounted for
+ by the CRC error counter. Finally the XEC (l3 / l4 checksum error) counter
+ was removed due to errata, see commit 256ff05a9cae for details.
+
+* **ixgbe: Removed CRC bytes from byte counter statistics.**
+
+* **i40e: Fixed base driver allocation when not using first numa node.**
+
+ Fixed i40e issue that occurred when a DPDK application didn't initialize
+ ports if memory wasn't available on socket 0.
+
+* **i40e: Fixed maximum of 64 queues per port.**
+
+ Fixed an issue in i40e where it would not support more than 64 queues per
+ port, even though the hardware actually supports it. The real number of
+ queues may vary, as long as the total number of queues used in PF, VFs, VMDq
+ and FD does not exceeds the hardware maximum.
+
+* **i40e: Fixed statistics of packets.**
+
+ Added discarding packets on VSI to the stats and rectify the old statistics.
+
+* **i40e: Fixed issue of not freeing memzone.**
+
+ Fixed an issue of not freeing a memzone in the call to free the memory for
+ adminq DMA.
+
+* **i40e: Removed CRC bytes from byte counter statistics.**
+
+* **mlx: Fixed driver loading.**
+
+ The mlx drivers were unable to load when built as a shared library,
+ due to a missing symbol in the mempool library.
+
+* **mlx4: Performance improvements.**
+
+ Fixed bugs in TX and RX flows that improves mlx4 performance.
+
+* **mlx4: Fixed TX loss after initialization.**
+
+* **mlx4: Fixed scattered TX with too many segments.**
+
+* **mlx4: Fixed memory registration for indirect mbuf data.**
+
+* **vhost: Fixed Qemu shutdown.**
+
+ Fixed issue with libvirt ``virsh destroy`` not killing the VM.
+
+* **virtio: Fixed crash after changing link state.**
+
+ Fixed IO permission in the interrupt handler.
+
+* **virtio: Fixed crash when releasing queue.**
+
+ Fixed issue when releasing null control queue.
+
+
+Libraries
+~~~~~~~~~
+
+* **hash: Fixed memory allocation of Cuckoo Hash key table.**
+
+ Fixed issue where an incorrect Cuckoo Hash key table size could be
+ calculated limiting the size to 4GB.
+
+* **hash: Fixed incorrect lookup if key is all zero.**
+
+ Fixed issue in hash library that occurred if an all zero
+ key was not added to the table and the key was looked up,
+ resulting in an incorrect hit.
+
+* **hash: Fixed thread scaling by reducing contention.**
+
+ Fixed issue in the hash library where, using multiple cores with
+ hardware transactional memory support, thread scaling did not work,
+ due to the global ring that is shared by all cores.
+
+
+Examples
+~~~~~~~~
+
+* **l3fwd: Fixed crash with IPv6.**
+
+* **vhost_xen: Fixed compile error.**
+
+
+Other
+~~~~~
+
+* This release drops compatibility with Linux kernel 2.6.33. The minimum
+ kernel requirement is now 2.6.34.
+