Mellanox Linux Driver Source code for Mellanox ConnectX-3, ConnectX-3 Pro, ConnectX-4 Lx and ConnectX-4 Ethernet adapters, Mellanox OFED_LINUX-4.1-1.0.2.0

Fixes:
------
- Fixed the issue where a compilation error in kernels of v4.6 or above used to occur due to a large stack size in the get_numa_phys_mask function.
- Fixed the issue of when prof_sel was invalid, mlx5_core failed upon debug print.
- Fixed the issue of when EEH was injected and the mlx4 tear down code was called, the eqs were not released, causing a page fault.
- Fixed the issue where kernel panic in mlx4_en_get_phys_port_id may occur during server reboot.
- Fixed the issue where kernel panic was observed on openibd stop as a result of querying non-existent bond slave.
- Fixed the issue of when the error flow was re-factored, the reading of the device caps was excluded from the error recovery flow.
- Fixed the issue where biosdavename running on Redhat 6.x with MLNX_OFED may show the same name to ConnectX-3 Eth port 1 and ConnectX-3 Eth port 2.
- Fixed the issue where memory allocation for CQ buffers used to fail when increasing the RX ring size.
- [ConnectX-3] Fixed the issue where multicast traffic over Raw Ethernet QP on virtual functions were received on the same QP (loopback).
- Fixed the issue of low performance when creating many address handles.
- Fixed an issue that caused TCP packets to be received in an out of order manner when Large Receive Offload (LRO) is on.
- Fixed the issue of when Spoof-check may have been turned on for MAC address 00:00:00:00:00:00
- Fixed the issue where very high stress on DC QP transport might have triggered NMI messages on specific servers.
- Fixed the issue of when the Kernel becomes out of memory upon driver start, it could crash on SLES 12 SP2.
- Fixed an issue where RDMACM connection used to fail upon high connection rate accompanied with the error message: RDMA_CM_EVENT_UNREACHABLE.
- Fixed the issue where an rdma_cm connection between a client and a server that were on the same host was not possible when working over VLAN interfaces.
- Fixed the issue where a race condition in the RoCE GID cache used to cause for the loss of IP-based GIDs.
- Fixed the issue where a kernel crash used to occur when RXe device was coupled with a virtual (dummy) device.
- Fixed SKB memory leak issue that was introduced in kernel 4.11, and added warning messages to the Soft RoCE driver for easy detection of future SKB leaks.
- Added log debug prints when changing HW configuration via DCB. To enable log debug prints, run: ethtool -s msglvl hw on/off
- Fixed a backport issue where IPv6 procedures were called while they were not supported in the underlying kernel.

Enhancements:
-----------------
- Changed the default RoCE mode on which RDMA CM runs to RoCEv2 instead of RoCEv1. RDMA_CM session requires both the client and server sides to support the same RoCE mode. Otherwise, the client will fail to connect to the server.

For mlx5 Driver:
- Added support for additional RoCE diagnostics and ECN congestion counters under /sys/class/infiniband/mlx5_0/ports/1/hw_counters/ directory.
- Added support for rx-fcs ethtool offload configuration. Normally, the FCS of the packet will be truncated by the ASIC hardware before sending it to the application socket buffer (skb). Ethtool allows to set the rx-fcs not to be truncated, but to pass it to the application for analysis.
- Added the option to enable PFC based on the DSCP value. Using this solution, VLAN headers will no longer be mandatory for use.
- ECN parameters have been moved to the following directory:/sys/kernel/debug/mlx5//cc_params/
- Added support for mlx_fs_dump, which is a python tool that prints the steering rules in a readable manner.
- Added the ability to disable probed VFs on the hypervisor.
- Improved performance by rendering Local loopback (unicast and multicast) disabled by mlx5 driver by default while local loopback is not in use.