Hi,
I just got a pair of these cards, and I've already ran into some serious issues. I'm running a KVM host on CentOS (6.6, kernel version 2.6.32-504.1.3.el6.x86_64), and one of the VMs, which is a HTTP proxy with moderate load (around 200mbit/s) with a routed interface, causes the host to crash in seconds with the bundled i40e driver (0.3.9). I upgraded the driver to the latest version (1.0.15) which doesn't crash the kernel anymore, but still drops the link quite frequently: Dec 13 18:20:04 88 kernel: i40e 0000:03:00.1: TX driver issue detected, PF reset issued Dec 13 18:20:05 88 kernel: i40e 0000:03:00.1: i40e_ptp_init: added PHC on eth4 Dec 13 18:20:05 88 kernel: i40e 0000:03:00.1: eth4: NIC Link is Up 10 Gbps Full Duplex, Flow Control: None Dec 13 18:20:05 88 kernel: i40e 0000:03:00.1: eth4: NIC Link is Down Dec 13 18:20:08 88 kernel: i40e 0000:03:00.1: eth4: NIC Link is Up 10 Gbps Full Duplex, Flow Control: None Dec 13 18:20:47 88 kernel: i40e 0000:03:00.1: TX driver issue detected, PF reset issued Dec 13 18:20:48 88 kernel: i40e 0000:03:00.1: i40e_ptp_init: added PHC on eth4 Dec 13 18:20:48 88 kernel: i40e 0000:03:00.1: eth4: NIC Link is Up 10 Gbps Full Duplex, Flow Control: None Dec 13 18:20:48 88 kernel: i40e 0000:03:00.1: eth4: NIC Link is Down Dec 13 18:20:51 88 kernel: i40e 0000:03:00.1: eth4: NIC Link is Up 10 Gbps Full Duplex, Flow Control: None Dec 13 18:20:52 88 kernel: i40e 0000:03:00.1: TX driver issue detected, PF reset issued Dec 13 18:20:52 88 kernel: i40e 0000:03:00.1: i40e_ptp_init: added PHC on eth4 Dec 13 18:20:52 88 kernel: i40e 0000:03:00.1: eth4: NIC Link is Up 10 Gbps Full Duplex, Flow Control: None Dec 13 18:20:52 88 kernel: i40e 0000:03:00.1: eth4: NIC Link is Down This happens even with GRO turned off. I ruled out hardware issues, because exactly the same happens with the second card in a second server. There's one more very troubling message I got from the driver during testing with a different SFP+: i40e 0000:03:00.0: The driver failed to link because an unqualified module was detected. Please don't start the same nonsense as with ixgbe, and provide us with an option to allow 'unqualified' transceivers to work. I've also tried intel.com for the latest driver, and it offered me ixgbe-3.22.3 for this card, but that 1, doesn't compile 2, doesn't support this card (according to the pci ids listed in ixgbe_type.h). Please fix that as well if you can. Thanks, Matyas PS: My cards are 8086:1572 (Intel Corporation Ethernet Controller X710 for 10GbE SFP+ (rev 01)), and the compile error for ixgbe was the following: In file included from /var/lib/dkms/ixgbe/3.22.3/build/ixgbe_main.c:51: /var/lib/dkms/ixgbe/3.22.3/build/ixgbe.h: In function 'ixgbe_qv_unlock_napi': /var/lib/dkms/ixgbe/3.22.3/build/ixgbe.h:572: error: too many arguments to function 'napi_gro_flush' /var/lib/dkms/ixgbe/3.22.3/build/ixgbe_main.c: At top level: /var/lib/dkms/ixgbe/3.22.3/build/ixgbe_main.c:9087: error: unknown field 'ndo_busy_poll' specified in initializer /var/lib/dkms/ixgbe/3.22.3/build/ixgbe_main.c:9087: warning: initialization from incompatible pointer type ------------------------------------------------------------------------------ Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server from Actuate! Instantly Supercharge Your Business Reports and Dashboards with Interactivity, Sharing, Native Excel Exports, App Integration & more Get technology previously reserved for billion-dollar corporations, FREE http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk _______________________________________________ E1000-devel mailing list E1000-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/e1000-devel To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
