On systems with 128 CPUs, turning off TSO results in errors.
Errors: i40e 0000:03:00.0: failed to get tracking for 1 vectors for VSI 400, err=-12 i40e 0000:03:00.0: Couldn't create FDir VSI i40e 0000:03:00.0: i40e_ptp_init: PTP not supported on eth0 i40e 0000:03:00.0: couldn't add VEB, err I40E_ERR_ADMIN_QUEUE_ERROR aq_err I40E_AQ_RC_ENOENT i40e 0000:03:00.0: rebuild of switch failed: -1, will try to set up simple PF connection i40e 0000:03:00.0 eth0: adding 00:10:e0:8a:24:b6 vid=0 From kernel log: i40e: Intel(R) Ethernet Connection XL710 Network Driver - version 1.4.8-k i40e: Copyright (c) 2013 - 2014 Intel Corporation. i40e 0000:03:00.0: fw 4.40.35115 api 1.4 nvm 4.53 0x80001e8c 0.0.0 i40e 0000:03:00.0: MAC address: 00:10:e0:8a:24:b6 i40e 0000:03:00.0: i40e_ptp_init: PTP not supported on eth0 i40e 0000:03:00.0: PCI-Express: Speed 8.0GT/s Width x8 i40e 0000:03:00.0: Features: PF-id[0] VFs: 32 VSIs: 34 QP: 128 RX: 1BUF RSS FD_ATR VxLAN VEPA As per the log above, feature FD_SB (sideband flow director)) is not enabled. Because there are no enough MSI-X vectors available. (Device function caps report 129 MSI-X vectors in this case. And driver reserved 1 of them for misc interrupt and rest 128 for 128 QP. So no vector left for FD_SB. Therefore driver disables FD_SB) However turning off TSO invokes i40e_set_ntuple() that enables FD_SB and returns true, issues reset in i40e_set_features() i.e i40e_do_reset. Later during reset, driver fails to find irq for FD_SB from irq pile (and it won't because there was no irq vector assigned for FD_SB). This results in the very first error, 'i40e 0000:03:00.0: failed to get tracking for 1 vectors for VSI 400, err=-12' I believe before enabling FD_SB in i40e_set_ntuple(), driver should check if MSI-X vector available for FD_SB. Sending patch in separate email. (FWIW, if number of CPUs reduced to 64, I don't see the issue described above because in that case out of 129 MSI-X vectors only 64 get assigned to QP. Remaining are used for features like FD_SB) -Tushar
