When a cpu is downed on a system, the irqs on the cpu are assigned to other cpus.  It is possible, however, that when a cpu is downed there aren't enough free vectors on the remaining cpus to account for the vectors from the cpu that is being downed.

This results in an interesting "overflow" condition where irqs are "assigned" to a CPU but are not handled.

For example, when downing cpus on a 1-64 logical processor system:

<snip>
[  232.021745] smpboot: CPU 61 is now offline
[  238.480275] smpboot: CPU 62 is now offline
 [  245.991080] ------------[ cut here ]------------
[  245.996270] WARNING: CPU: 0 PID: 0 at net/sched/sch_generic.c:264 dev_watchdog+0x246/0x250()
[  246.005688] NETDEV WATCHDOG: p786p1 (ixgbe): transmit queue 0 timed out
[  246.013070] Modules linked in: lockd sunrpc iTCO_wdt iTCO_vendor_support sb_edac ixgbe microcode e1000e pcspkr joydev edac_core lpc_ich ioatdma ptp mdio mfd_core i2c_i801 dca pps_core i2c_core wmi acpi_cpufreq isci libsas scsi_transport_sas
[  246.037633] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.12.0+ #14
[  246.044451] Hardware name: Intel Corporation S4600LH ........../SVRBD-ROW_T, BIOS SE5C600.86B.01.08.0003.022620131521 02/26/2013
[  246.057371]  0000000000000009 ffff88081fa03d40 ffffffff8164fbf6 ffff88081fa0ee48
[  246.065728]  ffff88081fa03d90 ffff88081fa03d80 ffffffff81054ecc ffff88081fa13040
[  246.074073]  0000000000000000 ffff88200cce0000 0000000000000040 0000000000000000
[  246.082430] Call Trace:
[  246.085174]  <IRQ>  [<ffffffff8164fbf6>] dump_stack+0x46/0x58
[  246.091633]  [<ffffffff81054ecc>] warn_slowpath_common+0x8c/0xc0
[  246.098352]  [<ffffffff81054fb6>] warn_slowpath_fmt+0x46/0x50
[  246.104786]  [<ffffffff815710d6>] dev_watchdog+0x246/0x250
[  246.110923]  [<ffffffff81570e90>] ? dev_deactivate_queue.constprop.31+0x80/0x80
[  246.119097]  [<ffffffff8106092a>] call_timer_fn+0x3a/0x110
[  246.125224]  [<ffffffff8106280f>] ? update_process_times+0x6f/0x80
[  246.132137]  [<ffffffff81570e90>] ? dev_deactivate_queue.constprop.31+0x80/0x80
[  246.140308]  [<ffffffff81061db0>] run_timer_softirq+0x1f0/0x2a0
[  246.146933]  [<ffffffff81059a80>] __do_softirq+0xe0/0x220
[  246.152976]  [<ffffffff8165fedc>] call_softirq+0x1c/0x30
[  246.158920]  [<ffffffff810045f5>] do_softirq+0x55/0x90
[  246.164670]  [<ffffffff81059d35>] irq_exit+0xa5/0xb0
[  246.170227]  [<ffffffff8166062a>] smp_apic_timer_interrupt+0x4a/0x60
[  246.177324]  [<ffffffff8165f40a>] apic_timer_interrupt+0x6a/0x70
[  246.184041]  <EOI>  [<ffffffff81505a1b>] ? cpuidle_enter_state+0x5b/0xe0
[  246.191559]  [<ffffffff81505a17>] ? cpuidle_enter_state+0x57/0xe0
[  246.198374]  [<ffffffff81505b5d>] cpuidle_idle_call+0xbd/0x200
[  246.204900]  [<ffffffff8100b7ae>] arch_cpu_idle+0xe/0x30
[  246.210846]  [<ffffffff810a47b0>] cpu_startup_entry+0xd0/0x250
[  246.217371]  [<ffffffff81646b47>] rest_init+0x77/0x80
[  246.223028]  [<ffffffff81d09e8e>] start_kernel+0x3ee/0x3fb
[  246.229165]  [<ffffffff81d0989f>] ? repair_env_string+0x5e/0x5e
[  246.235787]  [<ffffffff81d095a5>] x86_64_start_reservations+0x2a/0x2c
[  246.242990]  [<ffffffff81d0969f>] x86_64_start_kernel+0xf8/0xfc
[  246.249610] ---[ end trace fb74fdef54d79039 ]---
[  246.254807] ixgbe 0000:c2:00.0 p786p1: initiating reset due to tx timeout
[  246.262489] ixgbe 0000:c2:00.0 p786p1: Reset adapter
Last login: Mon Nov 11 08:35:14 from 10.18.17.119
[root@(none) ~]# [  246.792676] ixgbe 0000:c2:00.0 p786p1: detected SFP+: 5
[  249.231598] ixgbe 0000:c2:00.0 p786p1: NIC Link is Up 10 Gbps, Flow Control: RX/TX
[  246.792676] ixgbe 0000:c2:00.0 p786p1: detected SFP+: 5
[  249.231598] ixgbe 0000:c2:00.0 p786p1: NIC Link is Up 10 Gbps, Flow Control: RX/TX

(last lines keep repeating.  ixgbe driver is dead until module reload.)

Initially I thought this was a bug or race in the cpu down path wrt to unhandled events in the IRR of the downed cpu.  I debugged that code and confirmed that it works correctly.  I then turned to the ixgbe driver thinking that maybe there was some code there that did 

      if (this_cpu != assigned_cpu_for_queue)
              return -EBIGFATFAILURE;

but there isn't anything like that (thankfully) in the driver.

The issue comes down to this:  If the downed cpu has more vectors than are free on the remaining cpus on the system, it is possible that some vectors are "orphaned" even though they are assigned to a cpu.  In this case, since the ixgbe driver had a watchdog, the watchdog fired and notified that something was wrong.

NOTE:  The ixgbe driver assigned IIUC 2 MSI interrupts for a tx and rx queue/per cpu.  IMO, the ixgbe driver could have a cpu notifier that removes MSIs and queues as cpus go down.

I have a complete hack of a patch for this and am in the process of fixing it up, which checks the total number of available vectors on all cpus and compares it to the number needed by the cpu being downed.  I will post in a while.

P.