Comment 119 for bug 1403152

(In reply to Kevin from comment #15)
> Jan 11 16:17:44 <hostname> kernel: [ 2197.614240] lodebug: dev_hold 11
> dst_init
>
> (after __neigh_create function call, the reference count for loopback device
> is incorrectly increased to 13)
>
> Jan 11 16:20:41 <hostname> kernel: [ 2374.898917] lodebug: dev_hold 13
> __neigh_create

This is hard to believe or explain. The only way to increase the counter is to call dev_hold(), and whenever dev_hold() is called, a message saying so would be logged. The only explanation I can think of would be if the memory area holding the per-cpu counters would have been accidentally overwritten (say by overrun of an array in nearby memory).

But a more likely explanation is that there is a missing line in the log, either due to copy and paste error, or because somehow the message never made it to the log buffer (are there still contexts in which printk doesn't work?)

Question: do you see the same apparent +2 counter increase whenever you hit the "waiting for <dev> to become free" error, or was it a one-time thing?