Per-interface polling(4) status

[Chuck Swiger]

Ruslan Ermilov wrote:
[ ... ]
> Points I'm not sure about, and would like to hear the comments on:
> 
> - Should the initial polling status be on or off, on interfaces
>   that support it?  The current version of the patch makes it
>   off by default.

Given that polling will only take place if kern.polling.enable is set, 
IFF_POLLING can be enabled by default safely and changing the global sysctl 
will enable or disable all polling-capable interfaces, which retains the 
current behavior.

> - Should I add the code to ether_poll_register() to sanity check
>   that the interface doesn't attempt to register itself more
>   than once?

Sure.  I don't see how having an interface registered multiple times will do 
any good....?

-- 
-Chuck