[HEADSUP] big polling changes

[Yuriy N. Shkandybin]

Hello.

First of all hank you , Gleb, for your work with polling!

I've noticed next:
 
no polling
CPU states:  0.9% user,  0.0% nice,  5.8% system,  4.1% interrupt, 89.2% idle

polling
CPU states:  1.8% user,  0.0% nice,  5.8% system,  2.9% interrupt, 89.6% idle
1.5 times slowly then without polling enabled

polling + kern.polling.idle_poll=1
CPU states:  0.2% user,  0.0% nice, 59.3% system, 21.3% interrupt, 19.2% idle
but about +10 %  perfomance thoughput

I've tested with em, nge, em+nge.

Jura