vm_page: Track number of allocated NOFREE pages

[markj]

It's more a matter of taste than anything I guess, but I'd merge this with the previous patch. They are small enough that they can be reviewed as one patch, and it feels a bit silly to have a change which introduces a counter but doesn't use it. I almost pointed that out in the other revision before I saw this one, since it looked like an unintentional omission.

sys/vm/vm_page.c

2596

This can be done outside the lock; counter(9) is implemented using per-CPU counters that do not require explicit synchronization.

[bnovkov]

In D46979#1070935, @markj wrote:

It's more a matter of taste than anything I guess, but I'd merge this with the previous patch. They are small enough that they can be reviewed as one patch, and it feels a bit silly to have a change which introduces a counter but doesn't use it. I almost pointed that out in the other revision before I saw this one, since it looked like an unintentional omission.

Sure, I was on the fence about that from the beginning, I'll go ahead and merge it into the parent review.