Subject: Re: UBC vs page_idle_zero
To: None <>
From: Paul A Vixie <>
List: tech-kern
Date: 11/28/2000 10:26:02
i don't think the penalty for bzero() is all that high, and i don't mind
doing it on demand.  any threshold we pick will be suboptimal for SOMEONE
and adaptively deciding the right threshold for a given workload is a
recipe for bugs.  as long as we're synching dirty pages so that anything
with refcount==0 is immediately bzero()'able and usable, we'll do fine.