Subject: Re: question about vmstat -m
To: Wojciech Puchar <firstname.lastname@example.org>
From: enami tsugutomo <email@example.com>
Date: 01/24/2001 22:03:02
> Date: Wed, 24 Jan 2001 11:53:28 +0100 (CET)
> From: Wojciech Puchar <firstname.lastname@example.org>
> please tell me what is the pools below or where could i search for docs
> about it.
Please refer pool(9) for pool allocater itself. In short, it's a
memory (de)allocater for fixed sized object.
> a) why "rndsample" needs over half a megabyte memory?!
Size is a size of the object.
Requests/Fail/Releases are number of allocation request/failure count
for allocation/release requset.
Pgreq/Pgrel are how many times pages are requested or released (pool
system request memory in terms of ``page'' to VM system, and divide it
into each objects).
Npage is a number of page currently used by pool system for the
So, your system uses 2 pages for ``rndsample''. 2 page is 4k * 2 = 8k
on i386 for example. And 6248 - 6240 = 8 objects are actually
allocated by client of pool system.
> b) mclpl what is it and why is it so large
mclpl is a pool for mbuf cluster, typical the external storage of
mbuf. And, they used just heavily. Currently, just 24k is allocated
for mbuf cluster.
> c) what is sigapl, procpl
They are polls for struct sigacts and struct proc respectively.