seg_map and cachelist
From: star night (star_night_at_my-deja.com)
Date: 06/29/04
- Next message: noone: "Moving boot disk from a UltraAX-i2 to a UltraSPARC-IIi-cEngine"
- Previous message: Casper H.S. ***: "Re: Is iostream/fstream thread safe ?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 29 Jun 2004 02:36:19 -0700
While trying to understand Solaris file system caching process(Solaris
8), I am confused about couple of points
1. Is 12% of physical memory, the max possible for file system cache
or is it fixed with architecture(sun4u - 256MB --seg_map segment
size)?. I am inclined to think the former and seg_map segment is more
like a temporary place holder.
2. this question is based on Sun Doc ID74347 which I quote some of it
"Seg_map manages file system cache. When seg map uses
up all the available mapping, it starts paging against itself by
releasing
the pages that are not mapped and not dirty, into the cachelist and
thus
makes room for incoming pages filled by read()/write(). However, if
the
pages getting relased are dirty, they are not moved into the cache
list"
Does those mean that when seg_map is paged there is an additional
memory to memory copy from seg_map to the virtual address
representing that slot? and then mapping gets added to the cache list?
Thanks
SN
- Next message: noone: "Moving boot disk from a UltraAX-i2 to a UltraSPARC-IIi-cEngine"
- Previous message: Casper H.S. ***: "Re: Is iostream/fstream thread safe ?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]