Worlds greatest kernel

Sander Vesik sander at haldjas.folklore.ee
Mon Oct 13 02:29:37 PDT 2003


Kyle <kyleNOSPAM at xxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> This approach was originally taken by IBM a long time ago.  The paging
> algorithm eventually only has pages in memory that are used (by either the
> kernel or applications).  The definition of "used" can be quite varied but
> anything that would affect performance as you mentioned below would
> certainly qualify as "used".   The overall affect of this is that one gets a
> "no-fat" in memory execution state regardless of where the "fat" may end up
> (in kernel or user space).
> 

But in reality, paging of kernel memory - especially page tables - is one of the
things that makes SYSV suck. I don't think pagable kernel memory should be used
unless there are concrete real examples that show how it would be useful.

-- 
	Sander

+++ Out of cheese error +++





More information about the Kernel mailing list