lkml.org 
[lkml]   [1996]   [Nov]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Pageable kernel (was Bugs and wishes in memory management area)

Date: Wed, 20 Nov 1996 17:23:07 +1100 (EST)
From: Keith Owens <kaos@edison.dialix.com.au>

On Wed, 20 Nov 1996, David S. Miller wrote:
> No, you cannot page the memory a module uses. I leave the reader to
> consider the monsterous races assosciated with a kernel that can page
> itself.

It is not impossible in general but the specific design of the Linux
kernel makes it very difficult. For example, MVS can page most of the O/S
pages. The MVS nucleus ("kernel") is split into pageable and non-pageable
(fixed) areas, the former even use the MMU to map nucleus addresses to
physical addresses and the physical address changes everytime they are
paged in.

Just because it is possible doesn't mean it is not a stupid idea.

I think linux will become pagable as soon as it also begins to do
system calls via RPC...

Oh, and on fine grained locking, another example of how to "get there"
but again a stupid way to do it. The answer is not spin locks all
over the kernel, every textbook mutual exclusion primitive known to
man, and whatnot, to get SMP. I leave such monstrosities to Solaris
and others of it's kin, some of which need programs to scan their code
for race conditions.

I quit the day we need that sort of garbage for Linux.

The answer is with self locking data structures and careful coding. A
well thought out design, and not something meant to meet a deadline.

--------------------------------------------////
Yow! 10.49 MB/s remote host TCP bandwidth ////
over 100Mb/s ethernet. Beat that! ////
-----------------------------------------////__________ o
David S. Miller, davem@caip.rutgers.edu /_____________/ / // /_/ ><

\
 
 \ /
  Last update: 2005-03-22 13:38    [W:0.075 / U:0.016 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site