[MLton-devel] Kernel Mode Linux (KML)
Henry Cejtin
henry@sourcelight.com
Mon, 29 Apr 2002 16:08:18 -0500
- Previous message: [MLton-devel] CVSROOT avail,NONE,1.1 commit_prep.pl,NONE,1.1 commitcheck,NONE,1.1 cvs_acls.pl,NONE,1.1 log_accum.pl,NONE,1.1 checkoutlist,1.3,1.4 commitinfo,1.3,1.4 loginfo,1.11,1.12
- Messages sorted by:
[ date ]
[ thread ]
[ subject ]
[ author ]
I saw this a while ago. Note, it really is basically a joke. Even the
claims `faster than ordinary system calls in an order of magnitude', in
addition to being grammatically incorrect, are misleading. It is true that
the overhead of a procedure call vs. a system call is about a factor of 10,
but unless the system call does almost nothing, it doesn't effect performance
by any significant amount.
As to could we port to it, it would certainly not be trivial. A kernel space
thread does not control the memory mapping it sees. There are no really
effective ways of getting large continuous blocks of memory. Some of this
has, I remember reading, gotten a bit better in the 2.4 kernels, but not
much. In the 2.2 kernels, even getting 2 adjacent pages was hard (although
do-able). Thus you have to get memory in 4K chunks.
_______________________________________________
MLton-devel mailing list
MLton-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlton-devel
- Previous message: [MLton-devel] CVSROOT avail,NONE,1.1 commit_prep.pl,NONE,1.1 commitcheck,NONE,1.1 cvs_acls.pl,NONE,1.1 log_accum.pl,NONE,1.1 checkoutlist,1.3,1.4 commitinfo,1.3,1.4 loginfo,1.11,1.12
- Messages sorted by:
[ date ]
[ thread ]
[ subject ]
[ author ]