Subject: kern/9423: Cleanup/consolidation of MAXSLP
To: None <email@example.com>
From: None <firstname.lastname@example.org>
Date: 02/15/2000 13:05:34
>Synopsis: Cleanup/consolidation of MAXSLP
>Responsible: kern-bug-people (Kernel Bug People)
>Arrival-Date: Tue Feb 15 13:03:00 2000
>Originator: Erik Bertelsen
>Release: NetBSD-current 15 Feb 2000
System: NetBSD erik-be.uni-c.dk 1.4S NetBSD 1.4S (ERIKBE) #347: Mon Feb 14 07:51:33 MET 2000 email@example.com:/sw/NetBSD/src/sys/arch/i386/compile/ERIKBE i386
Recently (while hunting something else) I discovered that all NetBSD ports
define MAXSLP in arch/<port>/include/vmparam.h in the following way:
* The time for a process to be blocked before being very swappable.
* This is a number of seconds which the system takes as being a non-trivial
* amount of real time. You probably shouldn't change this;
* it is used in subtle ways (fractions and multiples of it are, that is, like
* half of a ``long time'', almost a long time, etc.)
* It is related to human patience and other factors which don't really
* change over time.
#define MAXSLP 20
The only use of MAXSLP in the kernel is to initialize maxslp in uvm/uvm_meter.c,
and maxslp is only used in that file and in uvm_glue.c.
The comment cited above is not entirely correct, as maxslp is used in straight
comparisons, and in one case it is halved, but no other fractions, multiples,
Move the definition of MAXSLP into a MI header file, possibly <uvm/???.h>
and possibly update the comment -- then delete all the identicatl definitions
from all the MD vmparam.h's (more than 20 of them).