Subject: bin/29981: vmstat(1) shows wrong system calls per interval
To: None <gnats-admin@netbsd.org, netbsd-bugs@netbsd.org>
From: Rui Paulo <rpaulo@netbsd-pt.org>
List: netbsd-bugs
Date: 04/15/2005 19:37:00
>Number: 29981
>Category: bin
>Synopsis: vmstat(1) shows wrong system calls per interval
>Confidential: no
>Severity: serious
>Priority: low
>Responsible: bin-bug-people
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Fri Apr 15 19:37:00 +0000 2005
>Originator: Rui Paulo <rpaulo@netbsd-pt.org>
>Release: NetBSD 3.0_BETA
>Organization:
>Environment:
System: NetBSD proton.fnop.net 3.0_BETA NetBSD 3.0_BETA (PROTON) #1: Thu Mar 24 20:38:18 WET 2005 rpaulo@proton.fnop.net:/home/rpaulo/netbsd/obj/sys/arch/i386/compile/PROTON i386
Architecture: i386
Machine: i386
>Description:
vmstat(1) shows a wrong number of system calls per interval in the first
cycle:
$ vmstat -c 3 5
procs memory page disks faults cpu
r b w avm fre flt re pi po fr sr f0 m0 w0 in sy cs us sy id
0 0 0 227212 79428 383 23 0 0 17 39 0 0 11 236 4294964578 664 28 8 64
1 0 0 227728 78912 15 0 0 0 0 0 0 0 2 334 4666 1174 10 4 86
1 0 0 227920 78720 39 0 0 0 0 0 0 0 1 209 3215 871 9 3 89
As you can see, the first value of system calls per interval is wrong. I suspect
this is due to the lack of proper initialisation of the variable in question
(maybe an int overflow?), though I haven't looked at the code, yet.
This is a desktop system up for 5 days ATM, and since I never saw this problem I'm
not going to reboot it right now so I can diagnose it further (I suspect that with
a reboot the problem goes away, so the category of this PR may be wrong - kern instead
of bin).
>How-To-Repeat:
run vmstat
>Fix: