Return to BSD News archive
Newsgroups: comp.os.386bsd.questions
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!bunyip.cc.uq.oz.au!munnari.oz.au!spool.mu.edu!howland.reston.ans.net!Germany.EU.net!EU.net!news.eunet.fi!KremlSun!satisfy.kiae.su!kiae!relcom!elvisti!store!office.elvisti.kiev.ua!stesin
From: stesin@elvisti.kiev.ua (Andrew V. Stesin)
Subject: `top' broken? (1.1.5) (was: Memory slowly disappear???)
X-Newsreader: TIN [version 1.2 PL2]
Sender: usenet@store.elvisti.kiev.ua (USENET subsystem)
Nntp-Posting-Host: office.elvisti.kiev.ua
Organization: Electronni Visti InformAgency (ElVisti)
Message-ID: <D1sEzx.Hv0@store.elvisti.kiev.ua>
References: <3e4ck1$o7o@uuneo.neosoft.com>
Date: Mon, 2 Jan 1995 16:55:09 GMT
Lines: 47
Thuy Mai (thuy@starbase.neosoft.com) wrote:
: I am running FreeBSD 1.5.???, when the machine just booted up,
: I used "top" to display the free memory (it shows ~22Mb).
: After three days, "top" shows that I have only about 8Mb left.
Running 1.1.5, I've noticed some strangeness with `top' I've
grabbed someday in summer as bin package; seems
it was top-3.1, but I'm not pretty sure.
First of all, it had a memory leak somewhere (?)
so it's groving in memory until some other processes
were being killed. Our machines are cheap (8 Mb RAM each)
so it did take no more than 3-4 days for this to happen.
Quitting top and restarting it a minute later helped.
The second thing: under heavy ( LA > 10 ) load with intensive swapping
top caused the system to fall down with page fault (page 0x0,
process id was always of top). Better to say, I think that there's
some nontrivial and hidden bug around, which may (or may not)
cause random page faults in kernel mode under a heavy load,
and the presence of the running top tickled that bug.
Maybe someone of wizards, like DG or John Dyson, will say more?
I removed those top binary a week ago and now I recognize
that I need this tool anyway; but don't know where the most
recent `top' source distribution lives (I want a full one,
not only FreeBSD port). Anyway, I'll give it
a chance after grabbing it, and I'm interested in how does it
work and how it's coded.
P.S. Don't know if it's important now, when people are mostly running
2.*, but here are some words about execve() bug in 1.1.5. Yes, thanks
to DG, Nate Williams and Terry Dwyer, there was a patch
long ago; this is patch016 and can be found at ref.tfs.com.
Terry also sent me the detailed description for all the patchkit and
affected bugs (written by Rod Ghrymes; some patches are considered good
and blessed, while some others aren't) with Terry's own comments,
that helped a lot. Thanks!
Merry Christmas and Happy New Year!
--
With best wishes -- Andrew.