*BSD News Article 26309


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!munnari.oz.au!bunyip.cc.uq.oz.au!harbinger.cc.monash.edu.au!aggedor.rmit.EDU.AU!goanna.cs.rmit.oz.au!numbat!s902150
From: s902150@numbat.cs.rmit.OZ.AU (Brian Havard)
Newsgroups: comp.os.386bsd.misc
Subject: xv3.00a on a 386-40 problem NetBSD 0.9 (solved)
Date: 20 Jan 94 13:00:30 GMT
Organization: Comp Sci, RMIT, Melbourne, Australia
Lines: 18
Message-ID: <s902150.759070830@numbat>
NNTP-Posting-Host: numbat.cs.rmit.oz.au
NNTP-Posting-User: s902150
Summary: Problem running xv on a 386-40 (no co-pro) with NetBSD 0.9 solved
Keywords: xv bug floor fixed

  You may remember a while back that some people reported problems running 
xv 3.00a on a 386-40 with no maths copro. The problem was that it took an
increadibly long time to start up and (in my case at least) got the colours
all wrong when it did get around to displaying the picture.
  Well I decided I was going to the botton of the problem so I compiled with
debug info, ran it from gdb and interruped it a little while into its looong
startup. The stack frame was very interesting. It contained several hundred
entries that when floor, ceil, floor, ceil, floor.....etc. The chain of calls
started with a call to floor so I wrote my own floor and linked it in. It
now works perfectly. 
  Is there a known bug in floor? Is it actually (as someone suggested) a bug
in the floating point emulator? It just seemed a bit silly to me.......

-- 
_______________________________________________________________________________
|  Brian Havard                 |  "He is not the messiah!                    |
|  s902150@numbat.cs.rmit.oz.au |  He's a very naughty boy!" - Life of Brian  |
-------------------------------------------------------------------------------