Newsgroups: comp.realtime,comp.os.qnx,comp.os.ms-windows.advocacy,comp.robotics
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!europa.eng.gtefsd.com!howland.reston.ans.net!cs.utexas.edu!utnut!nott!cunews!revcan!quantum!danh
From: danh@qnx.com (Dan Hildebrand)
Subject: Re: Real-time systems: Windows-NT or QNX
Message-ID: <apz0vb+@qnx.com>
Date: Mon, 17 Oct 94 17:02:01 GMT
Organization: QNX Software Systems
References: <FriOct14102309EST1994@eric> <37kql1$e4d@vixen.cso.uiuc.edu> <37ltcm$ged@sol.sun.csd.unb.ca> <1994Oct14.182316.28294@mprgate.mpr.ca>
Lines: 23
Xref: glinda.oz.cs.cmu.edu comp.realtime:7194 comp.os.qnx:2250 comp.os.ms-windows.advocacy:40668 comp.robotics:14462

In article <1994Oct14.182316.28294@mprgate.mpr.ca>,
John Altstadt <altstadt@mprgate.mpr.ca> wrote:
>In article <37ltcm$ged@sol.sun.csd.unb.ca>, newton@jupiter.sun.csd.unb.ca (Chris Newton CSD) writes:
>>    QNX on the other hand is solely 
>>built for real-time stuff as I understand it.
>
>I have seen some particularly horrible things happen with QNX V 3.xx
>where it decided to run an idle-priority, compute-bound task to the
>exclusion of the high-priority control tasks.  There were other bugs and
>mis-features, but this was the most spectacular since it occurred on the
>customer's site.  Hopefully V 4.xx is much better.

It turns out that this was the result of some floppy disk I/O being 
manually initiated from the console while the realtime process was running.  
I hate to criticize a previous product version, but that version of QNX 
(the pre-POSIX version, 4 years old now) had a single threaded filesystem, 
which had the problem of slowing hard disk I/O while floppy disk I/O was in 
progress.  QNX 4.xx has a multithreaded filesystem, avoiding this issue 
entirely.
-- 
Dan Hildebrand      danh@qnx.com         QNX Software Systems, Ltd.   
phone: (613) 591-0931 x204 (voice)       175 Terence Matthews          
       (613) 591-3579      (fax)         Kanata, Ontario, Canada K2M 1W8
