<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>
Date   : Thu, 28 Feb 2013 21:15:08 +0000
From   : jgh@... (J.G.Harston)
Subject: Odd bug in Watford DFS

When I worked at W(cough!) E(splutter!)  a customer phoned complaining 
about the Watford Electronics DFS he had recently bought - it would 
corrupt files occasionally, and in the applications he was using the 
occassionally happened with a high frequency. He explained the scenario 
and I was able to replicate it exactly, so offered to replace it with an 
Acorn DFS.

I've been trying to remember the events so I can document it.

I believe it must have been an 8271 DFS, as otherwise I couldn't have 
replaced it with the Acorn DFS.

 From memory, what happened was something along the lines of doing some 
sequence of disk operations all before the disk had stopped spinning, 
and something got written back incorrectly:

Has anybody encountered this?

DFS 0.90 had a similar problem, something like:
- Open an output or update file on one surface (eg 
ch1%=OPENUP":0.file")
- Open an output or update file on the other surface of the same drive 
(eg ch2%=OPENUP":2.file")
- Before the disk stops spinning:
-- Write some data to channel 1 (eg BPUT#ch1,0)
-- Close channel 1

The DFS sees the disk still spinning and as surface 2 was the last 
surface accessed, writes the current catalog to it - ie surface 1's 
catalog!

-- 
J.G.Harston - jgh@...      - mdfs.net
<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>