Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!cs.utexas.edu!uunet!mcsun!unido!fauern!immd4.informatik.uni-erlangen.de!mlelstv
From: mlelstv@immd4.informatik.uni-erlangen.de (Michael van Elst )
Newsgroups: comp.sys.amiga
Subject: Re: Mount Bug Warning (Was: Re: VD0 help)
Message-ID: <499@medusa.informatik.uni-erlangen.de>
Date: 14 Sep 89 12:23:57 GMT
References: <935@wsu-cs.uucp> <11102@boulder.Colorado.EDU> <26766@shemp.CS.UCLA.EDU> <20932@sequent.UUCP> <2493@abaa.UUCP> <838@madnix.UUCP> <19810@usc.edu>
Organization: IMMD IV, University of Erlangen, W-Germany
Lines: 27

papa@pollux.usc.edu (Marco Papa) writes:

>Quite true, Perry.  In fact, if one tried to do something like:
>Atalk3 DEVICE Serial.Device
>the device won't open, BUT a subsequent (correct) call:
>Atalk3 DEVICE serial.device
>will also fail, for the reasons Perry mentioned. At that point only a reboot
>will free the device. 

I got this bug when I wrote a hard disk driver and named the created
file different from what was mentioned in the Resident structure.
(That's not quite the same but a similar problem).

The bug is: ramlib.library of DOS or whoever checks if the name passed
to OpenDevice is entered into the device list. Now the device doesn't
see this name and installs itself with another name. The devicelist is
checked and the device requested is not found.

NOW: the device is UnLoadSeg'd from memory !!!!

the device will surely be destroyed during the next memory allocation.
I got the guru when first expunging the device resulting in #81000009
(memory freed already free).

				Michael van Elst

E-mail: UUCP: ...uunet!unido!fauern!immd4!mlelstv