Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!oliveb!amiga!kodiak
From: kodiak@amiga.UUCP (Robert R. Burns)
Newsgroups: comp.sys.amiga.tech
Subject: Re: Disk Questions for WorkBench replacement project
Message-ID: <3462@amiga.UUCP>
Date: 9 Mar 89 20:32:36 GMT
References: <21173@agate.BERKELEY.EDU> <821@microsoft.UUCP> <21279@agate.BERKELEY.EDU>
Reply-To: kodiak@tooter.UUCP (Robert R. Burns)
Organization: Commodore-Amiga Inc, Los Gatos CA
Lines: 31

In article ... (Yen Yuanchi Hsieh) writes:
>c60c-1ea@web-4g.berkeley.edu (David Navas) wrote:
>>1)  How does one go about locking the correct disk if two or more
>>have the same number.

>Fortunately, I've fixed (hopefully) that problem by doing some real nasty,
>stuff with the DeviceList.  I just pull the correct volume-node to the top
>and then do my regular Lock call.  I don't even want to think what will happen
>with 1.4 ....

1.  This solution does not feel multi-tasking safe.  What if two people are
    trying this technique at the same time?
    a.  Note: all access to the DeviceList must be done within Forbid/Permit
	protection.
    b.  Lock will Wait, and thus break any Forbid.  That's when the second
	person using this technique can swizzle you.

2.  My first suggestion is to match the ambiguous volume name to the
    non-ambiguous device name by scanning the device list and matching the
    "task" fields.  This suggestion falls flat when applied to handlers that
    have no corresponding device.

3.  My robust suggestion is to recognize that the "task" fields are actually
    the process message ports for the associated file handler, and to invoke
    the lock directly using the packet interface available at this port.  I
    leave the details to the reader... :-)
-- 
Bob Burns, amiga!kodiak        "five year    .         _
| /_  _|. _ |      Commodore __   service  _/ \_      |_) _ |_  _ )'
|<(_)(_)|(_\|<      /\ |  ||| _` /\  pin" `.(5).'     |_)(_\| )(_\ |
| \ Software    ___/..\|\/|||__|/..\___    /.^.\               Faith