RE: Locking and Sharing.

New Message Reply About this list Date view Thread view Subject view Author view Attachment view

From: Lance Kibblewhite (LanceKi@Attachmate.com)
Date: 07/13/98-04:24:20 PM Z


Message-ID: <E1CBD72DEEFDD011914400608CBA63D116218A@exch-mcl1.eco.twg.com>
From: Lance Kibblewhite <LanceKi@Attachmate.com>
Subject: RE: Locking and Sharing.
Date: Mon, 13 Jul 1998 14:24:20 -0700

> From: beame@mail1.tinet.ie [mailto:beame@mail1.tinet.ie]
> 
> Message-Id: <1998Jul13.213544+0100@panasonic.bws.com>
> Date: 13 Jul 1998 21:34:57 +0100
> MIME-Version: 1.0
> X-Mailer: Hummingbird 32-bit EMail Client Version 5.1.0.6
> 
> On Mon Jul 13 19:50:15 1998, Lance Kibblewhite wrote:
> > 
> > Ok, my problem is that I don't understand why they are 
> orthogonal and
> > distinct, as opposed to simply different perspectives of the same
> > semantic.
> > 
> > Can somebody please explain this.  I have seen the assertions, but a
> > description, and perhaps an example would be most useful.
> > 
> Given the following processes how can they be implemented with
> just locks:
> 
> Process 1:
> 
> 	OpenFile(Access = Read/Write, Share = Deny Write);
> 	LockBytes(File, offset = 10, length = 20);
> 	write(File, Offset = 10, Length = 20);
> 
> Process 2:
> 	OpenFile(Access = Read/Write, Share = Deny None);
> 	Open Fails because Access = Write!
> 
> Process 3:
> 	OpenFile(Access = Read, Share = Deny Read);
> 	Open Fail because Deny Read share clashes with Process 1 open
> 
> Process 4:
> 	OpenFile(Access = Read,  Share = Deny None); (succeeds)
> 	ReadBytes(File, Offset = 0, Length = 10); (succeeds)
> 	ReadBytes(File, Offset = 5, Length = 10);
> 	Read fails due to Lock of Process 1
> 	
> - Carl

If you defined 'just locks' to be the LOCKR and LOCKW type locks as in
the draft, then I agree, it would be very difficult, if not impossible.
That's why I proposed the change. And those changes would enable at
least the NFS protocol to support the above example correctly.

However, the point seems to be that with such a change this would be
difficult to implement on a Unix kernel, and that may be. So it's not
the PC perspective of this that I need, but the Unix server perspective.


New Message Reply About this list Date view Thread view Subject view Author view Attachment view

This archive was generated by hypermail 2.1.2 : 03/04/05-01:45:56 AM Z CST