do we need LOCK_CONFIRM?

From: Kendrick M. Smith (kmsmith@umich.edu)
Date: 10/17/01-05:48:18 PM Z

• Next message: Peter Åstrand: "ACCESS4_EXECUTE on directories"
• Previous message: Noveck, Dave: "RE: another question on the new .x"

Date: Wed, 17 Oct 2001 18:48:18 -0400 (EDT)
From: "Kendrick M. Smith" <kmsmith@umich.edu>
Subject: do we need LOCK_CONFIRM?
Message-ID: <Pine.SOL.4.33.0110171843470.9600-100000@tempest.gpcc.itd.umich.edu>


I don't see why we need a LOCK_CONFIRM operation to confirm the
instantation of a new lock_owner with LOCK.  I do understand why
OPEN_CONFIRM is necessary: if the OPEN uses a lockowner for the
first time, the server has no way of knowing whether the OPEN
is a replay or not, hence OPEN_CONFIRM.  However, the LOCK
operation always supplies the server with an up-to-date seqid
for an existing lockowner (and an up-to-date stateid), which
is enough to ensure that the LOCK is not a replay.

If I am missing some reason why LOCK_CONFIRM is needed, can
someone supply an explicit example showing how things could
go wrong without it?

Thanks,
Kendrick

• Next message: Peter Åstrand: "ACCESS4_EXECUTE on directories"
• Previous message: Noveck, Dave: "RE: another question on the new .x"

This archive was generated by hypermail 2.1.2 : 03/04/05-01:49:17 AM Z CST