Mailing-List: contact cygwin-help@sourceware.cygnus.com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner@sources.redhat.com Delivered-To: mailing list cygwin@sources.redhat.com Message-Id: <4.2.2.20010822180506.00ba33f0@mail.online.no> X-Sender: hardon@mail.online.no X-Mailer: QUALCOMM Windows Eudora Pro Version 4.2.2 Date: Wed, 22 Aug 2001 20:07:28 +0200 To: cygwin@cygwin.com From: Gunnar Andre Dalsnes Subject: Re: Help on posix file lock semantics Cc: Kurt Roeckx In-Reply-To: <20010822021939.A10144@ping.be> References: <4.2.2.20010822010059.00ba5600@mail.online.no> <4.2.2.20010821153100.00ba9ba0@mail.online.no> <4.2.2.20010820221811.00b9d9a0@mail.online.no> <4.2.2.20010819225911.00b94970@pop.online.no> <4.2.2.20010819225911.00b94970@pop.online.no> <20010820131824.A200@ping.be> <4.2.2.20010820221811.00b9d9a0@mail.online.no> <20010821121135.A7837@ping.be> <4.2.2.20010821153100.00ba9ba0@mail.online.no> <20010822003718.A9717@ping.be> <4.2.2.20010822010059.00ba5600@mail.online.no> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" At 22.08.01 02:19 , you wrote: >On Wed, Aug 22, 2001 at 01:31:39AM +0200, Gunnar Andre Dalsnes wrote: > > At 22.08.01 00:37 , you wrote: > > > > > >Those 2 regions follow each other. I think you're not supposed > > >to merge those regions even if you could. > > > > > >If you don't merge regions in any other case, why would you > > >suddenly want to merge them in the case of overlapping regions? > > > > But if the (linux) os kernel does it automagically for us, I want to do it too:-) > >The comment for posix_lock_file() in fs/locks.s says: >"We merge adjacent locks whenever possible.", so I guess it >always does. To sort this out ofa, I went out and got a copy of Suse, and this was what I got: -if any separate adjacent locks with same type were set, they merged. (regardless of type) -if a new lock overlapped existing lock with same type, they merged. (regardless of type) -if a new read lock overlapped existing write lock, the overlapped region changed to type read. -if a new write lock overlapped existing read lock, the overlapped region changed to type write. Not very unlike our assumptions:-) >Kurt Now I have enough info on this, to continue on the hard part, the coding... Thank you very much for your time, Kurt:-) Gunnar -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/