[PATCH][SAMBA3] use implicit oplocks to allow write caching
jpeach at sgi.com
Thu May 4 00:09:57 GMT 2006
On Thu, 23 Mar 2006 11:43 am, James Peach wrote:
> On Wed, 2006-03-22 at 16:23 -0800, Jeremy Allison wrote:
> > On Mon, Mar 20, 2006 at 11:35:19AM +1100, James Peach wrote:
> > > Hi all,
> > >
> > > The attached patch introduces a new IMPLICIT_OPLOCK modifier to standard
> > > exclusive oplocks. The purpose of this is to allow write caching support
> > > in the presence of CIFS clients that do not use oplocks.
> > >
> > > An implicit oplock is an exclusive oplock that is taken without the
> > > knowledge of the CIFS client. This means that when the oplock is broken,
> > > the client must not be notified. Other than that, there is no change to
> > > either the oplock implementation or semantics.
> > >
> > > Any a write cache that was established while an exclusive oplock was
> > > if force is removed once the oplock is broken.
> > >
> > > Jeremy and Volker, can you please cast an eye over this?
> > Ok, don't check this in yet please. I need to look over the implications
> > of adding a new type flag into the share mode db. The last oplock logic
> > bug was due to all possible type flags not being checked, so I really
> > want to invesitgate all the places we go over the entries first.
> > Give me a few days....
> No problemo
Jeremy, do you still want me to hold off with this? It passes the most
recent Samba4 smbtorture oplock tests ...
James Peach | jpeach at sgi.com | SGI Australian Software Group
I don't speak for SGI.
More information about the samba-technical