On Mon, Dec 06, 2021 at 05:00:47PM -0600, Eric Blake wrote:
> On Mon, Dec 06, 2021 at 02:40:45PM +0300, Vladimir Sementsov-Ogievskiy wrote:
> > > #### Simple reply message
> > >
> > > The simple reply message MUST be sent by the server in response to all
> > > requests if structured replies have not been negotiated using
> > > -`NBD_OPT_STRUCTURED_REPLY`. If structured replies have been negotiated,
> > > a simple
> > > -reply MAY be used as a reply to any request other than `NBD_CMD_READ`,
> > > -but only if the reply has no data payload. The message looks as
> > > -follows:
> > > +`NBD_OPT_STRUCTURED_REPLY`. If structured replies have been
> > > +negotiated, a simple reply MAY be used as a reply to any request other
> > > +than `NBD_CMD_READ`, but only if the reply has no data payload. If
> > > +extended headers were not negotiated using `NBD_OPT_EXTENDED_HEADERS`,
> > > +the message looks as follows:
> > >
> > > S: 32 bits, 0x67446698, magic (`NBD_SIMPLE_REPLY_MAGIC`; used to be
> > > `NBD_REPLY_MAGIC`)
> > > @@ -369,6 +398,16 @@ S: 64 bits, handle
> > > S: (*length* bytes of data if the request is of type `NBD_CMD_READ` and
> > > *error* is zero)
> > >
> > > +If extended headers were negotiated using `NBD_OPT_EXTENDED_HEADERS`,
> > > +the message looks like:
> > > +
> > > +S: 32 bits, 0x60d12fd6, magic (`NBD_SIMPLE_REPLY_EXT_MAGIC`)
> > > +S: 32 bits, error (MAY be zero)
> > > +S: 64 bits, handle
> > > +S: 128 bits, padding (MUST be zero)
> > > +S: (*length* bytes of data if the request is of type `NBD_CMD_READ` and
> > > + *error* is zero)
> > > +
> >
> > If we go this way, let's put payload length into padding: it will help to
> > make the protocol context-independent and less error-prone.
Agreed.
> Easy enough to do (the payload length will be 0 except for
> NBD_CMD_READ).
Indeed.
> > Or, the otherway, may be just forbid the payload for simple-64bit ? What's
> > the reason to allow 64bit requests without structured reply negotiation?
>
> The two happened to be orthogonal enough in my implementation. It was
> easy to demonstrate either one without the other, and it IS easier to
> write a client using non-structured replies (structured reads ARE
> tougher than simple reads, even if it is less efficient when it comes
> to reading zeros). But you are also right that we could require
> structured reads prior to allowing 64-bit operations, and then have
> only one supported reply type on the wire when negotiated. Wouter,
> which way do you prefer?
Given that I *still* haven't gotten around to implementing structured
replies for nbd-server, I'd prefer not to require it, but that's not
really a decent argument IMO :-)
[... I haven't read this in much detail yet, intend to do that later...]
--
w@uter.{be,co.za}
wouter@{grep.be,fosdem.org,debian.org}
