Re: Can an SSH_MSG_EXT_INFO have zero entries?

>> But I would say that, in as far as conformance to the spec goes,
>> that's a perfectly good SSH_MSG_EXT_INFO.
> It's technically legal according to the spec, but then so is a
> message with four billion extensions (assuming the packet length
> requirement is met).

...which it cannot be, because the packet length is a 4-byte field and
each extension requires at least 8 bytes, for the length fields.
(Indeed, to list four billion _distinct_ extensions, most of them will
require a minimum of 12 bytes each.)

> However, both of those values indicate that there's something
> severely wrong with the implementation sending it,

Well, as I said upthread, an implementation is free to refuse to
communicate on whatever grounds it pleases.

But, in the nr-extensions==0 case, I disagree that it indicates
anything at all wrong (or, at least, that it necessarily indicates
anything wrong; if the implementation is not supposed to send such
packets, it would of course indicate something wrong in that particular
case).  An explicit statement that none of the relevant extensions are
supported on the connection is, IMO, a perfectly reasonable thing.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		[email protected]
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B