[Guy Harris <gharris@xxxxxxxxx>]

> patch.1   is a one-line fix for packet-smb-pipe

I discovered that problem earlier while working on a version of the
remote API dissector ("dissect_pipe_lanman()") that'd use the descriptor
strings again, and checked in a fix (along with a fix to another
problem).
 
> patch.2  is the tvbuffified packet-smb-mailslot dissector.

Checked in, with some fixes:

	the data to hand to the protocol running atop the mailslot
	dissector starts at DataOffset - for now, I just use the tvbuff
	that starts at ParameterOffset, but, when the SMB dissector is
	tvbuffified, we should arguably construct a tvbuff solely for
	the parameters and a tvbuff solely for the data, and hand both
	of them to dissectors for protocols using pipes;

	don't construct a tvbuff for the setup area if SetupOffset is
	negative (meaning this is an interim response to a transaction).

> it also contains two small changes for packet-smb-pipe that made it
> dump core.  (in the nightly snapshot i downloaded yesterday, it might
> already have been fixed)

It was already fixed.