Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83 (MC830919); site haring.UUCP
Path: utzoo!linus!philabs!mcvax!haring!jim
From: jim@haring.UUCP (Jim McKie)
Newsgroups: net.bugs.4bsd
Subject: /etc/rmt protocol botch
Message-ID: <121@haring.UUCP>
Date: Tue, 20-Dec-83 09:05:59 EST
Article-I.D.: haring.121
Posted: Tue Dec 20 09:05:59 1983
Date-Received: Thu, 22-Dec-83 01:04:25 EST
Organization: CWI, Amsterdam
Lines: 35

The following is a bug report on a botch in the remote magtape protocol
described in RMT(8). I found this when writing a set of routines (topen,
tread, twrite, etc.) for a program to let us do remote tape-to-tape copies.

Jim McKie	    Centrum voor Wiskunde en Informatica	..mcvax!jim
---------------------------------------------------------------------------
Subject: /etc/rmt protocol botch
Index:	/usr/src/etc/rmt.c 4.2BSD

Description:
	The remote magtape protocol described in RMT(8) is not followed
	by the module "/etc/rmt" with respect to the 'S' command.

Repeat-By:
	Write a short program which creates a connection to the /etc/rmt
	on another machine then try an 'S' command - you will not get a
	valid response back (i.e. 'A') if the operation was successfull.

	/etc/rmt will send back the contents of a successfull MTIOCGET
	ioctl BEFORE it sends back the 'A' response.

Fix:
	Change the two lines in the code for the 'S' command from

		  (void) write(1, (char *)&mtget, sizeof (mtget));
		  goto respond;
	to
		  (void) sprintf(resp, "A%d\n", rval);
		  (void) write(1, resp, strlen(resp));
		  (void) write(1, (char *)&mtget, sizeof (mtget));
		  goto top;

	Another (fairly benign) bug is that all arrays declared of size
	'SSIZE' should be of size 'SSIZE+1', to deal with one of the
	terminating conditions of the 'for' loop in the function 'gets'.