badri@ur-helheim.UUCP (Badri Lokanathan) (02/19/86)
We run BSD4.2 and the 4.3(?) version of uucp. I am having the following problem with uux: I am trying to execute a command uux "remote\!who > local\!usr/spool/uucp/name" (The command "who" has been put in L.cmds for experimental purposes) The command gets executed on remote; I get mail from uucp saying that it worked. Then I go to the remote machine and find that uucico crashed while trying to copy the file over. Reading the core with adb gives bus error as the cause. The output of the command has been put in D./D.xxxx and C./C.xxxx also looks OK. I am stumped for an explanation! Incidentally I changed the version of uucp recently; the previous version had the same problem too. Everything else seems to work OK -eg uucp (both ways) and so on. Badri | ARPA : ur-helheim!badri@rochester.arpa | | UUCP : {allegra,cmcl2,decvax,harvard,seismo,topaz} | | !rochester!ur-helheim!badri |
honey@down.FUN (Peter Honeyman) (02/21/86)
regarding troubles with uux "remote!who > local!usr/spool/uucp/name" in 4.3 uucp, you probably meant uux "remote!who > local!/usr/spool/uucppublic/name" or uux "remote!who > local!~/name" for short -- local!usr/spool/uucp/name commits a multitude of sins. incidentally, this works in honey danber. peter
david@ukma.UUCP (David Herron, NPR Lover) (02/25/86)
I ran across a similar problem with the 4.2BSD uucp, but it didn't
happen in quite the same way as you described. Here's the bug report
I wrote up for it way back then. (This was almost 2 years ago)
Index: /usr/src/usr.bin/uucp/uuxqt.c 4.2BSD
Description: I was just setting up our first uucp connections and
was having trouble with a command like:
uux "ukgs\!who >ukma\!~uucp/ukgs.who"
[ukgs is a remote machine, ukma is the local machine,
ukgs is an 11/24 running 2.9BSD, ukma is a 750 running 4.2
-- David]
It was going and doing the execute but decided it could not
create the output file. Looking at the code (Ours is that
distributed from Berkeley with 4.2) I eventually found that
uuxqt was checking if it could create the output file for
any non-spool-area output file, even for files going to another
system. This doesn't make sense since uucp doesn't know
anything about its rights to create a file on another system.
Repeat by: Use the above command string. (make sure who is
executable from remote). Any command which produces output
and sent back to another system will do.
Fix: Apply the following code. Your line numbers will vary with
the mileage of your fingers and source file(s).
Notes: I would appreciate any comments on this patch. I don't know
if it creates any gaping security holes or not.
------------------------------------------------------------------
[I think the diff was run in the wrong order... believe me, I know
better now... -- David]
*** uuxqt.c Thu Aug 30 14:07:50 1984
--- uuxqt.c.orig Sat Jul 2 21:06:25 1983
***************
*** 1,24
#ifndef lint
static char sccsid[] = "@(#)uuxqt.c 5.2 (Berkeley) 7/2/83";
#endif
- /* CHANGES:
- *--- [22-Aug-84 dsh]: Problem: Output files to be routed back to another
- * system are not being routed there.
- * Diagnosis: The message being put in the log file is "who XQT DENIED".
- * The condition controlling the printing of this message is
- * "(argnok || badfiles)". The value of argnok was tested by defining
- * ALLOK which caused argok() to not test anything it was given. This
- * did not change the behavior of the program. Looking further, I
- * noticed that for X_STDOUT, chkpth() and chkperm() were being called
- * when the output file was to be sent back to a remote system.
- * Fix: Have the program check sysout to see if the file is to go
- * to another system and NOT execute chkpth() and chkperm() if so.
- */
-
-
#include "uucp.h"
#include <sys/types.h>
#include <sys/stat.h>
#ifdef NDIR
#include "ndir.h"
--- 1,9 -----
#ifndef lint
static char sccsid[] = "@(#)uuxqt.c 5.2 (Berkeley) 7/2/83";
#endif
#include "uucp.h"
#include <sys/types.h>
#include <sys/stat.h>
#ifdef NDIR
#include "ndir.h"
***************
*** 170,186
* vanilla spool file. DO check permissions
* of writing on a non-vanilla file */
i = 1;
if (fout[0] != '~' || prefix(sysout, Myname))
i = expfile(fout);
- /* CHANGE [21-Aug-84 dsh]: Problem: files going back to another system
- * don't make it.
- * Fix: Check to make sure that file is going to this system
- * before calling chkpth().
- */
- if (!prefix(sysout, Myname))
- i=0; /* sidesteps chkpth() */
if (i != 0
&& (chkpth("", "", fout)
|| chkperm(fout, (char *)1)))
badfiles = 1;
break;
--- 155,164 -----
* vanilla spool file. DO check permissions
* of writing on a non-vanilla file */
i = 1;
if (fout[0] != '~' || prefix(sysout, Myname))
i = expfile(fout);
if (i != 0
&& (chkpth("", "", fout)
|| chkperm(fout, (char *)1)))
badfiles = 1;
break;
--
David Herron, cbosgd!ukma!david, david@UKMA.BITNET, david@uky.csnet
^
Notice new and improved address---|
Postmaster for Kentucky
"'New and improved' is a misnomer" -- David Herron, 1986