ptrace: fix PTRACE_LISTEN race corrupting task->state
authorbsegall@google.com <bsegall@google.com>
Fri, 7 Apr 2017 23:04:51 +0000 (16:04 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 8 Apr 2017 07:47:48 +0000 (00:47 -0700)
In PT_SEIZED + LISTEN mode STOP/CONT signals cause a wakeup against
__TASK_TRACED.  If this races with the ptrace_unfreeze_traced at the end
of a PTRACE_LISTEN, this can wake the task /after/ the check against
__TASK_TRACED, but before the reset of state to TASK_TRACED.  This
causes it to instead clobber TASK_WAKING, allowing a subsequent wakeup
against TRACED while the task is still on the rq wake_list, corrupting
it.

Oleg said:
 "The kernel can crash or this can lead to other hard-to-debug problems.
  In short, "task->state = TASK_TRACED" in ptrace_unfreeze_traced()
  assumes that nobody else can wake it up, but PTRACE_LISTEN breaks the
  contract. Obviusly it is very wrong to manipulate task->state if this
  task is already running, or WAKING, or it sleeps again"

[akpm@linux-foundation.org: coding-style fixes]
Fixes: 9899d11f ("ptrace: ensure arch_ptrace/ptrace_request can never race with SIGKILL")
Link: http://lkml.kernel.org/r/xm26y3vfhmkp.fsf_-_@bsegall-linux.mtv.corp.google.com
Signed-off-by: Ben Segall <bsegall@google.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/ptrace.c

index 0af9287..266ddcc 100644 (file)
@@ -184,11 +184,17 @@ static void ptrace_unfreeze_traced(struct task_struct *task)
 
        WARN_ON(!task->ptrace || task->parent != current);
 
+       /*
+        * PTRACE_LISTEN can allow ptrace_trap_notify to wake us up remotely.
+        * Recheck state under the lock to close this race.
+        */
        spin_lock_irq(&task->sighand->siglock);
-       if (__fatal_signal_pending(task))
-               wake_up_state(task, __TASK_TRACED);
-       else
-               task->state = TASK_TRACED;
+       if (task->state == __TASK_TRACED) {
+               if (__fatal_signal_pending(task))
+                       wake_up_state(task, __TASK_TRACED);
+               else
+                       task->state = TASK_TRACED;
+       }
        spin_unlock_irq(&task->sighand->siglock);
 }