[PATCH] coredump: kill ptrace related stuff

With this patch zap_process() sets SIGNAL_GROUP_EXIT while sending SIGKILL to
the thread group.  This means that a TASK_TRACED task

	1. Will be awakened by signal_wake_up(1)

	2. Can't sleep again via ptrace_notify()

	3. Can't go to do_signal_stop() after return
	   from ptrace_stop() in get_signal_to_deliver()

So we can remove all ptrace related stuff from coredump path.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

fs/exec.c

@@ -1368,12 +1368,14 @@ static void format_corename(char *corename, const char *pattern, long signr)
 	*out_ptr = 0;
 }
 
-static void zap_process(struct task_struct *start, int *ptraced)
+static void zap_process(struct task_struct *start)
 {
 	struct task_struct *t;
 	unsigned long flags;
 
 	spin_lock_irqsave(&start->sighand->siglock, flags);
+	start->signal->flags = SIGNAL_GROUP_EXIT;
+	start->signal->group_stop_count = 0;
 
 	t = start;
 	do {
@@ -1381,22 +1383,17 @@ static void zap_process(struct task_struct *start, int *ptraced)
 			t->mm->core_waiters++;
 			sigaddset(&t->pending.signal, SIGKILL);
 			signal_wake_up(t, 1);
-
-			if (unlikely(t->ptrace) &&
-			    unlikely(t->parent->mm == t->mm))
-				*ptraced = 1;
 		}
 	} while ((t = next_thread(t)) != start);
 
 	spin_unlock_irqrestore(&start->sighand->siglock, flags);
 }
 
-static void zap_threads (struct mm_struct *mm)
+static void zap_threads(struct mm_struct *mm)
 {
 	struct task_struct *g, *p;
 	struct task_struct *tsk = current;
 	struct completion *vfork_done = tsk->vfork_done;
-	int traced = 0;
 
 	/*
 	 * Make sure nobody is waiting for us to release the VM,
@@ -1413,29 +1410,12 @@ static void zap_threads (struct mm_struct *mm)
 		do {
 			if (p->mm) {
 				if (p->mm == mm)
-					zap_process(p, &traced);
+					zap_process(p);
 				break;
 			}
 		} while ((p = next_thread(p)) != g);
 	}
 	read_unlock(&tasklist_lock);
-
-	if (unlikely(traced)) {
-		/*
-		 * We are zapping a thread and the thread it ptraces.
-		 * If the tracee went into a ptrace stop for exit tracing,
-		 * we could deadlock since the tracer is waiting for this
-		 * coredump to finish.  Detach them so they can both die.
-		 */
-		write_lock_irq(&tasklist_lock);
-		do_each_thread(g,p) {
-			if (mm == p->mm && p != tsk &&
-			    p->ptrace && p->parent->mm == mm) {
-				__ptrace_detach(p, 0);
-			}
-		} while_each_thread(g,p);
-		write_unlock_irq(&tasklist_lock);
-	}
 }
 
 static void coredump_wait(struct mm_struct *mm)

include/linux/ptrace.h

@@ -88,7 +88,6 @@ extern int ptrace_readdata(struct task_struct *tsk, unsigned long src, char __us
 extern int ptrace_writedata(struct task_struct *tsk, char __user *src, unsigned long dst, int len);
 extern int ptrace_attach(struct task_struct *tsk);
 extern int ptrace_detach(struct task_struct *, unsigned int);
-extern void __ptrace_detach(struct task_struct *, unsigned int);
 extern void ptrace_disable(struct task_struct *);
 extern int ptrace_check_attach(struct task_struct *task, int kill);
 extern int ptrace_request(struct task_struct *child, long request, long addr, long data);

kernel/ptrace.c

@@ -214,7 +214,7 @@ int ptrace_attach(struct task_struct *task)
 	return retval;
 }
 
-void __ptrace_detach(struct task_struct *child, unsigned int data)
+static inline void __ptrace_detach(struct task_struct *child, unsigned int data)
 {
 	child->exit_code = data;
 	/* .. re-parent .. */
@@ -233,6 +233,7 @@ int ptrace_detach(struct task_struct *child, unsigned int data)
 	ptrace_disable(child);
 
 	write_lock_irq(&tasklist_lock);
+	/* protect against de_thread()->release_task() */
 	if (child->ptrace)
 		__ptrace_detach(child, data);
 	write_unlock_irq(&tasklist_lock);

kernel/signal.c

@@ -1531,6 +1531,35 @@ static void do_notify_parent_cldstop(struct task_struct *tsk, int why)
 	spin_unlock_irqrestore(&sighand->siglock, flags);
 }
 
+static inline int may_ptrace_stop(void)
+{
+	if (!likely(current->ptrace & PT_PTRACED))
+		return 0;
+
+	if (unlikely(current->parent == current->real_parent &&
+		    (current->ptrace & PT_ATTACHED)))
+		return 0;
+
+	if (unlikely(current->signal == current->parent->signal) &&
+	    unlikely(current->signal->flags & SIGNAL_GROUP_EXIT))
+		return 0;
+
+	/*
+	 * Are we in the middle of do_coredump?
+	 * If so and our tracer is also part of the coredump stopping
+	 * is a deadlock situation, and pointless because our tracer
+	 * is dead so don't allow us to stop.
+	 * If SIGKILL was already sent before the caller unlocked
+	 * ->siglock we must see ->core_waiters != 0. Otherwise it
+	 * is safe to enter schedule().
+	 */
+	if (unlikely(current->mm->core_waiters) &&
+	    unlikely(current->mm == current->parent->mm))
+		return 0;
+
+	return 1;
+}
+
 /*
  * This must be called with current->sighand->siglock held.
  *
@@ -1559,11 +1588,7 @@ static void ptrace_stop(int exit_code, int nostop_code, siginfo_t *info)
 	spin_unlock_irq(&current->sighand->siglock);
 	try_to_freeze();
 	read_lock(&tasklist_lock);
-	if (likely(current->ptrace & PT_PTRACED) &&
-	    likely(current->parent != current->real_parent ||
-		   !(current->ptrace & PT_ATTACHED)) &&
-	    (likely(current->parent->signal != current->signal) ||
-	     !unlikely(current->signal->flags & SIGNAL_GROUP_EXIT))) {
+	if (may_ptrace_stop()) {
 		do_notify_parent_cldstop(current, CLD_TRAPPED);
 		read_unlock(&tasklist_lock);
 		schedule();