[PATCH 2/2] ptrace: ptrace_detach() can no longer race with SIGKILL

[Oleg Nesterov <oleg@redhat.com>]

ptrace_detach() re-checks ->ptrace under tasklist lock and calls
release_task() if __ptrace_detach() returns true. This was needed
because the __TASK_TRACED tracee could be killed/untraced, and it
could even pass exit_notify() before we take tasklist_lock.

But this is no longer possible after 9899d11f6544 "ptrace: ensure
arch_ptrace/ptrace_request can never race with SIGKILL". We can turn
these checks into WARN_ON() and remove release_task().

While at it, document the setting of child->exit_code.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 kernel/ptrace.c |   19 +++++++++----------
 1 files changed, 9 insertions(+), 10 deletions(-)

diff --git a/kernel/ptrace.c b/kernel/ptrace.c
index 5009263..23be9dd 100644
--- a/kernel/ptrace.c
+++ b/kernel/ptrace.c
@@ -456,8 +456,6 @@ static bool __ptrace_detach(struct task_struct *tracer, struct task_struct *p)
 
 static int ptrace_detach(struct task_struct *child, unsigned int data)
 {
-	bool dead = false;
-
 	if (!valid_signal(data))
 		return -EIO;
 
@@ -467,18 +465,19 @@ static int ptrace_detach(struct task_struct *child, unsigned int data)
 
 	write_lock_irq(&tasklist_lock);
 	/*
-	 * This child can be already killed. Make sure de_thread() or
-	 * our sub-thread doing do_wait() didn't do release_task() yet.
+	 * We rely on ptrace_freeze_traced(). It can't be killed and
+	 * untraced by another thread, it can't be a zombie.
 	 */
-	if (child->ptrace) {
-		child->exit_code = data;
-		dead = __ptrace_detach(current, child);
-	}
+	WARN_ON(!child->ptrace || child->exit_state);
+	/*
+	 * tasklist_lock avoids the race with wait_task_stopped(), see
+	 * the comment in ptrace_resume().
+	 */
+	child->exit_code = data;
+	__ptrace_detach(current, child);
 	write_unlock_irq(&tasklist_lock);
 
 	proc_ptrace_connector(child, PTRACE_DETACH);
-	if (unlikely(dead))
-		release_task(child);
 
 	return 0;
 }
-- 
1.5.5.1