Re: [syzbot] WARNING: locking bug in umh_complete

[Peter Zijlstra <peterz@infradead.org>]

On Fri, Feb 03, 2023 at 09:59:51PM +0900, Tetsuo Handa wrote:
> On 2023/02/03 21:30, Peter Zijlstra wrote:
> >> I think the right fix is to:
> >>
> >> 	state &= ~TASK_KILLABLE;
> > 
> > 	state &= ~__TASK_WAKEKILL;
> > 
> > we don't want to mask out UNINTERUPTIBLE, that would be bad.
> 
> This code was made killable as a solution for CVE-2012-4398.
> Although OOM reaper is available today, making back to unkillable is not smart.

Yes, I meant something like so.


diff --git a/kernel/umh.c b/kernel/umh.c
index 850631518665..0e69e1e4b6fe 100644
--- a/kernel/umh.c
+++ b/kernel/umh.c
@@ -438,21 +438,24 @@ int call_usermodehelper_exec(struct subprocess_info *sub_info, int wait)
 	if (wait == UMH_NO_WAIT)	/* task has freed sub_info */
 		goto unlock;
 
-	if (wait & UMH_KILLABLE)
-		state |= TASK_KILLABLE;
-
-	if (wait & UMH_FREEZABLE)
+	if (wait & UMH_FREEZABLE) {
 		state |= TASK_FREEZABLE;
 
-	retval = wait_for_completion_state(&done, state);
-	if (!retval)
-		goto wait_done;
-
 	if (wait & UMH_KILLABLE) {
+		retval = wait_for_completion_state(&done, state | TASK_KILLABLE);
+		if (!retval)
+			goto wait_done;
+
 		/* umh_complete() will see NULL and free sub_info */
 		if (xchg(&sub_info->complete, NULL))
 			goto unlock;
+
+		/*
+		 * fallthrough; in case of -ERESTARTSYS now do uninterruptible
+		 * wait_for_completion().
+		 */
 	}
+	wait_for_completion_state(&done, state);
 
 wait_done:
 	retval = sub_info->retval;