From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1753796AbcB2O1r (ORCPT <rfc822;w@1wt.eu>);
	Mon, 29 Feb 2016 09:27:47 -0500
Received: from mail-wm0-f49.google.com ([74.125.82.49]:37621 "EHLO
	mail-wm0-f49.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751253AbcB2O1o (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Mon, 29 Feb 2016 09:27:44 -0500
Message-ID: <1456756060.3488.120.camel@gmail.com>
Subject: Re: [ANNOUNCE] v4.4.3-rt9
From: Mike Galbraith <umgwanakikbuti@gmail.com>
To: Thomas Gleixner <tglx@linutronix.de>, LKML <linux-kernel@vger.kernel.org>
Cc: linux-rt-users <linux-rt-users@vger.kernel.org>,
        Sebastian Sewior <bigeasy@linutronix.de>,
        Steven Rostedt <rostedt@goodmis.org>
Date: Mon, 29 Feb 2016 15:27:40 +0100
In-Reply-To: <alpine.DEB.2.11.1602291212180.3638@nanos>
References: <alpine.DEB.2.11.1602291212180.3638@nanos>
Content-Type: text/plain; charset="UTF-8"
X-Mailer: Evolution 3.16.5 
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Mon, 2016-02-29 at 13:46 +0100, Thomas Gleixner wrote:
> Dear RT folks!
> 
> I'm pleased to announce the v4.4.3-rt9 patch set. v4.4.2-rt7 and v4.4.3-rt8
> are non-announced updates to incorporate the linux-4.4.y stable tree.
> 
> There is one change caused by the 4.4.3 update:
> 
>   The relaxed handling of dump_stack() on RT has been dropped as there is
>   actually a potential deadlock lurking around the corner. See: commit
>   d7ce36924344 upstream. This does not effect the other facilities which
>   gather stack traces.

Hrm.  I had rolled that dropped bit forward as below.  I was given
cause to do a very large pile of ltp oom4 testing (rt kernels will
livelock due to waitqueue workers waiting for kthreadd to get memory to
spawn a kworker thread, while stuck kworker holds manager mutex, unless
workers are run as rt tasks to keep us from getting that depleted in
the first place), which gives it oodles of exercise, and all _seemed_
well.  Only seemed?

--- a/lib/dump_stack.c	2016-02-29 14:20:29.512510444 +0100
+++ b/lib/dump_stack.c	2016-02-26 13:03:15.755297038 +0100
@@ -8,6 +8,7 @@
 #include <linux/sched.h>
 #include <linux/smp.h>
 #include <linux/atomic.h>
+#include <linux/locallock.h>
 
 static void __dump_stack(void)
 {
@@ -22,6 +23,7 @@ static void __dump_stack(void)
  */
 #ifdef CONFIG_SMP
 static atomic_t dump_lock = ATOMIC_INIT(-1);
+static DEFINE_LOCAL_IRQ_LOCK(dump_stack_irq_lock);
 
 asmlinkage __visible void dump_stack(void)
 {
@@ -35,7 +37,7 @@ asmlinkage __visible void dump_stack(voi
 	 * against other CPUs
 	 */
 retry:
-	local_irq_save(flags);
+	local_lock_irqsave(dump_stack_irq_lock, flags);
 	cpu = smp_processor_id();
 	old = atomic_cmpxchg(&dump_lock, -1, cpu);
 	if (old == -1) {
@@ -43,7 +45,7 @@ retry:
 	} else if (old == cpu) {
 		was_locked = 1;
 	} else {
-		local_irq_restore(flags);
+		local_unlock_irqrestore(dump_stack_irq_lock, flags);
 		cpu_relax();
 		goto retry;
 	}
@@ -53,7 +55,7 @@ retry:
 	if (!was_locked)
 		atomic_set(&dump_lock, -1);
 
-	local_irq_restore(flags);
+	local_unlock_irqrestore(dump_stack_irq_lock, flags);
 }
 #else
 asmlinkage __visible void dump_stack(void)