Date: Thu, 8 Jun 2017 14:28:14 -0700
From: Krister Johansen <kjlx@templeofstupid.com>
To: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Krister Johansen <kjlx@templeofstupid.com>,
        linux-kernel@vger.kernel.org, mingo@kernel.org, jiangshanlai@gmail.com,
        dipankar@in.ibm.com, akpm@linux-foundation.org,
        mathieu.desnoyers@efficios.com, josh@joshtriplett.org,
        tglx@linutronix.de, peterz@infradead.org, rostedt@goodmis.org,
        dhowells@redhat.com, edumazet@google.com, fweisbec@gmail.com,
        oleg@redhat.com, bobby.prani@gmail.com, stable@vger.kernel.org,
        gregkh@linuxfoundation.org
Subject: Re: [PATCH tip/core/rcu 45/88] rcu: Add memory barriers for NOCB
 leader wakeup
Message-ID: <20170608212814.GD2553@templeofstupid.com>
References: <20170525215934.GA11578@linux.vnet.ibm.com>
 <1495749601-21574-45-git-send-email-paulmck@linux.vnet.ibm.com>
 <20170608201148.GA2553@templeofstupid.com>
 <20170608205500.GC3721@linux.vnet.ibm.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20170608205500.GC3721@linux.vnet.ibm.com>
User-Agent: Mutt/1.5.24 (2015-08-30)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1928
Lines: 41

On Thu, Jun 08, 2017 at 01:55:00PM -0700, Paul E. McKenney wrote:
> On Thu, Jun 08, 2017 at 01:11:48PM -0700, Krister Johansen wrote:
> > May I impose upon you to CC this patch to stable, and tag it as fixing
> > abedf8e241?  I ran into this on a production 4.9 branch.  When I
> > debugged it, I discovered that it went all the way back to 4.6.  The
> > tl;dr is that at least for some environments, the missed wakeup
> > manifests itself as a series of hung-task warnings to console and if I'm
> > unlucky it can also generate a hang that can block interactive logins
> > via ssh.
> 
> Interesting!  This is the first that I have heard that this was anything
> other than a theoretical bug.  To the comment in your second URL, it is
> wise to recall that a seismologist was in fact arrested for failing to
> predict an earthquake.  Later acquitted/pardoned/whatever, but arrested
> nonetheless.  ;-)

Point taken.  I do realize that we all make mistakes, and certainly I do
too.  Perhaps I should have said that my survey of current callers of
swake_up() was enough to convince me that I didn't have an immediate
problem elsewhere, but that I'm not familiar enough with the code base
to make that statement with a lot of authority.  The concern being that if
the patch came from RT-linux where the barrier was present in
swake_up(), are there other places where swake_up() callers still assume
this is being handled on their behalf?

As part of this, I also pondered whether I should add a comment around
swake_up(), similar to what's already there for waitqueue_active.
I wasn't sure how subtle this is for other consumers, though.

> Silliness aside, does my patch actually fix your problem in practice as
> well as in theory?  If so, may I have your Tested-by?

Yes, it absolutely does.  Consider it given:

Tested-by: Krister Johansen <kjlx@templeofstupid.com>

> Impressive investigative effort, by the way!

Thanks!

-K