Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   "Rafael J. Wysocki" <rjw@rjwysocki.net>
To:     Michal Hocko <mhocko@kernel.org>
Cc:     Ingo Molnar <mingo@kernel.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Linux List Kernel Mailing <linux-kernel@vger.kernel.org>,
        "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>,
        Chanho Min <chanho.min@lge.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Peter Zijlstra <a.p.zijlstra@chello.nl>,
        Oleg Nesterov <oleg@redhat.com>, Pavel Machek <pavel@ucw.cz>
Subject: Re: [PATCH] Revert "exec: make de_thread() freezable (was: Re: Linux 4.20-rc4)
Date:   Mon, 03 Dec 2018 13:03:16 +0100
Message-ID: <1718869.PGhsHE7EHz@aspire.rjw.lan>
In-Reply-To: <20181203083942.GF31738@dhcp22.suse.cz>
References: <CAHk-=wgdsXyCaLsFEpyUpAeRqVS69u=xo4rzEN+cS=xwz2gajg@mail.gmail.com> <20181203074700.GA21240@gmail.com> <20181203083942.GF31738@dhcp22.suse.cz>
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="us-ascii"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Monday, December 3, 2018 9:39:42 AM CET Michal Hocko wrote:
> On Mon 03-12-18 08:47:00, Ingo Molnar wrote:
> [...]
> > I reviewed the ->cred_guard_mutex code, and the mutex is held across all 
> > of exec() - and we always did this.
> 
> Yes, this is something that has been pointed out during the review. Oleg
> has argued that making this path freezable is really hard and that we
> should be changing de_thread to sleep withtou cred_guard_mutex long term
> anyway (http://lkml.kernel.org/r/20181114143705.GB13885@redhat.com).
> 
> Failing suspend seems like a real problem while the lockdep one doesn't
> really reflect any real deadlock, right? So while the patch is not
> perfect it shouldn't make the situation much worse. Lockdep splat is
> certainly annoying but is it any worse than a suspend failing?
> 
> Now, I wouldn't mind to revert this because the code is really old and
> we haven't seen many bug reports about failing suspend yet. But what is
> the actual plan to make this work properly? Use
> freezable_schedule_unsafe instead? Freezer code has some
> fundamental design issues which are quite hard to get over.

I agree and we just need to look deeper into this.

I had hoped that this would work since you and Oleg agreed with it, but since
it doesn't, let's do a revert for now and get back to this later.

Thanks,
Rafael