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;
Message-ID: <1535718146.23560.80.camel@arista.com>
Subject: Re: [PATCH 3/4] tty: Lock tty pair in tty_init_dev()
From:   Dmitry Safonov <dima@arista.com>
To:     Jiri Slaby <jslaby@suse.cz>, linux-kernel@vger.kernel.org
Cc:     Daniel Axtens <dja@axtens.net>,
        Dmitry Safonov <0x7f454c46@gmail.com>,
        Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>,
        Dmitry Vyukov <dvyukov@google.com>,
        Tan Xiaojun <tanxiaojun@huawei.com>,
        Peter Hurley <peter@hurleysoftware.com>,
        Pasi =?ISO-8859-1?Q?K=E4rkk=E4inen?= <pasik@iki.fi>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Michael Neuling <mikey@neuling.org>,
        Mikulas Patocka <mpatocka@redhat.com>
Date:   Fri, 31 Aug 2018 13:22:26 +0100
In-Reply-To: <125964f4-ce26-ce43-b788-d3dbaef139b4@suse.cz>
References: <20180829022353.23568-1-dima@arista.com>
         <20180829022353.23568-4-dima@arista.com>
         <22935dc9-a9fe-5fc8-4c97-3f5c7b07b8c3@suse.cz>
         <1535560111.23560.61.camel@arista.com>
         <125964f4-ce26-ce43-b788-d3dbaef139b4@suse.cz>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, 2018-08-31 at 08:54 +0200, Jiri Slaby wrote:
> On 08/29/2018, 06:28 PM, Dmitry Safonov wrote:
> > On Wed, 2018-08-29 at 16:46 +0200, Jiri Slaby wrote:
> > > On 08/29/2018, 04:23 AM, Dmitry Safonov wrote:
> > > > It's safe to not lock both here - done to silence attempt
> > > > lockdep
> > > > assert in
> > > > tty_ldisc_open(), which will be added with following patch.
> > > 
> > > SOrry, could you elaborate here? I don't follow...
> > 
> > Sure, 4/4 patch adds lockdep_assert_held() into tty_ldisc_open().
> > Currently ldisc in tty->link isn't locked, which according to code
> > shouldn't be an issue, as far as I can see.
> > 
> > So, this patch silences lockdep warining by holding the semaphore,
> > which is slowpath anyway and doesn't case any new contention.
> 
> Eh, no... Adding a lock just to make lockdep happy is a no-go. The
> locking in tty is already complex enough.
> 
> > (actually, not holding the semaphore for slave might be an issue if
> > one
> > opens slave before it's fully initialized, but I'm not sure if it's
> > possible).
> 
> If that turns out to be a problem, we can apply the patch. BUt unless
> it
> is proven to be so (be it code review or a reproducer), let's leave
> the
> locking as it is.

Ok, than for v2 I'll just remove lockdep_assert() from
tty_ldisc_open().

-- 
Thanks,
             Dmitry