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;
MIME-Version: 1.0
References: <157558502272.10278.8718685637610645781.stgit@warthog.procyon.org.uk>
 <20191206135604.GB2734@twin.jikos.cz> <CAHk-=wiN_pWbcRaw5L-J2EFUyCn49Due0McwETKwmFFPp88K8Q@mail.gmail.com>
 <CAHk-=wjvO1V912ya=1rdXwrm1OBTi6GqnqryH_E8OR69cZuVOg@mail.gmail.com>
 <CAHk-=wizsHmCwUAyQKdU7hBPXHYQn-fOtJKBqMs-79br2pWxeQ@mail.gmail.com>
 <CAHk-=wjeG0q1vgzu4iJhW5juPkTsjTYmiqiMUYAebWW+0bam6w@mail.gmail.com> <9417.1576097731@warthog.procyon.org.uk>
In-Reply-To: <9417.1576097731@warthog.procyon.org.uk>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Wed, 11 Dec 2019 17:28:40 -0800
Message-ID: <CAHk-=wgVpV-O4zeO+xOMcOUyKZi+1kg+6Kmi-P7SUx-ydZm5Og@mail.gmail.com>
Subject: Re: [PATCH 0/2] pipe: Fixes [ver #2]
To:     David Howells <dhowells@redhat.com>
Cc:     David Sterba <dsterba@suse.cz>, Eric Biggers <ebiggers@kernel.org>,
        Al Viro <viro@zeniv.linux.org.uk>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@kernel.org>,
        Vincent Guittot <vincent.guittot@linaro.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Dec 11, 2019 at 12:55 PM David Howells <dhowells@redhat.com> wrote:
>
> Is it worth reverting:
>
>         commit f94df9890e98f2090c6a8d70c795134863b70201
>         Add wake_up_interruptible_sync_poll_locked()
>
> since you changed the code that was calling that new function and so it's no
> longer called?

You are sure you won't want that for the notification queue cases? I
guess they'll never want to "sync" part..

Looking at the regular pipe read/write code, maybe we'll want to try
it again - do the wakeup while we already have the spinlock, rather
than later. But I have this suspicion that that might just then push
things into mutex contention, so who knows..

Regardless, it's not going to happen for 5.5, so I guess we could
revert it and if we ever end up trying it again we can resurrect it.

              Linus