Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200721063258.17140-1-mhocko@kernel.org> <CAHk-=whewL14RgwLZTXcNAnrDPt0H+sRJS6iDq0oGb6zwaBMxg@mail.gmail.com>
 <CAHk-=whb0=rjc1WR+F_r_syw5Ld4=ebuNJmmpaPEzfjZRD5Y-w@mail.gmail.com>
 <alpine.LSU.2.11.2007221359450.1017@eggly.anvils> <CAHk-=wi=vuc6sdu0m9nYd3gb8x5Xgnc6=TH=DTOy7qU96rZ9nw@mail.gmail.com>
 <CAHk-=whEjnsANEhTA3aqpNLZ3vv7huP7QAmcAEd-GUxm2YMo-Q@mail.gmail.com>
 <20200723124749.GA7428@redhat.com> <CAHk-=wgyc7en4=HddEYiz_RKJXfqe1JYv3BzHc=+_wYq9ti+LQ@mail.gmail.com>
In-Reply-To: <CAHk-=wgyc7en4=HddEYiz_RKJXfqe1JYv3BzHc=+_wYq9ti+LQ@mail.gmail.com>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Thu, 23 Jul 2020 13:03:28 -0700
Message-ID: <CAHk-=whQK3OGwExTzCrwwvuuVaQAgs8KsR-Yv8m1BmXoNZZ=jQ@mail.gmail.com>
Subject: Re: [RFC PATCH] mm: silence soft lockups from unlock_page
To:     Oleg Nesterov <oleg@redhat.com>
Cc:     Hugh Dickins <hughd@google.com>, Michal Hocko <mhocko@kernel.org>,
        Linux-MM <linux-mm@kvack.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Tim Chen <tim.c.chen@linux.intel.com>,
        Michal Hocko <mhocko@suse.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Jul 23, 2020 at 10:32 AM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> So here's a v2, now as a "real" commit with a commit message and everything.

Oh, except it's broken.

Switching from the "am I still on the list" logic to the
"WQ_FLAG_WOKEN is set if woken" logic was all well and good, but I
missed the case where we did that trylock_page_bit_common().

It used to just not add the thing to the list if it would get the page
bit, and then the rest of the waiting logic looked at that and was
happy.

But now if it needs to actually fake that WQ_FLAG_WOKEN flag.

So that patch needs to do something like this:

        if (!trylock_page_bit_common(page, bit_nr, behavior))
                __add_wait_queue_entry_tail(q, wait);
        else
                wait->flags |= WQ_FLAG_WOKEN;

in there. Or maybe have that bit set originally, and clear it when we
add to the wait queue.

I'll send a new version after I actually test it.

                Linus