Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 5 Jul 2022 16:47:10 -0700
From:   Andrew Morton <akpm@linux-foundation.org>
To:     Matthew Wilcox <willy@infradead.org>
Cc:     Muchun Song <songmuchun@bytedance.com>, jgg@ziepe.ca,
        jhubbard@nvidia.com, william.kucharski@oracle.com,
        dan.j.williams@intel.com, jack@suse.cz, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        nvdimm@lists.linux.dev, stable@vger.kernel.org
Subject: Re: [PATCH v2] mm: fix missing wake-up event for FSDAX pages
Message-Id: <20220705164710.9541b5cf0e5819193213ea5c@linux-foundation.org>
In-Reply-To: <YsTLgQ45ESpsNEGV@casper.infradead.org>
References: <20220705123532.283-1-songmuchun@bytedance.com>
        <20220705141819.804eb972d43be3434dc70192@linux-foundation.org>
        <YsTLgQ45ESpsNEGV@casper.infradead.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Wed, 6 Jul 2022 00:38:41 +0100 Matthew Wilcox <willy@infradead.org> wrote:

> On Tue, Jul 05, 2022 at 02:18:19PM -0700, Andrew Morton wrote:
> > On Tue,  5 Jul 2022 20:35:32 +0800 Muchun Song <songmuchun@bytedance.com> wrote:
> > 
> > > FSDAX page refcounts are 1-based, rather than 0-based: if refcount is
> > > 1, then the page is freed.  The FSDAX pages can be pinned through GUP,
> > > then they will be unpinned via unpin_user_page() using a folio variant
> > > to put the page, however, folio variants did not consider this special
> > > case, the result will be to miss a wakeup event (like the user of
> > > __fuse_dax_break_layouts()).  Since FSDAX pages are only possible get
> > > by GUP users, so fix GUP instead of folio_put() to lower overhead.
> > > 
> > 
> > What are the user visible runtime effects of this bug?
> 
> "missing wake up event" seems pretty obvious to me?  Something goes to
> sleep waiting for a page to become unused, and is never woken.

No, missed wakeups are often obscured by another wakeup coming in
shortly afterwards.

If this wakeup is not one of these, then are there reports from the
softlockup detector?

Do we have reports of processes permanently stuck in D state?