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: <157225677483.3442.4227193290486305330.stgit@buzz>
In-Reply-To: <157225677483.3442.4227193290486305330.stgit@buzz>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Mon, 28 Oct 2019 13:39:46 +0100
Message-ID: <CAHk-=wjmLgo7DQT7Cy5rAGd=+2OK5Lqa8BN9qJFW1NPRoDfx5A@mail.gmail.com>
Subject: Re: [PATCH] mm/filemap: do not allocate cache pages beyond end of
 file at read
To:     Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Cc:     Linux-MM <linux-mm@kvack.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        Alexander Viro <viro@zeniv.linux.org.uk>,
        Johannes Weiner <hannes@cmpxchg.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Oct 28, 2019 at 10:59 AM Konstantin Khlebnikov
<khlebnikov@yandex-team.ru> wrote:
>
> Page cache could contain pages beyond end of file during write or
> if read races with truncate. But generic_file_buffered_read() always
> allocates unneeded pages beyond eof if somebody reads here and one
> extra page at the end if file size is page-aligned.

I wonder if we could just do something like this instead:

  diff --git a/mm/filemap.c b/mm/filemap.c
  index 85b7d087eb45..80b08433c93a 100644
  --- a/mm/filemap.c
  +++ b/mm/filemap.c
  @@ -2013,7 +2013,7 @@ static ssize_t generic_file_buffered_read(
        struct address_space *mapping = filp->f_mapping;
        struct inode *inode = mapping->host;
        struct file_ra_state *ra = &filp->f_ra;
  -     loff_t *ppos = &iocb->ki_pos;
  +     loff_t *ppos = &iocb->ki_pos, size;
        pgoff_t index;
        pgoff_t last_index;
        pgoff_t prev_index;
  @@ -2021,9 +2021,10 @@ static ssize_t generic_file_buffered_read(
        unsigned int prev_offset;
        int error = 0;

  -     if (unlikely(*ppos >= inode->i_sb->s_maxbytes))
  +     size = i_size_read(inode);
  +     if (unlikely(*ppos >= size))
                return 0;
  -     iov_iter_truncate(iter, inode->i_sb->s_maxbytes);
  +     iov_iter_truncate(iter, size);

        index = *ppos >> PAGE_SHIFT;
        prev_index = ra->prev_pos >> PAGE_SHIFT;

and yes, we still need to re-check the inode size after we've read the
page cache page (since it might have changed during the IO), but the
above seems fairly benign and simple.

Hmm?

              Linus