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;
MIME-Version: 1.0
References: <20221004001050.3269377-1-daeho43@gmail.com> <85cc583c-279c-2d65-4cfd-ebd54468c6f5@kernel.org>
In-Reply-To: <85cc583c-279c-2d65-4cfd-ebd54468c6f5@kernel.org>
From:   Daeho Jeong <daeho43@gmail.com>
Date:   Tue, 4 Oct 2022 10:12:38 -0700
Message-ID: <CACOAw_xR0sr3j=6v+7vqs77Hk=tZ3BqxKBFyKpawe_rpUJsf=Q@mail.gmail.com>
Subject: Re: [f2fs-dev] [PATCH v3 1/2] f2fs: correct i_size change for atomic writes
To:     Chao Yu <chao@kernel.org>
Cc:     linux-kernel@vger.kernel.org,
        linux-f2fs-devel@lists.sourceforge.net, kernel-team@android.com,
        Daeho Jeong <daehojeong@google.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Oct 4, 2022 at 8:41 AM Chao Yu <chao@kernel.org> wrote:
>
> On 2022/10/4 8:10, Daeho Jeong wrote:
> > From: Daeho Jeong <daehojeong@google.com>
> >
> > We need to make sure i_size doesn't change until atomic write commit is
> > successful and restore it when commit is failed.
> >
> > Signed-off-by: Daeho Jeong <daehojeong@google.com>
> > ---
> > v3: make sure inode is clean while atomic writing
> > ---
> >   fs/f2fs/f2fs.h    |  1 +
> >   fs/f2fs/file.c    | 18 +++++++++++-------
> >   fs/f2fs/inode.c   |  3 +++
> >   fs/f2fs/segment.c |  4 +++-
> >   4 files changed, 18 insertions(+), 8 deletions(-)
> >
> > diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> > index dee7b67a17a6..539da7f12cfc 100644
> > --- a/fs/f2fs/f2fs.h
> > +++ b/fs/f2fs/f2fs.h
> > @@ -821,6 +821,7 @@ struct f2fs_inode_info {
> >       unsigned int i_cluster_size;            /* cluster size */
> >
> >       unsigned int atomic_write_cnt;
> > +     loff_t original_i_size;         /* original i_size before atomic write */
> >   };
> >
> >   static inline void get_extent_info(struct extent_info *ext,
> > diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
> > index 5efe0e4a725a..ce2336d2f688 100644
> > --- a/fs/f2fs/file.c
> > +++ b/fs/f2fs/file.c
> > @@ -1989,6 +1989,7 @@ static int f2fs_ioc_start_atomic_write(struct file *filp)
> >       struct f2fs_inode_info *fi = F2FS_I(inode);
> >       struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
> >       struct inode *pinode;
> > +     loff_t isize;
> >       int ret;
> >
> >       if (!inode_owner_or_capable(mnt_userns, inode))
> > @@ -2047,7 +2048,12 @@ static int f2fs_ioc_start_atomic_write(struct file *filp)
> >               f2fs_up_write(&fi->i_gc_rwsem[WRITE]);
> >               goto out;
> >       }
> > -     f2fs_i_size_write(fi->cow_inode, i_size_read(inode));
> > +
> > +     f2fs_write_inode(inode, NULL);
> > +
> > +     isize = i_size_read(inode);
> > +     fi->original_i_size = isize;
> > +     f2fs_i_size_write(fi->cow_inode, isize);
> >
> >       spin_lock(&sbi->inode_lock[ATOMIC_FILE]);
> >       sbi->atomic_files++;
> > @@ -2087,16 +2093,14 @@ static int f2fs_ioc_commit_atomic_write(struct file *filp)
> >
> >       if (f2fs_is_atomic_file(inode)) {
> >               ret = f2fs_commit_atomic_write(inode);
> > -             if (ret)
> > -                     goto unlock_out;
> > -
> > -             ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true);
> >               if (!ret)
> > -                     f2fs_abort_atomic_write(inode, false);
> > +                     ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true);
> > +
> > +             f2fs_abort_atomic_write(inode, ret);
> >       } else {
> >               ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 1, false);
> >       }
> > -unlock_out:
> > +
> >       inode_unlock(inode);
> >       mnt_drop_write_file(filp);
> >       return ret;
> > diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c
> > index cde0a3dc80c3..64d7772b4cd9 100644
> > --- a/fs/f2fs/inode.c
> > +++ b/fs/f2fs/inode.c
> > @@ -30,6 +30,9 @@ void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync)
> >       if (f2fs_inode_dirtied(inode, sync))
> >               return;
> >
> > +     if (f2fs_is_atomic_file(inode))
> > +             return;
> > +
> >       mark_inode_dirty_sync(inode);
> >   }
> >
> > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> > index 460048f3c850..143b7ea0fb8e 100644
> > --- a/fs/f2fs/segment.c
> > +++ b/fs/f2fs/segment.c
> > @@ -193,8 +193,10 @@ void f2fs_abort_atomic_write(struct inode *inode, bool clean)
> >       if (!f2fs_is_atomic_file(inode))
> >               return;
> >
> > -     if (clean)
> > +     if (clean) {
> >               truncate_inode_pages_final(inode->i_mapping);
> > +             f2fs_i_size_write(inode, fi->original_i_size);
>
> inode has atomic_write flag, it can not be set dirtied here, how
> about writing i_size ater clear_inode_flag(inode, FI_ATOMIC_FILE)?

Oh, right.
Thanks~!

>
> Thanks,
>
> > +     }
> >       clear_inode_flag(fi->cow_inode, FI_COW_FILE);
> >       iput(fi->cow_inode);
> >       fi->cow_inode = NULL;