linux-kernel - Re: [PATCH] f2fs: optimize f2fs DIO overwrites

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CACOAw_xpcjaSPXTrPaZZzed6UbfLpdBEww8HDmUHU3yacaq7sg@mail.gmail.com>
Date: Tue, 11 Mar 2025 14:14:14 -0700
From: Daeho Jeong <daeho43@...il.com>
To: Chao Yu <chao@...nel.org>
Cc: Yohan Joung <jyh429@...il.com>, jaegeuk@...nel.org, 
	linux-f2fs-devel@...ts.sourceforge.net, linux-kernel@...r.kernel.org, 
	Yohan Joung <yohan.joung@...com>
Subject: Re: [PATCH] f2fs: optimize f2fs DIO overwrites

On Tue, Mar 11, 2025 at 5:00 AM Chao Yu <chao@...nel.org> wrote:
>
> On 3/7/25 22:56, Yohan Joung wrote:
> > this is unnecessary when we know we are overwriting already allocated
> > blocks and the overhead of starting a transaction can be significant
> > especially for multithreaded workloads doing small writes.
>
> Hi Yohan,
>
> So you're trying to avoid f2fs_map_lock() in dio write path, right?
>
> Thanks,
>
> >
> > Signed-off-by: Yohan Joung <yohan.joung@...com>
> > ---
> >  fs/f2fs/data.c | 20 ++++++++++++++++++++
> >  fs/f2fs/f2fs.h |  1 +
> >  fs/f2fs/file.c |  5 ++++-
> >  3 files changed, 25 insertions(+), 1 deletion(-)
> >
> > diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
> > index 09437dbd1b42..728630037b74 100644
> > --- a/fs/f2fs/data.c
> > +++ b/fs/f2fs/data.c
> > @@ -4267,6 +4267,26 @@ static int f2fs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> >       return 0;
> >  }
> >
> > +static int f2fs_iomap_overwrite_begin(struct inode *inode, loff_t offset,
> > +             loff_t length, unsigned flags, struct iomap *iomap,
> > +             struct iomap *srcmap)
> > +{
> > +     int ret;
> > +
> > +     /*
> > +      * Even for writes we don't need to allocate blocks, so just pretend
> > +      * we are reading to save overhead of starting a transaction.
> > +      */
> > +     flags &= ~IOMAP_WRITE;
> > +     ret = f2fs_iomap_begin(inode, offset, length, flags, iomap, srcmap);
> > +     WARN_ON_ONCE(!ret && iomap->type != IOMAP_MAPPED);
> > +     return ret;
> > +}
> > +
> >  const struct iomap_ops f2fs_iomap_ops = {
> >       .iomap_begin    = f2fs_iomap_begin,
> >  };
> > +
> > +const struct iomap_ops f2fs_iomap_overwrite_ops = {
> > +     .iomap_begin    = f2fs_iomap_overwrite_begin,
> > +};
> > diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> > index c6cc2694f9ac..0511ab5ed42a 100644
> > --- a/fs/f2fs/f2fs.h
> > +++ b/fs/f2fs/f2fs.h
> > @@ -3936,6 +3936,7 @@ void f2fs_destroy_post_read_processing(void);
> >  int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
> >  void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
> >  extern const struct iomap_ops f2fs_iomap_ops;
> > +extern const struct iomap_ops f2fs_iomap_overwrite_ops;
> >
> >  static inline struct page *f2fs_find_data_page(struct inode *inode,
> >               pgoff_t index, pgoff_t *next_pgofs)
> > diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
> > index 82b21baf5628..bb2fe6dac9b6 100644
> > --- a/fs/f2fs/file.c
> > +++ b/fs/f2fs/file.c
> > @@ -4985,6 +4985,7 @@ static ssize_t f2fs_dio_write_iter(struct kiocb *iocb, struct iov_iter *from,
> >       const ssize_t count = iov_iter_count(from);
> >       unsigned int dio_flags;
> >       struct iomap_dio *dio;
> > +     const struct iomap_ops *iomap_ops = &f2fs_iomap_ops;
> >       ssize_t ret;
> >
> >       trace_f2fs_direct_IO_enter(inode, iocb, count, WRITE);
> > @@ -5025,7 +5026,9 @@ static ssize_t f2fs_dio_write_iter(struct kiocb *iocb, struct iov_iter *from,
> >       dio_flags = 0;
> >       if (pos + count > inode->i_size)
> >               dio_flags |= IOMAP_DIO_FORCE_WAIT;
> > -     dio = __iomap_dio_rw(iocb, from, &f2fs_iomap_ops,
> > +     else if (f2fs_overwrite_io(inode, pos, count))
> > +             iomap_ops = &f2fs_iomap_overwrite_ops;
> > +     dio = __iomap_dio_rw(iocb, from, iomap_ops,
> >                            &f2fs_iomap_dio_write_ops, dio_flags, NULL, 0);
> >       if (IS_ERR_OR_NULL(dio)) {
> >               ret = PTR_ERR_OR_ZERO(dio);

I think we can add the overwrite check in f2fs_iomap_begin() before
setting the map.m_may_create, instead of adding a new function
f2fs_iomap_overwrite_begin().
What do you think?

>