On Mon, Apr 23, 2018 at 3:53 PM, Vivek Goyal <vgoyal@xxxxxxxxxx> wrote:
> On Mon, Apr 23, 2018 at 03:39:45PM +0200, Miklos Szeredi wrote:
>> On Mon, Apr 23, 2018 at 3:36 PM, Vivek Goyal <vgoyal@xxxxxxxxxx> wrote:
>> > On Thu, Apr 12, 2018 at 05:08:04PM +0200, Miklos Szeredi wrote:
>> >> Implement stacked fsync().
>> >>
>> >> Signed-off-by: Miklos Szeredi <mszeredi@xxxxxxxxxx>
>> >> ---
>> >> fs/overlayfs/file.c | 20 ++++++++++++++++++++
>> >> 1 file changed, 20 insertions(+)
>> >>
>> >> diff --git a/fs/overlayfs/file.c b/fs/overlayfs/file.c
>> >> index b98204c1c19c..4417527667ff 100644
>> >> --- a/fs/overlayfs/file.c
>> >> +++ b/fs/overlayfs/file.c
>> >> @@ -222,10 +222,30 @@ static ssize_t ovl_write_iter(struct kiocb *iocb, struct iov_iter *iter)
>> >> return ret;
>> >> }
>> >>
>> >> +static int ovl_fsync(struct file *file, loff_t start, loff_t end, int datasync)
>> >> +{
>> >> + struct fd real;
>> >> + const struct cred *old_cred;
>> >> + int ret;
>> >> +
>> >> + ret = ovl_real_file(file, &real);
>> >> + if (ret)
>> >> + return ret;
>> >> +
>> >> + old_cred = ovl_override_creds(file_inode(file)->i_sb);
>> >> + ret = vfs_fsync_range(real.file, start, end, datasync);
>> >> + revert_creds(old_cred);
>> >
>> > Can we avoid calling fsync() on real file if it is not upper. Is it worth
>> > optimizing.
>>
>> Not sure it's worth bothering with. If caller of fsync(2) didn't
>> worry about cost, then why should we?
>
> I was thinking more from the point of view of metadata copy up patches.
> For a metacopy file, I was thinking if I can just issue fsync() on upper
> file and skip it on lower file.
Ah, in that case I agree with doing fsync only on upper. If there's
a choice in implementing the given functionality (and performance
doesn't matter) then the simplest one should be chosen.
Thanks,
Miklos
