Re: [PATCH v2 6/8] vfs: copy_file_range should update file timestamps

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Dave Chinner <david@xxxxxxxxxxxxx> writes:

> On Mon, May 27, 2019 at 03:35:39PM +0100, Luis Henriques wrote:
>> On Sun, May 26, 2019 at 09:10:57AM +0300, Amir Goldstein wrote:
>> > From: Dave Chinner <dchinner@xxxxxxxxxx>
>> > 
>> > Timestamps are not updated right now, so programs looking for
>> > timestamp updates for file modifications (like rsync) will not
>> > detect that files have changed. We are also accessing the source
>> > data when doing a copy (but not when cloning) so we need to update
>> > atime on the source file as well.
>> > 
>> > Signed-off-by: Dave Chinner <dchinner@xxxxxxxxxx>
>> > Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx>
>> > ---
>> >  fs/read_write.c | 10 ++++++++++
>> >  1 file changed, 10 insertions(+)
>> > 
>> > diff --git a/fs/read_write.c b/fs/read_write.c
>> > index e16bcafc0da2..4b23a86aacd9 100644
>> > --- a/fs/read_write.c
>> > +++ b/fs/read_write.c
>> > @@ -1576,6 +1576,16 @@ int generic_copy_file_range_prep(struct file *file_in, struct file *file_out)
>> >  
>> >  	WARN_ON_ONCE(!inode_is_locked(file_inode(file_out)));
>> >  
>> > +	/* Update source timestamps, because we are accessing file data */
>> > +	file_accessed(file_in);
>> > +
>> > +	/* Update destination timestamps, since we can alter file contents. */
>> > +	if (!(file_out->f_mode & FMODE_NOCMTIME)) {
>> > +		ret = file_update_time(file_out);
>> > +		if (ret)
>> > +			return ret;
>> > +	}
>> > +
>> 
>> Is this the right place for updating the timestamps?  I see that in same
>> cases we may be updating the timestamp even if there was an error and no
>> copy was performed.  For example, if file_remove_privs fails.
>
> It's the same place we do it for read - file_accessed() is called
> before we do the IO - and the same place for write -
> file_update_time() is called before we copy data into the pagecache
> or do direct IO. As such, it really doesn't matter if it is before
> or after file_remove_privs() - the IO can still fail for many
> reasons after we've updated the timestamps and in some of the
> failure cases (e.g. we failed the sync at the end of an O_DSYNC
> buffered write) we still want the timestamps to be modified because
> the data and/or user visible metadata /may/ have been changed.
>
> cfr operates under the same constraints as read() and write(), so we
> need to update the timestamps up front regardless of whether the
> copy ends up succeeding or not....

Great, thanks for explaining it.  It now makes sense, even for
consistency, to have this operation here.

Cheers,
-- 
Luis



[Index of Archives]     [Linux Filesystem Development]     [Linux USB Development]     [Linux Media Development]     [Video for Linux]     [Linux NILFS]     [Linux Audio Users]     [Yosemite Info]     [Linux SCSI]

  Powered by Linux