On Wed, Oct 10, 2018 at 3:11 AM Darrick J. Wong <darrick.wong@xxxxxxxxxx> wrote:
>
> From: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
>
> File range remapping, if allowed to run past the destination file's EOF,
> is an optimization on a regular file write. Regular file writes that
> extend the file length are subject to various constraints which are not
> checked by range cloning.
>
> This is a correctness problem because we're never allowed to touch
> ranges that the page cache can't support (s_maxbytes); we're not
> supposed to deal with large offsets (MAX_NON_LFS) if O_LARGEFILE isn't
> set; and we must obey resource limits (RLIMIT_FSIZE).
>
> Therefore, add these checks to the new generic_remap_checks function so
> that we curtail unexpected behavior.
>
> Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> ---
> mm/filemap.c | 39 +++++++++++++++++++++++++++++++++++++++
> 1 file changed, 39 insertions(+)
>
>
> diff --git a/mm/filemap.c b/mm/filemap.c
> index 14041a8468ba..59056bd9c58a 100644
> --- a/mm/filemap.c
> +++ b/mm/filemap.c
> @@ -2974,6 +2974,27 @@ inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from)
> }
> EXPORT_SYMBOL(generic_write_checks);
>
> +static int
> +generic_remap_check_limits(struct file *file, loff_t pos, uint64_t *count)
> +{
> + struct inode *inode = file->f_mapping->host;
> +
> + /* Don't exceed the LFS limits. */
> + if (unlikely(pos + *count > MAX_NON_LFS &&
> + !(file->f_flags & O_LARGEFILE))) {
> + if (pos >= MAX_NON_LFS)
> + return -EFBIG;
> + *count = min(*count, MAX_NON_LFS - (uint64_t)pos);
> + }
> +
> + /* Don't operate on ranges the page cache doesn't support. */
> + if (unlikely(pos >= inode->i_sb->s_maxbytes))
> + return -EFBIG;
> +
> + *count = min(*count, inode->i_sb->s_maxbytes - (uint64_t)pos);
> + return 0;
> +}
> +
Sorry. I haven't explained myself properly last time.
What I meant is that it hurts my eyes to see generic_write_checks() and
generic_remap_check_limits() which from the line of (limit != RLIM_INFINITY)
are exactly the same thing. Yes, generic_remap_check_limits() uses
iov_iter_truncate(), but that's a minor semantic change - it can be easily
resolved by creating a dummy iter in generic_remap_checks() instead of
passing int *count.
You could say that this is nit picking, but the very reason this patch
set exists
it because clone/dedup implementation did not use the same range checks
of write to begin with, so it just seems wrong to diverge them at this point.
So to be clear, I suggest that generic_write_checks() should use your
generic_remap_check_limits() helper.
If you disagree and others can live with this minor duplication, fine by me.
Thanks,
Amir.
