[4/4] Btrfs: remove no longer needed range length checks for deduplication
diff mbox series

Message ID 20181212180559.15249-5-fdmanana@kernel.org
State New
Headers show
Series
  • Btrfs: a few more cleanups and fixes for clone/deduplication
Related show

Commit Message

Filipe Manana Dec. 12, 2018, 6:05 p.m. UTC
From: Filipe Manana <fdmanana@suse.com>

Comparing the content of the pages in the range to deduplicate is now done
by the generic helper generic_remap_file_range_prep(), which takes care of
ensuring we do not compare/deduplicate undefined data beyond a file's eof
(range from eof to the next block boundary). So remove these checks which
are now redundant.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
---
 fs/btrfs/ioctl.c | 19 ++-----------------
 1 file changed, 2 insertions(+), 17 deletions(-)

Comments

Nikolay Borisov Dec. 13, 2018, 12:20 p.m. UTC | #1
On 12.12.18 г. 20:05 ч., fdmanana@kernel.org wrote:
> From: Filipe Manana <fdmanana@suse.com>
> 
> Comparing the content of the pages in the range to deduplicate is now done
> by the generic helper generic_remap_file_range_prep(), which takes care of

very minor nit: the checks are performed in generic_remap_checks which
is called by generic_remap_file_range_prep. But even without this change
LGTM.

> ensuring we do not compare/deduplicate undefined data beyond a file's eof
> (range from eof to the next block boundary). So remove these checks which
> are now redundant.
> 
> Signed-off-by: Filipe Manana <fdmanana@suse.com>
> ---
>  fs/btrfs/ioctl.c | 19 ++-----------------
>  1 file changed, 2 insertions(+), 17 deletions(-)
> 
> diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
> index 4e9efc93340e..3a27efa2b955 100644
> --- a/fs/btrfs/ioctl.c
> +++ b/fs/btrfs/ioctl.c
> @@ -3206,31 +3206,16 @@ static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
>  	inode_lock_nested(inode2, I_MUTEX_CHILD);
>  }
>  
> -static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 olen,
> +static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 len,
>  				   struct inode *dst, u64 dst_loff)
>  {
> -	u64 bs = BTRFS_I(src)->root->fs_info->sb->s_blocksize;
>  	int ret;
> -	u64 len = olen;
> -
> -	if (loff + len == src->i_size)
> -		len = ALIGN(src->i_size, bs) - loff;
> -	/*
> -	 * For same inode case we don't want our length pushed out past i_size
> -	 * as comparing that data range makes no sense.
> -	 *
> -	 * This effectively means we require aligned extents for the single
> -	 * inode case, whereas the other cases allow an unaligned length so long
> -	 * as it ends at i_size.
> -	 */
> -	if (dst == src && len != olen)
> -		return -EINVAL;
>  
>  	/*
>  	 * Lock destination range to serialize with concurrent readpages().
>  	 */
>  	lock_extent(&BTRFS_I(dst)->io_tree, dst_loff, dst_loff + len - 1);
> -	ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
> +	ret = btrfs_clone(src, dst, loff, len, len, dst_loff, 1);
>  	unlock_extent(&BTRFS_I(dst)->io_tree, dst_loff, dst_loff + len - 1);
>  
>  	return ret;
>
Filipe Manana Jan. 31, 2019, 4:31 p.m. UTC | #2
On Wed, Dec 12, 2018 at 6:07 PM <fdmanana@kernel.org> wrote:
>
> From: Filipe Manana <fdmanana@suse.com>
>
> Comparing the content of the pages in the range to deduplicate is now done
> by the generic helper generic_remap_file_range_prep(), which takes care of
> ensuring we do not compare/deduplicate undefined data beyond a file's eof
> (range from eof to the next block boundary). So remove these checks which
> are now redundant.
>
> Signed-off-by: Filipe Manana <fdmanana@suse.com>

Any reason why this was not yet picked?
Thanks.

> ---
>  fs/btrfs/ioctl.c | 19 ++-----------------
>  1 file changed, 2 insertions(+), 17 deletions(-)
>
> diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
> index 4e9efc93340e..3a27efa2b955 100644
> --- a/fs/btrfs/ioctl.c
> +++ b/fs/btrfs/ioctl.c
> @@ -3206,31 +3206,16 @@ static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
>         inode_lock_nested(inode2, I_MUTEX_CHILD);
>  }
>
> -static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 olen,
> +static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 len,
>                                    struct inode *dst, u64 dst_loff)
>  {
> -       u64 bs = BTRFS_I(src)->root->fs_info->sb->s_blocksize;
>         int ret;
> -       u64 len = olen;
> -
> -       if (loff + len == src->i_size)
> -               len = ALIGN(src->i_size, bs) - loff;
> -       /*
> -        * For same inode case we don't want our length pushed out past i_size
> -        * as comparing that data range makes no sense.
> -        *
> -        * This effectively means we require aligned extents for the single
> -        * inode case, whereas the other cases allow an unaligned length so long
> -        * as it ends at i_size.
> -        */
> -       if (dst == src && len != olen)
> -               return -EINVAL;
>
>         /*
>          * Lock destination range to serialize with concurrent readpages().
>          */
>         lock_extent(&BTRFS_I(dst)->io_tree, dst_loff, dst_loff + len - 1);
> -       ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
> +       ret = btrfs_clone(src, dst, loff, len, len, dst_loff, 1);
>         unlock_extent(&BTRFS_I(dst)->io_tree, dst_loff, dst_loff + len - 1);
>
>         return ret;
> --
> 2.11.0
>
Filipe Manana Feb. 12, 2019, 5:58 p.m. UTC | #3
On Thu, Jan 31, 2019 at 4:31 PM Filipe Manana <fdmanana@kernel.org> wrote:
>
> On Wed, Dec 12, 2018 at 6:07 PM <fdmanana@kernel.org> wrote:
> >
> > From: Filipe Manana <fdmanana@suse.com>
> >
> > Comparing the content of the pages in the range to deduplicate is now done
> > by the generic helper generic_remap_file_range_prep(), which takes care of
> > ensuring we do not compare/deduplicate undefined data beyond a file's eof
> > (range from eof to the next block boundary). So remove these checks which
> > are now redundant.
> >
> > Signed-off-by: Filipe Manana <fdmanana@suse.com>
>
> Any reason why this was not yet picked?
> Thanks.

Another ping. Thanks.

>
> > ---
> >  fs/btrfs/ioctl.c | 19 ++-----------------
> >  1 file changed, 2 insertions(+), 17 deletions(-)
> >
> > diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
> > index 4e9efc93340e..3a27efa2b955 100644
> > --- a/fs/btrfs/ioctl.c
> > +++ b/fs/btrfs/ioctl.c
> > @@ -3206,31 +3206,16 @@ static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
> >         inode_lock_nested(inode2, I_MUTEX_CHILD);
> >  }
> >
> > -static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 olen,
> > +static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 len,
> >                                    struct inode *dst, u64 dst_loff)
> >  {
> > -       u64 bs = BTRFS_I(src)->root->fs_info->sb->s_blocksize;
> >         int ret;
> > -       u64 len = olen;
> > -
> > -       if (loff + len == src->i_size)
> > -               len = ALIGN(src->i_size, bs) - loff;
> > -       /*
> > -        * For same inode case we don't want our length pushed out past i_size
> > -        * as comparing that data range makes no sense.
> > -        *
> > -        * This effectively means we require aligned extents for the single
> > -        * inode case, whereas the other cases allow an unaligned length so long
> > -        * as it ends at i_size.
> > -        */
> > -       if (dst == src && len != olen)
> > -               return -EINVAL;
> >
> >         /*
> >          * Lock destination range to serialize with concurrent readpages().
> >          */
> >         lock_extent(&BTRFS_I(dst)->io_tree, dst_loff, dst_loff + len - 1);
> > -       ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
> > +       ret = btrfs_clone(src, dst, loff, len, len, dst_loff, 1);
> >         unlock_extent(&BTRFS_I(dst)->io_tree, dst_loff, dst_loff + len - 1);
> >
> >         return ret;
> > --
> > 2.11.0
> >
David Sterba Feb. 18, 2019, 3:10 p.m. UTC | #4
On Thu, Jan 31, 2019 at 04:31:49PM +0000, Filipe Manana wrote:
> On Wed, Dec 12, 2018 at 6:07 PM <fdmanana@kernel.org> wrote:
> >
> > From: Filipe Manana <fdmanana@suse.com>
> >
> > Comparing the content of the pages in the range to deduplicate is now done
> > by the generic helper generic_remap_file_range_prep(), which takes care of
> > ensuring we do not compare/deduplicate undefined data beyond a file's eof
> > (range from eof to the next block boundary). So remove these checks which
> > are now redundant.
> >
> > Signed-off-by: Filipe Manana <fdmanana@suse.com>
> 
> Any reason why this was not yet picked?

Added to misc-next now.

Patch
diff mbox series

diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 4e9efc93340e..3a27efa2b955 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -3206,31 +3206,16 @@  static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
 	inode_lock_nested(inode2, I_MUTEX_CHILD);
 }
 
-static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 olen,
+static int btrfs_extent_same_range(struct inode *src, u64 loff, u64 len,
 				   struct inode *dst, u64 dst_loff)
 {
-	u64 bs = BTRFS_I(src)->root->fs_info->sb->s_blocksize;
 	int ret;
-	u64 len = olen;
-
-	if (loff + len == src->i_size)
-		len = ALIGN(src->i_size, bs) - loff;
-	/*
-	 * For same inode case we don't want our length pushed out past i_size
-	 * as comparing that data range makes no sense.
-	 *
-	 * This effectively means we require aligned extents for the single
-	 * inode case, whereas the other cases allow an unaligned length so long
-	 * as it ends at i_size.
-	 */
-	if (dst == src && len != olen)
-		return -EINVAL;
 
 	/*
 	 * Lock destination range to serialize with concurrent readpages().
 	 */
 	lock_extent(&BTRFS_I(dst)->io_tree, dst_loff, dst_loff + len - 1);
-	ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
+	ret = btrfs_clone(src, dst, loff, len, len, dst_loff, 1);
 	unlock_extent(&BTRFS_I(dst)->io_tree, dst_loff, dst_loff + len - 1);
 
 	return ret;