[PATCH] iomap: iomap that extends beyond EOF should be marked dirty

[PATCH] iomap: iomap that extends beyond EOF should be marked dirty

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

When doing a direct IO that spans the current EOF, and there are
written blocks beyond EOF that extend beyond the current write, the
only metadata update that needs to be done is a file size extension.

However, we don't mark such iomaps as IOMAP_F_DIRTY to indicate that
there is IO completion metadata updates required, and hence we may
fail to correctly sync file size extensions made in IO completion
when O_DSYNC writes are beingt used and the hardware supports FUA.

Hence when setting IOMAP_F_DIRTY, we need to also take into account
whether the iomap spans the current EOF. If it does, then we need to
mark it dirty so that IO completion will call generic_write_sync()
to flush the inode size update to stable storage correctly.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 fs/ext4/inode.c       | 9 ++++++++-
 fs/xfs/xfs_iomap.c    | 8 ++++++++
 include/linux/iomap.h | 2 ++
 3 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 516faa280ced..e9dc52537e5b 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -3523,9 +3523,16 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
 			return ret;
 	}
 
+	/*
+	 * Writes that span EOF might trigger an IO size update on completion,
+	 * so consider them to be dirty for the purposes of O_DSYNC even if
+	 * there is no other metadata changes being made or are pending here.
+	 */
 	iomap->flags = 0;
-	if (ext4_inode_datasync_dirty(inode))
+	if (ext4_inode_datasync_dirty(inode) ||
+	    offset + length > i_size_read(inode))
 		iomap->flags |= IOMAP_F_DIRTY;
+
 	iomap->bdev = inode->i_sb->s_bdev;
 	iomap->dax_dev = sbi->s_daxdev;
 	iomap->offset = (u64)first_block << blkbits;
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index f780e223b118..38be06f19ea2 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -722,6 +722,14 @@ xfs_file_iomap_begin_delay(
 		xfs_trim_extent(&imap, cmap.br_startoff, cmap.br_blockcount);
 		shared = true;
 	}
+
+	/*
+	 * Writes that span EOF might trigger an IO size update on completion,
+	 * so consider them to be dirty for the purposes of O_DSYNC even if
+	 * there is no other metadata changes being made or are pending here.
+	 */
+	if (offset + count > i_size_read(inode))
+		iomap->flags |= IOMAP_F_DIRTY;
 	error = xfs_bmbt_to_iomap(ip, iomap, &imap, shared);
 out_unlock:
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
diff --git a/include/linux/iomap.h b/include/linux/iomap.h
index 7aa5d6117936..24bd227d59f9 100644
--- a/include/linux/iomap.h
+++ b/include/linux/iomap.h
@@ -32,6 +32,8 @@ struct vm_fault;
  *
  * IOMAP_F_DIRTY indicates the inode has uncommitted metadata needed to access
  * written data and requires fdatasync to commit them to persistent storage.
+ * This needs to take into account metadata changes that *may* be made at IO
+ * completion, such as file size updates from direct IO.
  */
 #define IOMAP_F_NEW		0x01	/* blocks have been newly allocated */
 #define IOMAP_F_DIRTY		0x02	/* uncommitted metadata */
-- 
2.23.0.rc1

Re: [PATCH] iomap: iomap that extends beyond EOF should be marked dirty

[Dave Chinner]

On Wed, Oct 16, 2019 at 04:11:01PM +1100, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> When doing a direct IO that spans the current EOF, and there are
> written blocks beyond EOF that extend beyond the current write, the
> only metadata update that needs to be done is a file size extension.
> 
> However, we don't mark such iomaps as IOMAP_F_DIRTY to indicate that
> there is IO completion metadata updates required, and hence we may
> fail to correctly sync file size extensions made in IO completion
> when O_DSYNC writes are beingt used and the hardware supports FUA.
> 
> Hence when setting IOMAP_F_DIRTY, we need to also take into account
> whether the iomap spans the current EOF. If it does, then we need to
> mark it dirty so that IO completion will call generic_write_sync()
> to flush the inode size update to stable storage correctly.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
> ---
>  fs/ext4/inode.c       | 9 ++++++++-
>  fs/xfs/xfs_iomap.c    | 8 ++++++++
>  include/linux/iomap.h | 2 ++
>  3 files changed, 18 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 516faa280ced..e9dc52537e5b 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -3523,9 +3523,16 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
>  			return ret;
>  	}
>  
> +	/*
> +	 * Writes that span EOF might trigger an IO size update on completion,
> +	 * so consider them to be dirty for the purposes of O_DSYNC even if
> +	 * there is no other metadata changes being made or are pending here.
> +	 */
>  	iomap->flags = 0;
> -	if (ext4_inode_datasync_dirty(inode))
> +	if (ext4_inode_datasync_dirty(inode) ||
> +	    offset + length > i_size_read(inode))
>  		iomap->flags |= IOMAP_F_DIRTY;
> +
>  	iomap->bdev = inode->i_sb->s_bdev;
>  	iomap->dax_dev = sbi->s_daxdev;
>  	iomap->offset = (u64)first_block << blkbits;
> diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
> index f780e223b118..38be06f19ea2 100644
> --- a/fs/xfs/xfs_iomap.c
> +++ b/fs/xfs/xfs_iomap.c
> @@ -722,6 +722,14 @@ xfs_file_iomap_begin_delay(
>  		xfs_trim_extent(&imap, cmap.br_startoff, cmap.br_blockcount);
>  		shared = true;
>  	}
> +
> +	/*
> +	 * Writes that span EOF might trigger an IO size update on completion,
> +	 * so consider them to be dirty for the purposes of O_DSYNC even if
> +	 * there is no other metadata changes being made or are pending here.
> +	 */
> +	if (offset + count > i_size_read(inode))
> +		iomap->flags |= IOMAP_F_DIRTY;
>  	error = xfs_bmbt_to_iomap(ip, iomap, &imap, shared);
>  out_unlock:
>  	xfs_iunlock(ip, XFS_ILOCK_EXCL);

Urk, self-nack.

I put this in the wrong function. Too much spaghetti that all looks
similar in the XFS iomap code now. Will repost in a minute.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH] iomap: iomap that extends beyond EOF should be marked dirty

[Darrick J. Wong]

On Wed, Oct 16, 2019 at 04:11:01PM +1100, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> When doing a direct IO that spans the current EOF, and there are
> written blocks beyond EOF that extend beyond the current write, the
> only metadata update that needs to be done is a file size extension.
> 
> However, we don't mark such iomaps as IOMAP_F_DIRTY to indicate that
> there is IO completion metadata updates required, and hence we may
> fail to correctly sync file size extensions made in IO completion
> when O_DSYNC writes are beingt used and the hardware supports FUA.
> 
> Hence when setting IOMAP_F_DIRTY, we need to also take into account
> whether the iomap spans the current EOF. If it does, then we need to
> mark it dirty so that IO completion will call generic_write_sync()
> to flush the inode size update to stable storage correctly.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
> ---
>  fs/ext4/inode.c       | 9 ++++++++-
>  fs/xfs/xfs_iomap.c    | 8 ++++++++
>  include/linux/iomap.h | 2 ++
>  3 files changed, 18 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 516faa280ced..e9dc52537e5b 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -3523,9 +3523,16 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
>  			return ret;
>  	}
>  
> +	/*
> +	 * Writes that span EOF might trigger an IO size update on completion,
> +	 * so consider them to be dirty for the purposes of O_DSYNC even if
> +	 * there is no other metadata changes being made or are pending here.
> +	 */
>  	iomap->flags = 0;
> -	if (ext4_inode_datasync_dirty(inode))
> +	if (ext4_inode_datasync_dirty(inode) ||
> +	    offset + length > i_size_read(inode))
>  		iomap->flags |= IOMAP_F_DIRTY;
> +
>  	iomap->bdev = inode->i_sb->s_bdev;
>  	iomap->dax_dev = sbi->s_daxdev;
>  	iomap->offset = (u64)first_block << blkbits;
> diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
> index f780e223b118..38be06f19ea2 100644
> --- a/fs/xfs/xfs_iomap.c
> +++ b/fs/xfs/xfs_iomap.c
> @@ -722,6 +722,14 @@ xfs_file_iomap_begin_delay(
>  		xfs_trim_extent(&imap, cmap.br_startoff, cmap.br_blockcount);
>  		shared = true;
>  	}
> +
> +	/*
> +	 * Writes that span EOF might trigger an IO size update on completion,
> +	 * so consider them to be dirty for the purposes of O_DSYNC even if
> +	 * there is no other metadata changes being made or are pending here.
> +	 */
> +	if (offset + count > i_size_read(inode))
> +		iomap->flags |= IOMAP_F_DIRTY;

This ought to be in xfs_direct_write_iomap_begin(), right?

(Hoping to see another rev of Christoph's iomap cleanup series... ;))

--D

>  	error = xfs_bmbt_to_iomap(ip, iomap, &imap, shared);
>  out_unlock:
>  	xfs_iunlock(ip, XFS_ILOCK_EXCL);
> diff --git a/include/linux/iomap.h b/include/linux/iomap.h
> index 7aa5d6117936..24bd227d59f9 100644
> --- a/include/linux/iomap.h
> +++ b/include/linux/iomap.h
> @@ -32,6 +32,8 @@ struct vm_fault;
>   *
>   * IOMAP_F_DIRTY indicates the inode has uncommitted metadata needed to access
>   * written data and requires fdatasync to commit them to persistent storage.
> + * This needs to take into account metadata changes that *may* be made at IO
> + * completion, such as file size updates from direct IO.
>   */
>  #define IOMAP_F_NEW		0x01	/* blocks have been newly allocated */
>  #define IOMAP_F_DIRTY		0x02	/* uncommitted metadata */
> -- 
> 2.23.0.rc1
> 

[PATCH v2] iomap: iomap that extends beyond EOF should be marked dirty

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

When doing a direct IO that spans the current EOF, and there are
written blocks beyond EOF that extend beyond the current write, the
only metadata update that needs to be done is a file size extension.

However, we don't mark such iomaps as IOMAP_F_DIRTY to indicate that
there is IO completion metadata updates required, and hence we may
fail to correctly sync file size extensions made in IO completion
when O_DSYNC writes are being used and the hardware supports FUA.

Hence when setting IOMAP_F_DIRTY, we need to also take into account
whether the iomap spans the current EOF. If it does, then we need to
mark it dirty so that IO completion will call generic_write_sync()
to flush the inode size update to stable storage correctly.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
V2: put the XFS code in the right function, tested on non-FUA
capable hardware, too.

 fs/ext4/inode.c       | 9 ++++++++-
 fs/xfs/xfs_iomap.c    | 7 +++++++
 include/linux/iomap.h | 2 ++
 3 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 516faa280ced..e9dc52537e5b 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -3523,9 +3523,16 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
 			return ret;
 	}
 
+	/*
+	 * Writes that span EOF might trigger an IO size update on completion,
+	 * so consider them to be dirty for the purposes of O_DSYNC even if
+	 * there is no other metadata changes being made or are pending here.
+	 */
 	iomap->flags = 0;
-	if (ext4_inode_datasync_dirty(inode))
+	if (ext4_inode_datasync_dirty(inode) ||
+	    offset + length > i_size_read(inode))
 		iomap->flags |= IOMAP_F_DIRTY;
+
 	iomap->bdev = inode->i_sb->s_bdev;
 	iomap->dax_dev = sbi->s_daxdev;
 	iomap->offset = (u64)first_block << blkbits;
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index f780e223b118..32993c2acbd9 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -1049,6 +1049,13 @@ xfs_file_iomap_begin(
 	trace_xfs_iomap_alloc(ip, offset, length, XFS_DATA_FORK, &imap);
 
 out_finish:
+	/*
+	 * Writes that span EOF might trigger an IO size update on completion,
+	 * so consider them to be dirty for the purposes of O_DSYNC even if
+	 * there is no other metadata changes pending or have been made here.
+	 */
+	if ((flags & IOMAP_WRITE) && offset + length > i_size_read(inode))
+		iomap->flags |= IOMAP_F_DIRTY;
 	return xfs_bmbt_to_iomap(ip, iomap, &imap, shared);
 
 out_found:
diff --git a/include/linux/iomap.h b/include/linux/iomap.h
index 7aa5d6117936..24bd227d59f9 100644
--- a/include/linux/iomap.h
+++ b/include/linux/iomap.h
@@ -32,6 +32,8 @@ struct vm_fault;
  *
  * IOMAP_F_DIRTY indicates the inode has uncommitted metadata needed to access
  * written data and requires fdatasync to commit them to persistent storage.
+ * This needs to take into account metadata changes that *may* be made at IO
+ * completion, such as file size updates from direct IO.
  */
 #define IOMAP_F_NEW		0x01	/* blocks have been newly allocated */
 #define IOMAP_F_DIRTY		0x02	/* uncommitted metadata */

Re: [PATCH] iomap: iomap that extends beyond EOF should be marked dirty

[Christoph Hellwig]

On Tue, Oct 15, 2019 at 10:27:13PM -0700, Darrick J. Wong wrote:
> > +	/*
> > +	 * Writes that span EOF might trigger an IO size update on completion,
> > +	 * so consider them to be dirty for the purposes of O_DSYNC even if
> > +	 * there is no other metadata changes being made or are pending here.
> > +	 */
> > +	if (offset + count > i_size_read(inode))
> > +		iomap->flags |= IOMAP_F_DIRTY;
> 
> This ought to be in xfs_direct_write_iomap_begin(), right?
> 
> (Hoping to see another rev of Christoph's iomap cleanup series... ;))

I need to finish off all the nitpicks on the first iomap series..

Also we'll want this patch in first as it is 5.4 / -stable material,
so I'll need to rebase on top of that as well.

Re: [PATCH v2] iomap: iomap that extends beyond EOF should be marked dirty

[Christoph Hellwig]

I wonder if the i_size check should be done in core, similar to the
iomap_block_needs_zeroing helper.  But independent of what is nicer
this version does look correct to me:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v2] iomap: iomap that extends beyond EOF should be marked dirty

[Theodore Y. Ts'o]

On Wed, Oct 16, 2019 at 05:06:04PM +1100, Dave Chinner wrote:
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 516faa280ced..e9dc52537e5b 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -3523,9 +3523,16 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
>  			return ret;
>  	}
>  
> +	/*
> +	 * Writes that span EOF might trigger an IO size update on completion,
> +	 * so consider them to be dirty for the purposes of O_DSYNC even if
> +	 * there is no other metadata changes being made or are pending here.
> +	 */
>  	iomap->flags = 0;
> -	if (ext4_inode_datasync_dirty(inode))
> +	if (ext4_inode_datasync_dirty(inode) ||
> +	    offset + length > i_size_read(inode))
>  		iomap->flags |= IOMAP_F_DIRTY;
> +
>  	iomap->bdev = inode->i_sb->s_bdev;
>  	iomap->dax_dev = sbi->s_daxdev;
>  	iomap->offset = (u64)first_block << blkbits;

Ext4 is not currently using iomap for any kind of writing right now,
so perhaps this should land via Matthew's patchset?

					- Ted

Re: [PATCH v2] iomap: iomap that extends beyond EOF should be marked dirty

[Christoph Hellwig]

On Thu, Oct 17, 2019 at 08:29:11AM -0400, Theodore Y. Ts'o wrote:
> > +	/*
> > +	 * Writes that span EOF might trigger an IO size update on completion,
> > +	 * so consider them to be dirty for the purposes of O_DSYNC even if
> > +	 * there is no other metadata changes being made or are pending here.
> > +	 */
> >  	iomap->flags = 0;
> > -	if (ext4_inode_datasync_dirty(inode))
> > +	if (ext4_inode_datasync_dirty(inode) ||
> > +	    offset + length > i_size_read(inode))
> >  		iomap->flags |= IOMAP_F_DIRTY;
> > +
> >  	iomap->bdev = inode->i_sb->s_bdev;
> >  	iomap->dax_dev = sbi->s_daxdev;
> >  	iomap->offset = (u64)first_block << blkbits;
> 
> Ext4 is not currently using iomap for any kind of writing right now,
> so perhaps this should land via Matthew's patchset?

It does for DAX, which is one of the consumers of IOMAP_F_DIRTY.

Re: [PATCH v2] iomap: iomap that extends beyond EOF should be marked dirty

[Theodore Y. Ts'o]

On Thu, Oct 17, 2019 at 07:17:05AM -0700, Christoph Hellwig wrote:
> > 
> > Ext4 is not currently using iomap for any kind of writing right now,
> > so perhaps this should land via Matthew's patchset?
> 
> It does for DAX, which is one of the consumers of IOMAP_F_DIRTY.

Ah, right, I had forgotten about DAX.

					- Ted