[PATCH v21 08/30] splice: Make splice from a DAX file use copy_splice_read()

[PATCH v21 08/30] splice: Make splice from a DAX file use copy_splice_read()

[David Howells]

Make a read splice from a DAX file go directly to copy_splice_read() to do
the reading as filemap_splice_read() is unlikely to find any pagecache to
splice.

I think this affects only erofs, Ext2, Ext4, fuse and XFS.

Signed-off-by: David Howells <dhowells@redhat.com>
cc: Christoph Hellwig <hch@lst.de>
cc: Al Viro <viro@zeniv.linux.org.uk>
cc: Jens Axboe <axboe@kernel.dk>
cc: linux-erofs@lists.ozlabs.org
cc: linux-ext4@vger.kernel.org
cc: linux-xfs@vger.kernel.org
cc: linux-fsdevel@vger.kernel.org
cc: linux-block@vger.kernel.org
cc: linux-mm@kvack.org
---

Notes:
    ver #21)
     - Don't need #ifdef CONFIG_FS_DAX as IS_DAX() is false if !CONFIG_FS_DAX.
     - Needs to be in vfs_splice_read(), not generic_file_splice_read().

 fs/splice.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/fs/splice.c b/fs/splice.c
index 76126b1aafcb..8268248df3a9 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -908,10 +908,10 @@ long vfs_splice_read(struct file *in, loff_t *ppos,
 	if (unlikely(!in->f_op->splice_read))
 		return warn_unsupported(in, "read");
 	/*
-	 * O_DIRECT doesn't deal with the pagecache, so we allocate a buffer,
-	 * copy into it and splice that into the pipe.
+	 * O_DIRECT and DAX don't deal with the pagecache, so we allocate a
+	 * buffer, copy into it and splice that into the pipe.
 	 */
-	if ((in->f_flags & O_DIRECT))
+	if ((in->f_flags & O_DIRECT) || IS_DAX(in->f_mapping->host))
 		return copy_splice_read(in, ppos, pipe, len, flags);
 	return in->f_op->splice_read(in, ppos, pipe, len, flags);
 }

[PATCH v21 24/30] xfs: Provide a splice-read stub

[David Howells]

Provide a splice_read stub for XFS.  This does a stat count and a shutdown
check before proceeding, then emits a new trace line and locks the inode
across the call to filemap_splice_read() and adds to the stats afterwards.
Splicing from direct I/O or DAX is handled by the caller.

Signed-off-by: David Howells <dhowells@redhat.com>
cc: Christoph Hellwig <hch@lst.de>
cc: Al Viro <viro@zeniv.linux.org.uk>
cc: Jens Axboe <axboe@kernel.dk>
cc: Darrick J. Wong <djwong@kernel.org>
cc: linux-xfs@vger.kernel.org
cc: linux-fsdevel@vger.kernel.org
cc: linux-block@vger.kernel.org
cc: linux-mm@kvack.org
---
 fs/xfs/xfs_file.c  | 30 +++++++++++++++++++++++++++++-
 fs/xfs/xfs_trace.h |  2 +-
 2 files changed, 30 insertions(+), 2 deletions(-)

diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index aede746541f8..08d632668e94 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -306,6 +306,34 @@ xfs_file_read_iter(
 	return ret;
 }
 
+STATIC ssize_t
+xfs_file_splice_read(
+	struct file		*in,
+	loff_t			*ppos,
+	struct pipe_inode_info	*pipe,
+	size_t			len,
+	unsigned int		flags)
+{
+	struct inode		*inode = file_inode(in);
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	ssize_t			ret = 0;
+
+	XFS_STATS_INC(mp, xs_read_calls);
+
+	if (xfs_is_shutdown(mp))
+		return -EIO;
+
+	trace_xfs_file_splice_read(ip, *ppos, len);
+
+	xfs_ilock(ip, XFS_IOLOCK_SHARED);
+	ret = filemap_splice_read(in, ppos, pipe, len, flags);
+	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
+	if (ret > 0)
+		XFS_STATS_ADD(mp, xs_read_bytes, ret);
+	return ret;
+}
+
 /*
  * Common pre-write limit and setup checks.
  *
@@ -1423,7 +1451,7 @@ const struct file_operations xfs_file_operations = {
 	.llseek		= xfs_file_llseek,
 	.read_iter	= xfs_file_read_iter,
 	.write_iter	= xfs_file_write_iter,
-	.splice_read	= generic_file_splice_read,
+	.splice_read	= xfs_file_splice_read,
 	.splice_write	= iter_file_splice_write,
 	.iopoll		= iocb_bio_iopoll,
 	.unlocked_ioctl	= xfs_file_ioctl,
diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
index cd4ca5b1fcb0..4db669203149 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -1445,7 +1445,6 @@ DEFINE_RW_EVENT(xfs_file_direct_write);
 DEFINE_RW_EVENT(xfs_file_dax_write);
 DEFINE_RW_EVENT(xfs_reflink_bounce_dio_write);
 
-
 DECLARE_EVENT_CLASS(xfs_imap_class,
 	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count,
 		 int whichfork, struct xfs_bmbt_irec *irec),
@@ -1535,6 +1534,7 @@ DEFINE_SIMPLE_IO_EVENT(xfs_zero_eof);
 DEFINE_SIMPLE_IO_EVENT(xfs_end_io_direct_write);
 DEFINE_SIMPLE_IO_EVENT(xfs_end_io_direct_write_unwritten);
 DEFINE_SIMPLE_IO_EVENT(xfs_end_io_direct_write_append);
+DEFINE_SIMPLE_IO_EVENT(xfs_file_splice_read);
 
 DECLARE_EVENT_CLASS(xfs_itrunc_class,
 	TP_PROTO(struct xfs_inode *ip, xfs_fsize_t new_size),

[PATCH v21 25/30] zonefs: Provide a splice-read stub

[David Howells]

Provide a splice_read stub for zonefs.  This does some checks before
proceeding and locks the inode across the call to filemap_splice_read() and
a size check in case of truncation.  Splicing from direct I/O is handled by
the caller.

Signed-off-by: David Howells <dhowells@redhat.com>
cc: Christoph Hellwig <hch@lst.de>
cc: Al Viro <viro@zeniv.linux.org.uk>
cc: Jens Axboe <axboe@kernel.dk>
cc: Darrick J. Wong <djwong@kernel.org>
cc: linux-xfs@vger.kernel.org
cc: linux-fsdevel@vger.kernel.org
cc: linux-block@vger.kernel.org
cc: linux-mm@kvack.org
---
 fs/zonefs/file.c | 40 +++++++++++++++++++++++++++++++++++++++-
 1 file changed, 39 insertions(+), 1 deletion(-)

diff --git a/fs/zonefs/file.c b/fs/zonefs/file.c
index 132f01d3461f..65d4c4fe6364 100644
--- a/fs/zonefs/file.c
+++ b/fs/zonefs/file.c
@@ -752,6 +752,44 @@ static ssize_t zonefs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
 	return ret;
 }
 
+static ssize_t zonefs_file_splice_read(struct file *in, loff_t *ppos,
+				       struct pipe_inode_info *pipe,
+				       size_t len, unsigned int flags)
+{
+	struct inode *inode = file_inode(in);
+	struct zonefs_inode_info *zi = ZONEFS_I(inode);
+	struct zonefs_zone *z = zonefs_inode_zone(inode);
+	loff_t isize;
+	ssize_t ret = 0;
+
+	/* Offline zones cannot be read */
+	if (unlikely(IS_IMMUTABLE(inode) && !(inode->i_mode & 0777)))
+		return -EPERM;
+
+	if (*ppos >= z->z_capacity)
+		return 0;
+
+	inode_lock_shared(inode);
+
+	/* Limit read operations to written data */
+	mutex_lock(&zi->i_truncate_mutex);
+	isize = i_size_read(inode);
+	if (*ppos >= isize)
+		len = 0;
+	else
+		len = min_t(loff_t, len, isize - *ppos);
+	mutex_unlock(&zi->i_truncate_mutex);
+
+	if (len > 0) {
+		ret = filemap_splice_read(in, ppos, pipe, len, flags);
+		if (ret == -EIO)
+			zonefs_io_error(inode, false);
+	}
+
+	inode_unlock_shared(inode);
+	return ret;
+}
+
 /*
  * Write open accounting is done only for sequential files.
  */
@@ -896,7 +934,7 @@ const struct file_operations zonefs_file_operations = {
 	.llseek		= zonefs_file_llseek,
 	.read_iter	= zonefs_file_read_iter,
 	.write_iter	= zonefs_file_write_iter,
-	.splice_read	= generic_file_splice_read,
+	.splice_read	= zonefs_file_splice_read,
 	.splice_write	= iter_file_splice_write,
 	.iopoll		= iocb_bio_iopoll,
 };

Re: [PATCH v21 08/30] splice: Make splice from a DAX file use copy_splice_read()

[Christoph Hellwig]

Looks good:

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

Re: [PATCH v21 24/30] xfs: Provide a splice-read stub

[Christoph Hellwig]

Same comment about stub sounding a bit weird as for ext4 and various
patches not in my area, but the changes themselves looks good:

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

Re: [PATCH v21 08/30] splice: Make splice from a DAX file use copy_splice_read()

[Christian Brauner]

On Sat, May 20, 2023 at 01:00:27AM +0100, David Howells wrote:
> Make a read splice from a DAX file go directly to copy_splice_read() to do
> the reading as filemap_splice_read() is unlikely to find any pagecache to
> splice.
> 
> I think this affects only erofs, Ext2, Ext4, fuse and XFS.
> 
> Signed-off-by: David Howells <dhowells@redhat.com>
> cc: Christoph Hellwig <hch@lst.de>
> cc: Al Viro <viro@zeniv.linux.org.uk>
> cc: Jens Axboe <axboe@kernel.dk>
> cc: linux-erofs@lists.ozlabs.org
> cc: linux-ext4@vger.kernel.org
> cc: linux-xfs@vger.kernel.org
> cc: linux-fsdevel@vger.kernel.org
> cc: linux-block@vger.kernel.org
> cc: linux-mm@kvack.org
> ---

Fwiw, O_DIRECT and DAX could've just been folded into one patch imho.
Reviewed-by: Christian Brauner <brauner@kernel.org>

Re: [PATCH v21 08/30] splice: Make splice from a DAX file use copy_splice_read()

[Theodore Ts'o]

On Sat, May 20, 2023 at 01:00:27AM +0100, David Howells wrote:
> Make a read splice from a DAX file go directly to copy_splice_read() to do
> the reading as filemap_splice_read() is unlikely to find any pagecache to
> splice.
> 
> I think this affects only erofs, Ext2, Ext4, fuse and XFS.
> 
> Signed-off-by: David Howells <dhowells@redhat.com>
> cc: Christoph Hellwig <hch@lst.de>
> cc: Al Viro <viro@zeniv.linux.org.uk>
> cc: Jens Axboe <axboe@kernel.dk>
> cc: linux-erofs@lists.ozlabs.org
> cc: linux-ext4@vger.kernel.org

Reviewed-by: Theodore Ts'o <tytso@mit.edu>

Re: [PATCH v21 08/30] splice: Make splice from a DAX file use copy_splice_read()

[Gao Xiang]

On 2023/5/20 17:00, David Howells wrote:
> Make a read splice from a DAX file go directly to copy_splice_read() to do
> the reading as filemap_splice_read() is unlikely to find any pagecache to
> splice.
> 
> I think this affects only erofs, Ext2, Ext4, fuse and XFS.
> 
> Signed-off-by: David Howells <dhowells@redhat.com>
> cc: Christoph Hellwig <hch@lst.de>
> cc: Al Viro <viro@zeniv.linux.org.uk>
> cc: Jens Axboe <axboe@kernel.dk>
> cc: linux-erofs@lists.ozlabs.org
> cc: linux-ext4@vger.kernel.org
> cc: linux-xfs@vger.kernel.org
> cc: linux-fsdevel@vger.kernel.org
> cc: linux-block@vger.kernel.org
> cc: linux-mm@kvack.org

Reviewed-by: Gao Xiang <hsiangkao@linux.alibaba.com>

Thanks,
Gao Xiang