Merge branch 'splice-2.6.23' of git://git.kernel.dk/data/git/linux-2.6-block

* 'splice-2.6.23' of git://git.kernel.dk/data/git/linux-2.6-block:
  pipe: add documentation and comments
  pipe: change the ->pin() operation to ->confirm()
  Remove remnants of sendfile()
  xip sendfile removal
  splice: completely document external interface with kerneldoc
  sendfile: remove bad_sendfile() from bad_file_ops
  shmem: convert to using splice instead of sendfile()
  relay: use splice_to_pipe() instead of open-coding the pipe loop
  pipe: allow passing around of ops private pointer
  splice: divorce the splice structure/function definitions from the pipe header
  splice: relay support
  sendfile: convert nfsd to splice_direct_to_actor()
  sendfile: convert nfs to using splice_read()
  loop: convert to using splice_direct_to_actor() instead of sendfile()
  splice: add void cookie to the actor data
  sendfile: kill generic_file_sendfile()
  sendfile: remove .sendfile from filesystems that use generic_file_sendfile()
  sys_sendfile: switch to using ->splice_read, if available
  vmsplice: add vmsplice-to-user support
  splice: abstract out actor data

Documentation/DocBook/kernel-api.tmpl

@@ -643,4 +643,15 @@ X!Idrivers/video/console/fonts.c
 !Edrivers/spi/spi.c
   </chapter>
 
+  <chapter id="splice">
+      <title>splice API</title>
+  <para>)
+	splice is a method for moving blocks of data around inside the
+	kernel, without continually transferring it between the kernel
+	and user space.
+  </para>
+!Iinclude/linux/splice.h
+!Ffs/splice.c
+  </chapter>
+
 </book>

drivers/block/loop.c

@@ -74,6 +74,7 @@
 #include <linux/highmem.h>
 #include <linux/gfp.h>
 #include <linux/kthread.h>
+#include <linux/splice.h>
 
 #include <asm/uaccess.h>
 
@@ -401,50 +402,73 @@ struct lo_read_data {
 };
 
 static int
-lo_read_actor(read_descriptor_t *desc, struct page *page,
-	      unsigned long offset, unsigned long size)
+lo_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
+		struct splice_desc *sd)
 {
-	unsigned long count = desc->count;
-	struct lo_read_data *p = desc->arg.data;
+	struct lo_read_data *p = sd->u.data;
 	struct loop_device *lo = p->lo;
+	struct page *page = buf->page;
 	sector_t IV;
+	size_t size;
+	int ret;
 
-	IV = ((sector_t) page->index << (PAGE_CACHE_SHIFT - 9))+(offset >> 9);
+	ret = buf->ops->confirm(pipe, buf);
+	if (unlikely(ret))
+		return ret;
 
-	if (size > count)
-		size = count;
+	IV = ((sector_t) page->index << (PAGE_CACHE_SHIFT - 9)) +
+							(buf->offset >> 9);
+	size = sd->len;
+	if (size > p->bsize)
+		size = p->bsize;
 
-	if (lo_do_transfer(lo, READ, page, offset, p->page, p->offset, size, IV)) {
-		size = 0;
+	if (lo_do_transfer(lo, READ, page, buf->offset, p->page, p->offset, size, IV)) {
 		printk(KERN_ERR "loop: transfer error block %ld\n",
 		       page->index);
-		desc->error = -EINVAL;
+		size = -EINVAL;
 	}
 
 	flush_dcache_page(p->page);
 
-	desc->count = count - size;
-	desc->written += size;
-	p->offset += size;
+	if (size > 0)
+		p->offset += size;
+
 	return size;
 }
 
+static int
+lo_direct_splice_actor(struct pipe_inode_info *pipe, struct splice_desc *sd)
+{
+	return __splice_from_pipe(pipe, sd, lo_splice_actor);
+}
+
 static int
 do_lo_receive(struct loop_device *lo,
 	      struct bio_vec *bvec, int bsize, loff_t pos)
 {
 	struct lo_read_data cookie;
+	struct splice_desc sd;
 	struct file *file;
-	int retval;
+	long retval;
 
 	cookie.lo = lo;
 	cookie.page = bvec->bv_page;
 	cookie.offset = bvec->bv_offset;
 	cookie.bsize = bsize;
+
+	sd.len = 0;
+	sd.total_len = bvec->bv_len;
+	sd.flags = 0;
+	sd.pos = pos;
+	sd.u.data = &cookie;
+
 	file = lo->lo_backing_file;
-	retval = file->f_op->sendfile(file, &pos, bvec->bv_len,
-			lo_read_actor, &cookie);
-	return (retval < 0)? retval: 0;
+	retval = splice_direct_to_actor(file, &sd, lo_direct_splice_actor);
+
+	if (retval < 0)
+		return retval;
+
+	return 0;
 }
 
 static int
@@ -679,8 +703,8 @@ static int loop_change_fd(struct loop_device *lo, struct file *lo_file,
 	if (!S_ISREG(inode->i_mode) && !S_ISBLK(inode->i_mode))
 		goto out_putf;
 
-	/* new backing store needs to support loop (eg sendfile) */
-	if (!inode->i_fop->sendfile)
+	/* new backing store needs to support loop (eg splice_read) */
+	if (!inode->i_fop->splice_read)
 		goto out_putf;
 
 	/* size of the new backing store needs to be the same */
@@ -760,7 +784,7 @@ static int loop_set_fd(struct loop_device *lo, struct file *lo_file,
 		 * If we can't read - sorry. If we only can't write - well,
 		 * it's going to be read-only.
 		 */
-		if (!file->f_op->sendfile)
+		if (!file->f_op->splice_read)
 			goto out_putf;
 		if (aops->prepare_write && aops->commit_write)
 			lo_flags |= LO_FLAGS_USE_AOPS;

drivers/char/mem.c

@@ -24,7 +24,7 @@
 #include <linux/crash_dump.h>
 #include <linux/backing-dev.h>
 #include <linux/bootmem.h>
-#include <linux/pipe_fs_i.h>
+#include <linux/splice.h>
 #include <linux/pfn.h>
 
 #include <asm/uaccess.h>

fs/adfs/file.c

@@ -33,7 +33,7 @@ const struct file_operations adfs_file_operations = {
 	.fsync		= file_fsync,
 	.write		= do_sync_write,
 	.aio_write	= generic_file_aio_write,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 const struct inode_operations adfs_file_inode_operations = {

fs/affs/file.c

@@ -35,7 +35,7 @@ const struct file_operations affs_file_operations = {
 	.open		= affs_file_open,
 	.release	= affs_file_release,
 	.fsync		= file_fsync,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 const struct inode_operations affs_file_inode_operations = {

fs/afs/file.c

@@ -32,7 +32,7 @@ const struct file_operations afs_file_operations = {
 	.aio_read	= generic_file_aio_read,
 	.aio_write	= afs_file_write,
 	.mmap		= generic_file_readonly_mmap,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 	.fsync		= afs_fsync,
 };
 

fs/bad_inode.c

@@ -114,12 +114,6 @@ static int bad_file_lock(struct file *file, int cmd, struct file_lock *fl)
 	return -EIO;
 }
 
-static ssize_t bad_file_sendfile(struct file *in_file, loff_t *ppos,
-			size_t count, read_actor_t actor, void *target)
-{
-	return -EIO;
-}
-
 static ssize_t bad_file_sendpage(struct file *file, struct page *page,
 			int off, size_t len, loff_t *pos, int more)
 {
@@ -182,7 +176,6 @@ static const struct file_operations bad_file_ops =
 	.aio_fsync	= bad_file_aio_fsync,
 	.fasync		= bad_file_fasync,
 	.lock		= bad_file_lock,
-	.sendfile	= bad_file_sendfile,
 	.sendpage	= bad_file_sendpage,
 	.get_unmapped_area = bad_file_get_unmapped_area,
 	.check_flags	= bad_file_check_flags,

fs/bfs/file.c

@@ -24,7 +24,7 @@ const struct file_operations bfs_file_operations = {
 	.write		= do_sync_write,
 	.aio_write	= generic_file_aio_write,
 	.mmap		= generic_file_mmap,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 static int bfs_move_block(unsigned long from, unsigned long to, struct super_block *sb)

fs/block_dev.c

@@ -1346,7 +1346,6 @@ const struct file_operations def_blk_fops = {
 #ifdef CONFIG_COMPAT
 	.compat_ioctl	= compat_blkdev_ioctl,
 #endif
-	.sendfile	= generic_file_sendfile,
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= generic_file_splice_write,
 };

fs/cifs/cifsfs.c

@@ -616,7 +616,7 @@ const struct file_operations cifs_file_ops = {
 	.fsync = cifs_fsync,
 	.flush = cifs_flush,
 	.mmap  = cifs_file_mmap,
-	.sendfile = generic_file_sendfile,
+	.splice_read = generic_file_splice_read,
 	.llseek = cifs_llseek,
 #ifdef CONFIG_CIFS_POSIX
 	.ioctl	= cifs_ioctl,
@@ -637,7 +637,7 @@ const struct file_operations cifs_file_direct_ops = {
 	.lock = cifs_lock,
 	.fsync = cifs_fsync,
 	.flush = cifs_flush,
-	.sendfile = generic_file_sendfile, /* BB removeme BB */
+	.splice_read = generic_file_splice_read,
 #ifdef CONFIG_CIFS_POSIX
 	.ioctl  = cifs_ioctl,
 #endif /* CONFIG_CIFS_POSIX */
@@ -656,7 +656,7 @@ const struct file_operations cifs_file_nobrl_ops = {
 	.fsync = cifs_fsync,
 	.flush = cifs_flush,
 	.mmap  = cifs_file_mmap,
-	.sendfile = generic_file_sendfile,
+	.splice_read = generic_file_splice_read,
 	.llseek = cifs_llseek,
 #ifdef CONFIG_CIFS_POSIX
 	.ioctl	= cifs_ioctl,
@@ -676,7 +676,7 @@ const struct file_operations cifs_file_direct_nobrl_ops = {
 	.release = cifs_close,
 	.fsync = cifs_fsync,
 	.flush = cifs_flush,
-	.sendfile = generic_file_sendfile, /* BB removeme BB */
+	.splice_read = generic_file_splice_read,
 #ifdef CONFIG_CIFS_POSIX
 	.ioctl  = cifs_ioctl,
 #endif /* CONFIG_CIFS_POSIX */

fs/coda/file.c

@@ -47,8 +47,9 @@ coda_file_read(struct file *coda_file, char __user *buf, size_t count, loff_t *p
 }
 
 static ssize_t
-coda_file_sendfile(struct file *coda_file, loff_t *ppos, size_t count,
-		   read_actor_t actor, void *target)
+coda_file_splice_read(struct file *coda_file, loff_t *ppos,
+		      struct pipe_inode_info *pipe, size_t count,
+		      unsigned int flags)
 {
 	struct coda_file_info *cfi;
 	struct file *host_file;
@@ -57,10 +58,10 @@ coda_file_sendfile(struct file *coda_file, loff_t *ppos, size_t count,
 	BUG_ON(!cfi || cfi->cfi_magic != CODA_MAGIC);
 	host_file = cfi->cfi_container;
 
-	if (!host_file->f_op || !host_file->f_op->sendfile)
+	if (!host_file->f_op || !host_file->f_op->splice_read)
 		return -EINVAL;
 
-	return host_file->f_op->sendfile(host_file, ppos, count, actor, target);
+	return host_file->f_op->splice_read(host_file, ppos, pipe, count,flags);
 }
 
 static ssize_t
@@ -295,6 +296,6 @@ const struct file_operations coda_file_operations = {
 	.flush		= coda_flush,
 	.release	= coda_release,
 	.fsync		= coda_fsync,
-	.sendfile	= coda_file_sendfile,
+	.splice_read	= coda_file_splice_read,
 };
 

fs/ecryptfs/file.c

@@ -338,16 +338,17 @@ static int ecryptfs_fasync(int fd, struct file *file, int flag)
 	return rc;
 }
 
-static ssize_t ecryptfs_sendfile(struct file *file, loff_t * ppos,
-				 size_t count, read_actor_t actor, void *target)
+static ssize_t ecryptfs_splice_read(struct file *file, loff_t * ppos,
+				    struct pipe_inode_info *pipe, size_t count,
+				    unsigned int flags)
 {
 	struct file *lower_file = NULL;
 	int rc = -EINVAL;
 
 	lower_file = ecryptfs_file_to_lower(file);
-	if (lower_file->f_op && lower_file->f_op->sendfile)
-		rc = lower_file->f_op->sendfile(lower_file, ppos, count,
-						actor, target);
+	if (lower_file->f_op && lower_file->f_op->splice_read)
+		rc = lower_file->f_op->splice_read(lower_file, ppos, pipe,
+						count, flags);
 
 	return rc;
 }
@@ -364,7 +365,7 @@ const struct file_operations ecryptfs_dir_fops = {
 	.release = ecryptfs_release,
 	.fsync = ecryptfs_fsync,
 	.fasync = ecryptfs_fasync,
-	.sendfile = ecryptfs_sendfile,
+	.splice_read = ecryptfs_splice_read,
 };
 
 const struct file_operations ecryptfs_main_fops = {
@@ -381,7 +382,7 @@ const struct file_operations ecryptfs_main_fops = {
 	.release = ecryptfs_release,
 	.fsync = ecryptfs_fsync,
 	.fasync = ecryptfs_fasync,
-	.sendfile = ecryptfs_sendfile,
+	.splice_read = ecryptfs_splice_read,
 };
 
 static int

fs/ext2/file.c

@@ -53,7 +53,6 @@ const struct file_operations ext2_file_operations = {
 	.open		= generic_file_open,
 	.release	= ext2_release_file,
 	.fsync		= ext2_sync_file,
-	.sendfile	= generic_file_sendfile,
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= generic_file_splice_write,
 };
@@ -71,7 +70,6 @@ const struct file_operations ext2_xip_file_operations = {
 	.open		= generic_file_open,
 	.release	= ext2_release_file,
 	.fsync		= ext2_sync_file,
-	.sendfile	= xip_file_sendfile,
 };
 #endif
 

fs/ext3/file.c

@@ -120,7 +120,6 @@ const struct file_operations ext3_file_operations = {
 	.open		= generic_file_open,
 	.release	= ext3_release_file,
 	.fsync		= ext3_sync_file,
-	.sendfile	= generic_file_sendfile,
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= generic_file_splice_write,
 };

fs/ext4/file.c

@@ -120,7 +120,6 @@ const struct file_operations ext4_file_operations = {
 	.open		= generic_file_open,
 	.release	= ext4_release_file,
 	.fsync		= ext4_sync_file,
-	.sendfile	= generic_file_sendfile,
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= generic_file_splice_write,
 };

fs/fat/file.c

@@ -134,7 +134,7 @@ const struct file_operations fat_file_operations = {
 	.release	= fat_file_release,
 	.ioctl		= fat_generic_ioctl,
 	.fsync		= file_fsync,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 static int fat_cont_expand(struct inode *inode, loff_t size)

fs/fuse/file.c

@@ -802,7 +802,7 @@ static const struct file_operations fuse_file_operations = {
 	.release	= fuse_release,
 	.fsync		= fuse_fsync,
 	.lock		= fuse_file_lock,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 static const struct file_operations fuse_direct_io_file_operations = {
@@ -814,7 +814,7 @@ static const struct file_operations fuse_direct_io_file_operations = {
 	.release	= fuse_release,
 	.fsync		= fuse_fsync,
 	.lock		= fuse_file_lock,
-	/* no mmap and sendfile */
+	/* no mmap and splice_read */
 };
 
 static const struct address_space_operations fuse_file_aops  = {

fs/gfs2/ops_file.c

@@ -635,7 +635,6 @@ const struct file_operations gfs2_file_fops = {
 	.release	= gfs2_close,
 	.fsync		= gfs2_fsync,
 	.lock		= gfs2_lock,
-	.sendfile	= generic_file_sendfile,
 	.flock		= gfs2_flock,
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= generic_file_splice_write,

fs/hfs/inode.c

@@ -607,7 +607,7 @@ static const struct file_operations hfs_file_operations = {
 	.write		= do_sync_write,
 	.aio_write	= generic_file_aio_write,
 	.mmap		= generic_file_mmap,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 	.fsync		= file_fsync,
 	.open		= hfs_file_open,
 	.release	= hfs_file_release,

fs/hfsplus/inode.c

@@ -288,7 +288,7 @@ static const struct file_operations hfsplus_file_operations = {
 	.write		= do_sync_write,
 	.aio_write	= generic_file_aio_write,
 	.mmap		= generic_file_mmap,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 	.fsync		= file_fsync,
 	.open		= hfsplus_file_open,
 	.release	= hfsplus_file_release,

fs/hostfs/hostfs_kern.c

@@ -390,7 +390,7 @@ int hostfs_fsync(struct file *file, struct dentry *dentry, int datasync)
 static const struct file_operations hostfs_file_fops = {
 	.llseek		= generic_file_llseek,
 	.read		= do_sync_read,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 	.aio_read	= generic_file_aio_read,
 	.aio_write	= generic_file_aio_write,
 	.write		= do_sync_write,

fs/hpfs/file.c

@@ -129,7 +129,7 @@ const struct file_operations hpfs_file_ops =
 	.mmap		= generic_file_mmap,
 	.release	= hpfs_file_release,
 	.fsync		= hpfs_file_fsync,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 const struct inode_operations hpfs_file_iops =

fs/jffs2/file.c

@@ -47,7 +47,7 @@ const struct file_operations jffs2_file_operations =
 	.ioctl =	jffs2_ioctl,
 	.mmap =		generic_file_readonly_mmap,
 	.fsync =	jffs2_fsync,
-	.sendfile =	generic_file_sendfile
+	.splice_read =	generic_file_splice_read,
 };
 
 /* jffs2_file_inode_operations */

fs/jfs/file.c

@@ -108,7 +108,6 @@ const struct file_operations jfs_file_operations = {
 	.aio_read	= generic_file_aio_read,
 	.aio_write	= generic_file_aio_write,
 	.mmap		= generic_file_mmap,
-	.sendfile	= generic_file_sendfile,
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= generic_file_splice_write,
 	.fsync		= jfs_fsync,

fs/minix/file.c

@@ -23,7 +23,7 @@ const struct file_operations minix_file_operations = {
 	.aio_write	= generic_file_aio_write,
 	.mmap		= generic_file_mmap,
 	.fsync		= minix_sync_file,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 const struct inode_operations minix_file_inode_operations = {

fs/nfs/file.c

@@ -41,7 +41,9 @@ static int nfs_file_open(struct inode *, struct file *);
 static int nfs_file_release(struct inode *, struct file *);
 static loff_t nfs_file_llseek(struct file *file, loff_t offset, int origin);
 static int  nfs_file_mmap(struct file *, struct vm_area_struct *);
-static ssize_t nfs_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
+static ssize_t nfs_file_splice_read(struct file *filp, loff_t *ppos,
+					struct pipe_inode_info *pipe,
+					size_t count, unsigned int flags);
 static ssize_t nfs_file_read(struct kiocb *, const struct iovec *iov,
 				unsigned long nr_segs, loff_t pos);
 static ssize_t nfs_file_write(struct kiocb *, const struct iovec *iov,
@@ -65,7 +67,7 @@ const struct file_operations nfs_file_operations = {
 	.fsync		= nfs_fsync,
 	.lock		= nfs_lock,
 	.flock		= nfs_flock,
-	.sendfile	= nfs_file_sendfile,
+	.splice_read	= nfs_file_splice_read,
 	.check_flags	= nfs_check_flags,
 };
 
@@ -224,20 +226,21 @@ nfs_file_read(struct kiocb *iocb, const struct iovec *iov,
 }
 
 static ssize_t
-nfs_file_sendfile(struct file *filp, loff_t *ppos, size_t count,
-		read_actor_t actor, void *target)
+nfs_file_splice_read(struct file *filp, loff_t *ppos,
+		     struct pipe_inode_info *pipe, size_t count,
+		     unsigned int flags)
 {
 	struct dentry *dentry = filp->f_path.dentry;
 	struct inode *inode = dentry->d_inode;
 	ssize_t res;
 
-	dfprintk(VFS, "nfs: sendfile(%s/%s, %lu@%Lu)\n",
+	dfprintk(VFS, "nfs: splice_read(%s/%s, %lu@%Lu)\n",
 		dentry->d_parent->d_name.name, dentry->d_name.name,
 		(unsigned long) count, (unsigned long long) *ppos);
 
 	res = nfs_revalidate_mapping(inode, filp->f_mapping);
 	if (!res)
-		res = generic_file_sendfile(filp, ppos, count, actor, target);
+		res = generic_file_splice_read(filp, ppos, pipe, count, flags);
 	return res;
 }
 

fs/nfsd/vfs.c

@@ -23,7 +23,7 @@
 #include <linux/file.h>
 #include <linux/mount.h>
 #include <linux/major.h>
-#include <linux/ext2_fs.h>
+#include <linux/splice.h>
 #include <linux/proc_fs.h>
 #include <linux/stat.h>
 #include <linux/fcntl.h>
@@ -801,26 +801,32 @@ nfsd_get_raparms(dev_t dev, ino_t ino)
 }
 
 /*
- * Grab and keep cached pages assosiated with a file in the svc_rqst
- * so that they can be passed to the netowork sendmsg/sendpage routines
- * directrly. They will be released after the sending has completed.
+ * Grab and keep cached pages associated with a file in the svc_rqst
+ * so that they can be passed to the network sendmsg/sendpage routines
+ * directly. They will be released after the sending has completed.
  */
 static int
-nfsd_read_actor(read_descriptor_t *desc, struct page *page, unsigned long offset , unsigned long size)
+nfsd_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
+		  struct splice_desc *sd)
 {
-	unsigned long count = desc->count;
-	struct svc_rqst *rqstp = desc->arg.data;
+	struct svc_rqst *rqstp = sd->u.data;
 	struct page **pp = rqstp->rq_respages + rqstp->rq_resused;
+	struct page *page = buf->page;
+	size_t size;
+	int ret;
+
+	ret = buf->ops->confirm(pipe, buf);
+	if (unlikely(ret))
+		return ret;
 
-	if (size > count)
-		size = count;
+	size = sd->len;
 
 	if (rqstp->rq_res.page_len == 0) {
 		get_page(page);
 		put_page(*pp);
 		*pp = page;
 		rqstp->rq_resused++;
-		rqstp->rq_res.page_base = offset;
+		rqstp->rq_res.page_base = buf->offset;
 		rqstp->rq_res.page_len = size;
 	} else if (page != pp[-1]) {
 		get_page(page);
@@ -832,11 +838,15 @@ nfsd_read_actor(read_descriptor_t *desc, struct page *page, unsigned long offset
 	} else
 		rqstp->rq_res.page_len += size;
 
-	desc->count = count - size;
-	desc->written += size;
 	return size;
 }
 
+static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
+				    struct splice_desc *sd)
+{
+	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
+}
+
 static __be32
 nfsd_vfs_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
               loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
@@ -861,10 +871,15 @@ nfsd_vfs_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
 	if (ra && ra->p_set)
 		file->f_ra = ra->p_ra;
 
-	if (file->f_op->sendfile && rqstp->rq_sendfile_ok) {
-		rqstp->rq_resused = 1;
-		host_err = file->f_op->sendfile(file, &offset, *count,
-						 nfsd_read_actor, rqstp);
+	if (file->f_op->splice_read && rqstp->rq_splice_ok) {
+		struct splice_desc sd = {
+			.len		= 0,
+			.total_len	= *count,
+			.pos		= offset,
+			.u.data		= rqstp,
+		};
+
+		host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
 	} else {
 		oldfs = get_fs();
 		set_fs(KERNEL_DS);

fs/ntfs/file.c

@@ -2276,7 +2276,7 @@ const struct file_operations ntfs_file_ops = {
 						    mounted filesystem. */
 	.mmap		= generic_file_mmap,	 /* Mmap file. */
 	.open		= ntfs_file_open,	 /* Open file. */
-	.sendfile	= generic_file_sendfile, /* Zero-copy data send with
+	.splice_read	= generic_file_splice_read /* Zero-copy data send with
 						    the data source being on
 						    the ntfs partition.  We do
 						    not need to care about the

fs/ocfs2/file.c

@@ -31,7 +31,7 @@
 #include <linux/pagemap.h>
 #include <linux/uio.h>
 #include <linux/sched.h>
-#include <linux/pipe_fs_i.h>
+#include <linux/splice.h>
 #include <linux/mount.h>
 #include <linux/writeback.h>
 
@@ -1583,7 +1583,7 @@ static int ocfs2_splice_write_actor(struct pipe_inode_info *pipe,
 	ssize_t copied = 0;
 	struct ocfs2_splice_write_priv sp;
 
-	ret = buf->ops->pin(pipe, buf);
+	ret = buf->ops->confirm(pipe, buf);
 	if (ret)
 		goto out;
 
@@ -1604,7 +1604,7 @@ static int ocfs2_splice_write_actor(struct pipe_inode_info *pipe,
 		 * might enter ocfs2_buffered_write_cluster() more
 		 * than once, so keep track of our progress here.
 		 */
-		copied = ocfs2_buffered_write_cluster(sd->file,
+		copied = ocfs2_buffered_write_cluster(sd->u.file,
 						      (loff_t)sd->pos + total,
 						      count,
 						      ocfs2_map_and_write_splice_data,
@@ -1636,9 +1636,14 @@ static ssize_t __ocfs2_file_splice_write(struct pipe_inode_info *pipe,
 	int ret, err;
 	struct address_space *mapping = out->f_mapping;
 	struct inode *inode = mapping->host;
-
-	ret = __splice_from_pipe(pipe, out, ppos, len, flags,
-				 ocfs2_splice_write_actor);
+	struct splice_desc sd = {
+		.total_len = len,
+		.flags = flags,
+		.pos = *ppos,
+		.u.file = out,
+	};
+
+	ret = __splice_from_pipe(pipe, &sd, ocfs2_splice_write_actor);
 	if (ret > 0) {
 		*ppos += ret;
 
@@ -1817,7 +1822,6 @@ const struct inode_operations ocfs2_special_file_iops = {
 const struct file_operations ocfs2_fops = {
 	.read		= do_sync_read,
 	.write		= do_sync_write,
-	.sendfile	= generic_file_sendfile,
 	.mmap		= ocfs2_mmap,
 	.fsync		= ocfs2_sync_file,
 	.release	= ocfs2_file_release,

fs/pipe.c

@@ -164,6 +164,20 @@ static void anon_pipe_buf_release(struct pipe_inode_info *pipe,
 		page_cache_release(page);
 }
 
+/**
+ * generic_pipe_buf_map - virtually map a pipe buffer
+ * @pipe:	the pipe that the buffer belongs to
+ * @buf:	the buffer that should be mapped
+ * @atomic:	whether to use an atomic map
+ *
+ * Description:
+ *	This function returns a kernel virtual address mapping for the
+ *	passed in @pipe_buffer. If @atomic is set, an atomic map is provided
+ *	and the caller has to be careful not to fault before calling
+ *	the unmap function.
+ *
+ *	Note that this function occupies KM_USER0 if @atomic != 0.
+ */
 void *generic_pipe_buf_map(struct pipe_inode_info *pipe,
 			   struct pipe_buffer *buf, int atomic)
 {
@@ -175,6 +189,15 @@ void *generic_pipe_buf_map(struct pipe_inode_info *pipe,
 	return kmap(buf->page);
 }
 
+/**
+ * generic_pipe_buf_unmap - unmap a previously mapped pipe buffer
+ * @pipe:	the pipe that the buffer belongs to
+ * @buf:	the buffer that should be unmapped
+ * @map_data:	the data that the mapping function returned
+ *
+ * Description:
+ *	This function undoes the mapping that ->map() provided.
+ */
 void generic_pipe_buf_unmap(struct pipe_inode_info *pipe,
 			    struct pipe_buffer *buf, void *map_data)
 {
@@ -185,11 +208,28 @@ void generic_pipe_buf_unmap(struct pipe_inode_info *pipe,
 		kunmap(buf->page);
 }
 
+/**
+ * generic_pipe_buf_steal - attempt to take ownership of a @pipe_buffer
+ * @pipe:	the pipe that the buffer belongs to
+ * @buf:	the buffer to attempt to steal
+ *
+ * Description:
+ *	This function attempts to steal the @struct page attached to
+ *	@buf. If successful, this function returns 0 and returns with
+ *	the page locked. The caller may then reuse the page for whatever
+ *	he wishes, the typical use is insertion into a different file
+ *	page cache.
+ */
 int generic_pipe_buf_steal(struct pipe_inode_info *pipe,
 			   struct pipe_buffer *buf)
 {
 	struct page *page = buf->page;
 
+	/*
+	 * A reference of one is golden, that means that the owner of this
+	 * page is the only one holding a reference to it. lock the page
+	 * and return OK.
+	 */
 	if (page_count(page) == 1) {
 		lock_page(page);
 		return 0;
@@ -198,12 +238,32 @@ int generic_pipe_buf_steal(struct pipe_inode_info *pipe,
 	return 1;
 }
 
-void generic_pipe_buf_get(struct pipe_inode_info *info, struct pipe_buffer *buf)
+/**
+ * generic_pipe_buf_get - get a reference to a @struct pipe_buffer
+ * @pipe:	the pipe that the buffer belongs to
+ * @buf:	the buffer to get a reference to
+ *
+ * Description:
+ *	This function grabs an extra reference to @buf. It's used in
+ *	in the tee() system call, when we duplicate the buffers in one
+ *	pipe into another.
+ */
+void generic_pipe_buf_get(struct pipe_inode_info *pipe, struct pipe_buffer *buf)
 {
 	page_cache_get(buf->page);
 }
 
-int generic_pipe_buf_pin(struct pipe_inode_info *info, struct pipe_buffer *buf)
+/**
+ * generic_pipe_buf_confirm - verify contents of the pipe buffer
+ * @pipe:	the pipe that the buffer belongs to
+ * @buf:	the buffer to confirm
+ *
+ * Description:
+ *	This function does nothing, because the generic pipe code uses
+ *	pages that are always good when inserted into the pipe.
+ */
+int generic_pipe_buf_confirm(struct pipe_inode_info *info,
+			     struct pipe_buffer *buf)
 {
 	return 0;
 }
@@ -212,7 +272,7 @@ static const struct pipe_buf_operations anon_pipe_buf_ops = {
 	.can_merge = 1,
 	.map = generic_pipe_buf_map,
 	.unmap = generic_pipe_buf_unmap,
-	.pin = generic_pipe_buf_pin,
+	.confirm = generic_pipe_buf_confirm,
 	.release = anon_pipe_buf_release,
 	.steal = generic_pipe_buf_steal,
 	.get = generic_pipe_buf_get,
@@ -252,7 +312,7 @@ pipe_read(struct kiocb *iocb, const struct iovec *_iov,
 			if (chars > total_len)
 				chars = total_len;
 
-			error = ops->pin(pipe, buf);
+			error = ops->confirm(pipe, buf);
 			if (error) {
 				if (!ret)
 					error = ret;
@@ -373,7 +433,7 @@ pipe_write(struct kiocb *iocb, const struct iovec *_iov,
 			int error, atomic = 1;
 			void *addr;
 
-			error = ops->pin(pipe, buf);
+			error = ops->confirm(pipe, buf);
 			if (error)
 				goto out;
 

fs/qnx4/file.c

@@ -25,7 +25,7 @@ const struct file_operations qnx4_file_operations =
 	.read		= do_sync_read,
 	.aio_read	= generic_file_aio_read,
 	.mmap		= generic_file_mmap,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 #ifdef CONFIG_QNX4FS_RW
 	.write		= do_sync_write,
 	.aio_write	= generic_file_aio_write,

fs/ramfs/file-mmu.c

@@ -41,7 +41,7 @@ const struct file_operations ramfs_file_operations = {
 	.aio_write	= generic_file_aio_write,
 	.mmap		= generic_file_mmap,
 	.fsync		= simple_sync_file,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 	.llseek		= generic_file_llseek,
 };
 

fs/ramfs/file-nommu.c

@@ -42,7 +42,7 @@ const struct file_operations ramfs_file_operations = {
 	.write			= do_sync_write,
 	.aio_write		= generic_file_aio_write,
 	.fsync			= simple_sync_file,
-	.sendfile		= generic_file_sendfile,
+	.splice_read		= generic_file_splice_read,
 	.llseek			= generic_file_llseek,
 };
 

fs/read_write.c

@@ -15,6 +15,7 @@
 #include <linux/module.h>
 #include <linux/syscalls.h>
 #include <linux/pagemap.h>
+#include <linux/splice.h>
 #include "read_write.h"
 
 #include <asm/uaccess.h>
@@ -25,7 +26,7 @@ const struct file_operations generic_ro_fops = {
 	.read		= do_sync_read,
 	.aio_read	= generic_file_aio_read,
 	.mmap		= generic_file_readonly_mmap,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 EXPORT_SYMBOL(generic_ro_fops);
@@ -708,7 +709,7 @@ static ssize_t do_sendfile(int out_fd, int in_fd, loff_t *ppos,
 	struct inode * in_inode, * out_inode;
 	loff_t pos;
 	ssize_t retval;
-	int fput_needed_in, fput_needed_out;
+	int fput_needed_in, fput_needed_out, fl;
 
 	/*
 	 * Get input file, and verify that it is ok..
@@ -723,7 +724,7 @@ static ssize_t do_sendfile(int out_fd, int in_fd, loff_t *ppos,
 	in_inode = in_file->f_path.dentry->d_inode;
 	if (!in_inode)
 		goto fput_in;
-	if (!in_file->f_op || !in_file->f_op->sendfile)
+	if (!in_file->f_op || !in_file->f_op->splice_read)
 		goto fput_in;
 	retval = -ESPIPE;
 	if (!ppos)
@@ -776,7 +777,18 @@ static ssize_t do_sendfile(int out_fd, int in_fd, loff_t *ppos,
 		count = max - pos;
 	}
 
-	retval = in_file->f_op->sendfile(in_file, ppos, count, file_send_actor, out_file);
+	fl = 0;
+#if 0
+	/*
+	 * We need to debate whether we can enable this or not. The
+	 * man page documents EAGAIN return for the output at least,
+	 * and the application is arguably buggy if it doesn't expect
+	 * EAGAIN on a non-blocking file descriptor.
+	 */
+	if (in_file->f_flags & O_NONBLOCK)
+		fl = SPLICE_F_NONBLOCK;
+#endif
+	retval = do_splice_direct(in_file, ppos, out_file, count, fl);
 
 	if (retval > 0) {
 		add_rchar(current, retval);

fs/reiserfs/file.c

@@ -1531,7 +1531,6 @@ const struct file_operations reiserfs_file_operations = {
 	.open = generic_file_open,
 	.release = reiserfs_file_release,
 	.fsync = reiserfs_sync_file,
-	.sendfile = generic_file_sendfile,
 	.aio_read = generic_file_aio_read,
 	.aio_write = generic_file_aio_write,
 	.splice_read = generic_file_splice_read,

fs/smbfs/file.c

@@ -262,8 +262,9 @@ smb_file_mmap(struct file * file, struct vm_area_struct * vma)
 }
 
 static ssize_t
-smb_file_sendfile(struct file *file, loff_t *ppos,
-		  size_t count, read_actor_t actor, void *target)
+smb_file_splice_read(struct file *file, loff_t *ppos,
+		     struct pipe_inode_info *pipe, size_t count,
+		     unsigned int flags)
 {
 	struct dentry *dentry = file->f_path.dentry;
 	ssize_t status;
@@ -277,7 +278,7 @@ smb_file_sendfile(struct file *file, loff_t *ppos,
 			 DENTRY_PATH(dentry), status);
 		goto out;
 	}
-	status = generic_file_sendfile(file, ppos, count, actor, target);
+	status = generic_file_splice_read(file, ppos, pipe, count, flags);
 out:
 	return status;
 }
@@ -416,7 +417,7 @@ const struct file_operations smb_file_operations =
 	.open		= smb_file_open,
 	.release	= smb_file_release,
 	.fsync		= smb_fsync,
-	.sendfile	= smb_file_sendfile,
+	.splice_read	= smb_file_splice_read,
 };
 
 const struct inode_operations smb_file_inode_operations =

fs/splice.c

@@ -20,7 +20,7 @@
 #include <linux/fs.h>
 #include <linux/file.h>
 #include <linux/pagemap.h>
-#include <linux/pipe_fs_i.h>
+#include <linux/splice.h>
 #include <linux/mm_inline.h>
 #include <linux/swap.h>
 #include <linux/writeback.h>
@@ -29,22 +29,6 @@
 #include <linux/syscalls.h>
 #include <linux/uio.h>
 
-struct partial_page {
-	unsigned int offset;
-	unsigned int len;
-};
-
-/*
- * Passed to splice_to_pipe
- */
-struct splice_pipe_desc {
-	struct page **pages;		/* page map */
-	struct partial_page *partial;	/* pages[] may not be contig */
-	int nr_pages;			/* number of pages in map */
-	unsigned int flags;		/* splice flags */
-	const struct pipe_buf_operations *ops;/* ops associated with output pipe */
-};
-
 /*
  * Attempt to steal a page from a pipe buffer. This should perhaps go into
  * a vm helper function, it's already simplified quite a bit by the
@@ -101,8 +85,12 @@ static void page_cache_pipe_buf_release(struct pipe_inode_info *pipe,
 	buf->flags &= ~PIPE_BUF_FLAG_LRU;
 }
 
-static int page_cache_pipe_buf_pin(struct pipe_inode_info *pipe,
-				   struct pipe_buffer *buf)
+/*
+ * Check whether the contents of buf is OK to access. Since the content
+ * is a page cache page, IO may be in flight.
+ */
+static int page_cache_pipe_buf_confirm(struct pipe_inode_info *pipe,
+				       struct pipe_buffer *buf)
 {
 	struct page *page = buf->page;
 	int err;
@@ -143,7 +131,7 @@ static const struct pipe_buf_operations page_cache_pipe_buf_ops = {
 	.can_merge = 0,
 	.map = generic_pipe_buf_map,
 	.unmap = generic_pipe_buf_unmap,
-	.pin = page_cache_pipe_buf_pin,
+	.confirm = page_cache_pipe_buf_confirm,
 	.release = page_cache_pipe_buf_release,
 	.steal = page_cache_pipe_buf_steal,
 	.get = generic_pipe_buf_get,
@@ -163,18 +151,25 @@ static const struct pipe_buf_operations user_page_pipe_buf_ops = {
 	.can_merge = 0,
 	.map = generic_pipe_buf_map,
 	.unmap = generic_pipe_buf_unmap,
-	.pin = generic_pipe_buf_pin,
+	.confirm = generic_pipe_buf_confirm,
 	.release = page_cache_pipe_buf_release,
 	.steal = user_page_pipe_buf_steal,
 	.get = generic_pipe_buf_get,
 };
 
-/*
- * Pipe output worker. This sets up our pipe format with the page cache
- * pipe buffer operations. Otherwise very similar to the regular pipe_writev().
+/**
+ * splice_to_pipe - fill passed data into a pipe
+ * @pipe:	pipe to fill
+ * @spd:	data to fill
+ *
+ * Description:
+ *    @spd contains a map of pages and len/offset tupples, a long with
+ *    the struct pipe_buf_operations associated with these pages. This
+ *    function will link that data to the pipe.
+ *
  */
-static ssize_t splice_to_pipe(struct pipe_inode_info *pipe,
-			      struct splice_pipe_desc *spd)
+ssize_t splice_to_pipe(struct pipe_inode_info *pipe,
+		       struct splice_pipe_desc *spd)
 {
 	unsigned int spd_pages = spd->nr_pages;
 	int ret, do_wakeup, page_nr;
@@ -201,6 +196,7 @@ static ssize_t splice_to_pipe(struct pipe_inode_info *pipe,
 			buf->page = spd->pages[page_nr];
 			buf->offset = spd->partial[page_nr].offset;
 			buf->len = spd->partial[page_nr].len;
+			buf->private = spd->partial[page_nr].private;
 			buf->ops = spd->ops;
 			if (spd->flags & SPLICE_F_GIFT)
 				buf->flags |= PIPE_BUF_FLAG_GIFT;
@@ -295,11 +291,6 @@ __generic_file_splice_read(struct file *in, loff_t *ppos,
 	 */
 	page_cache_readahead(mapping, &in->f_ra, in, index, nr_pages);
 
-	/*
-	 * Now fill in the holes:
-	 */
-	error = 0;
-
 	/*
 	 * Lookup the (hopefully) full range of pages we need.
 	 */
@@ -307,8 +298,9 @@ __generic_file_splice_read(struct file *in, loff_t *ppos,
 
 	/*
 	 * If find_get_pages_contig() returned fewer pages than we needed,
-	 * allocate the rest.
+	 * allocate the rest and fill in the holes.
 	 */
+	error = 0;
 	index += spd.nr_pages;
 	while (spd.nr_pages < nr_pages) {
 		/*
@@ -470,11 +462,16 @@ __generic_file_splice_read(struct file *in, loff_t *ppos,
 /**
  * generic_file_splice_read - splice data from file to a pipe
  * @in:		file to splice from
+ * @ppos:	position in @in
  * @pipe:	pipe to splice to
  * @len:	number of bytes to splice
  * @flags:	splice modifier flags
  *
- * Will read pages from given file and fill them into a pipe.
+ * Description:
+ *    Will read pages from given file and fill them into a pipe. Can be
+ *    used as long as the address_space operations for the source implements
+ *    a readpage() hook.
+ *
  */
 ssize_t generic_file_splice_read(struct file *in, loff_t *ppos,
 				 struct pipe_inode_info *pipe, size_t len,
@@ -528,11 +525,11 @@ EXPORT_SYMBOL(generic_file_splice_read);
 static int pipe_to_sendpage(struct pipe_inode_info *pipe,
 			    struct pipe_buffer *buf, struct splice_desc *sd)
 {
-	struct file *file = sd->file;
+	struct file *file = sd->u.file;
 	loff_t pos = sd->pos;
 	int ret, more;
 
-	ret = buf->ops->pin(pipe, buf);
+	ret = buf->ops->confirm(pipe, buf);
 	if (!ret) {
 		more = (sd->flags & SPLICE_F_MORE) || sd->len < sd->total_len;
 
@@ -566,7 +563,7 @@ static int pipe_to_sendpage(struct pipe_inode_info *pipe,
 static int pipe_to_file(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
 			struct splice_desc *sd)
 {
-	struct file *file = sd->file;
+	struct file *file = sd->u.file;
 	struct address_space *mapping = file->f_mapping;
 	unsigned int offset, this_len;
 	struct page *page;
@@ -576,7 +573,7 @@ static int pipe_to_file(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
 	/*
 	 * make sure the data in this buffer is uptodate
 	 */
-	ret = buf->ops->pin(pipe, buf);
+	ret = buf->ops->confirm(pipe, buf);
 	if (unlikely(ret))
 		return ret;
 
@@ -663,36 +660,37 @@ static int pipe_to_file(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
 	return ret;
 }
 
-/*
- * Pipe input worker. Most of this logic works like a regular pipe, the
- * key here is the 'actor' worker passed in that actually moves the data
- * to the wanted destination. See pipe_to_file/pipe_to_sendpage above.
+/**
+ * __splice_from_pipe - splice data from a pipe to given actor
+ * @pipe:	pipe to splice from
+ * @sd:		information to @actor
+ * @actor:	handler that splices the data
+ *
+ * Description:
+ *    This function does little more than loop over the pipe and call
+ *    @actor to do the actual moving of a single struct pipe_buffer to
+ *    the desired destination. See pipe_to_file, pipe_to_sendpage, or
+ *    pipe_to_user.
+ *
  */
-ssize_t __splice_from_pipe(struct pipe_inode_info *pipe,
-			   struct file *out, loff_t *ppos, size_t len,
-			   unsigned int flags, splice_actor *actor)
+ssize_t __splice_from_pipe(struct pipe_inode_info *pipe, struct splice_desc *sd,
+			   splice_actor *actor)
 {
 	int ret, do_wakeup, err;
-	struct splice_desc sd;
 
 	ret = 0;
 	do_wakeup = 0;
 
-	sd.total_len = len;
-	sd.flags = flags;
-	sd.file = out;
-	sd.pos = *ppos;
-
 	for (;;) {
 		if (pipe->nrbufs) {
 			struct pipe_buffer *buf = pipe->bufs + pipe->curbuf;
 			const struct pipe_buf_operations *ops = buf->ops;
 
-			sd.len = buf->len;
-			if (sd.len > sd.total_len)
-				sd.len = sd.total_len;
+			sd->len = buf->len;
+			if (sd->len > sd->total_len)
+				sd->len = sd->total_len;
 
-			err = actor(pipe, buf, &sd);
+			err = actor(pipe, buf, sd);
 			if (err <= 0) {
 				if (!ret && err != -ENODATA)
 					ret = err;
@@ -704,10 +702,10 @@ ssize_t __splice_from_pipe(struct pipe_inode_info *pipe,
 			buf->offset += err;
 			buf->len -= err;
 
-			sd.len -= err;
-			sd.pos += err;
-			sd.total_len -= err;
-			if (sd.len)
+			sd->len -= err;
+			sd->pos += err;
+			sd->total_len -= err;
+			if (sd->len)
 				continue;
 
 			if (!buf->len) {
@@ -719,7 +717,7 @@ ssize_t __splice_from_pipe(struct pipe_inode_info *pipe,
 					do_wakeup = 1;
 			}
 
-			if (!sd.total_len)
+			if (!sd->total_len)
 				break;
 		}
 
@@ -732,7 +730,7 @@ ssize_t __splice_from_pipe(struct pipe_inode_info *pipe,
 				break;
 		}
 
-		if (flags & SPLICE_F_NONBLOCK) {
+		if (sd->flags & SPLICE_F_NONBLOCK) {
 			if (!ret)
 				ret = -EAGAIN;
 			break;
@@ -766,12 +764,32 @@ ssize_t __splice_from_pipe(struct pipe_inode_info *pipe,
 }
 EXPORT_SYMBOL(__splice_from_pipe);
 
+/**
+ * splice_from_pipe - splice data from a pipe to a file
+ * @pipe:	pipe to splice from
+ * @out:	file to splice to
+ * @ppos:	position in @out
+ * @len:	how many bytes to splice
+ * @flags:	splice modifier flags
+ * @actor:	handler that splices the data
+ *
+ * Description:
+ *    See __splice_from_pipe. This function locks the input and output inodes,
+ *    otherwise it's identical to __splice_from_pipe().
+ *
+ */
 ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out,
 			 loff_t *ppos, size_t len, unsigned int flags,
 			 splice_actor *actor)
 {
 	ssize_t ret;
 	struct inode *inode = out->f_mapping->host;
+	struct splice_desc sd = {
+		.total_len = len,
+		.flags = flags,
+		.pos = *ppos,
+		.u.file = out,
+	};
 
 	/*
 	 * The actor worker might be calling ->prepare_write and
@@ -780,7 +798,7 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out,
 	 * pipe->inode, we have to order lock acquiry here.
 	 */
 	inode_double_lock(inode, pipe->inode);
-	ret = __splice_from_pipe(pipe, out, ppos, len, flags, actor);
+	ret = __splice_from_pipe(pipe, &sd, actor);
 	inode_double_unlock(inode, pipe->inode);
 
 	return ret;
@@ -790,12 +808,14 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out,
  * generic_file_splice_write_nolock - generic_file_splice_write without mutexes
  * @pipe:	pipe info
  * @out:	file to write to
+ * @ppos:	position in @out
  * @len:	number of bytes to splice
  * @flags:	splice modifier flags
  *
- * Will either move or copy pages (determined by @flags options) from
- * the given pipe inode to the given file. The caller is responsible
- * for acquiring i_mutex on both inodes.
+ * Description:
+ *    Will either move or copy pages (determined by @flags options) from
+ *    the given pipe inode to the given file. The caller is responsible
+ *    for acquiring i_mutex on both inodes.
  *
  */
 ssize_t
@@ -804,6 +824,12 @@ generic_file_splice_write_nolock(struct pipe_inode_info *pipe, struct file *out,
 {
 	struct address_space *mapping = out->f_mapping;
 	struct inode *inode = mapping->host;
+	struct splice_desc sd = {
+		.total_len = len,
+		.flags = flags,
+		.pos = *ppos,
+		.u.file = out,
+	};
 	ssize_t ret;
 	int err;
 
@@ -811,7 +837,7 @@ generic_file_splice_write_nolock(struct pipe_inode_info *pipe, struct file *out,
 	if (unlikely(err))
 		return err;
 
-	ret = __splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
+	ret = __splice_from_pipe(pipe, &sd, pipe_to_file);
 	if (ret > 0) {
 		unsigned long nr_pages;
 
@@ -841,11 +867,13 @@ EXPORT_SYMBOL(generic_file_splice_write_nolock);
  * generic_file_splice_write - splice data from a pipe to a file
  * @pipe:	pipe info
  * @out:	file to write to
+ * @ppos:	position in @out
  * @len:	number of bytes to splice
  * @flags:	splice modifier flags
  *
- * Will either move or copy pages (determined by @flags options) from
- * the given pipe inode to the given file.
+ * Description:
+ *    Will either move or copy pages (determined by @flags options) from
+ *    the given pipe inode to the given file.
  *
  */
 ssize_t
@@ -896,13 +924,15 @@ EXPORT_SYMBOL(generic_file_splice_write);
 
 /**
  * generic_splice_sendpage - splice data from a pipe to a socket
- * @inode:	pipe inode
+ * @pipe:	pipe to splice from
  * @out:	socket to write to
+ * @ppos:	position in @out
  * @len:	number of bytes to splice
  * @flags:	splice modifier flags
  *
- * Will send @len bytes from the pipe to a network socket. No data copying
- * is involved.
+ * Description:
+ *    Will send @len bytes from the pipe to a network socket. No data copying
+ *    is involved.
  *
  */
 ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, struct file *out,
@@ -956,14 +986,27 @@ static long do_splice_to(struct file *in, loff_t *ppos,
 	return in->f_op->splice_read(in, ppos, pipe, len, flags);
 }
 
-long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
-		      size_t len, unsigned int flags)
+/**
+ * splice_direct_to_actor - splices data directly between two non-pipes
+ * @in:		file to splice from
+ * @sd:		actor information on where to splice to
+ * @actor:	handles the data splicing
+ *
+ * Description:
+ *    This is a special case helper to splice directly between two
+ *    points, without requiring an explicit pipe. Internally an allocated
+ *    pipe is cached in the process, and reused during the life time of
+ *    that process.
+ *
+ */
+ssize_t splice_direct_to_actor(struct file *in, struct splice_desc *sd,
+			       splice_direct_actor *actor)
 {
 	struct pipe_inode_info *pipe;
 	long ret, bytes;
-	loff_t out_off;
 	umode_t i_mode;
-	int i;
+	size_t len;
+	int i, flags;
 
 	/*
 	 * We require the input being a regular file, as we don't want to
@@ -999,7 +1042,13 @@ long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
 	 */
 	ret = 0;
 	bytes = 0;
-	out_off = 0;
+	len = sd->total_len;
+	flags = sd->flags;
+
+	/*
+	 * Don't block on output, we have to drain the direct pipe.
+	 */
+	sd->flags &= ~SPLICE_F_NONBLOCK;
 
 	while (len) {
 		size_t read_len, max_read_len;
@@ -1009,19 +1058,19 @@ long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
 		 */
 		max_read_len = min(len, (size_t)(PIPE_BUFFERS*PAGE_SIZE));
 
-		ret = do_splice_to(in, ppos, pipe, max_read_len, flags);
+		ret = do_splice_to(in, &sd->pos, pipe, max_read_len, flags);
 		if (unlikely(ret < 0))
 			goto out_release;
 
 		read_len = ret;
+		sd->total_len = read_len;
 
 		/*
 		 * NOTE: nonblocking mode only applies to the input. We
 		 * must not do the output in nonblocking mode as then we
 		 * could get stuck data in the internal pipe:
 		 */
-		ret = do_splice_from(pipe, out, &out_off, read_len,
-				     flags & ~SPLICE_F_NONBLOCK);
+		ret = actor(pipe, sd);
 		if (unlikely(ret < 0))
 			goto out_release;
 
@@ -1066,6 +1115,48 @@ long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
 		return bytes;
 
 	return ret;
+
+}
+EXPORT_SYMBOL(splice_direct_to_actor);
+
+static int direct_splice_actor(struct pipe_inode_info *pipe,
+			       struct splice_desc *sd)
+{
+	struct file *file = sd->u.file;
+
+	return do_splice_from(pipe, file, &sd->pos, sd->total_len, sd->flags);
+}
+
+/**
+ * do_splice_direct - splices data directly between two files
+ * @in:		file to splice from
+ * @ppos:	input file offset
+ * @out:	file to splice to
+ * @len:	number of bytes to splice
+ * @flags:	splice modifier flags
+ *
+ * Description:
+ *    For use by do_sendfile(). splice can easily emulate sendfile, but
+ *    doing it in the application would incur an extra system call
+ *    (splice in + splice out, as compared to just sendfile()). So this helper
+ *    can splice directly through a process-private pipe.
+ *
+ */
+long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
+		      size_t len, unsigned int flags)
+{
+	struct splice_desc sd = {
+		.len		= len,
+		.total_len	= len,
+		.flags		= flags,
+		.pos		= *ppos,
+		.u.file		= out,
+	};
+	size_t ret;
+
+	ret = splice_direct_to_actor(in, &sd, direct_splice_actor);
+	*ppos = sd.pos;
+	return ret;
 }
 
 /*
@@ -1248,28 +1339,131 @@ static int get_iovec_page_array(const struct iovec __user *iov,
 	return error;
 }
 
+static int pipe_to_user(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
+			struct splice_desc *sd)
+{
+	char *src;
+	int ret;
+
+	ret = buf->ops->confirm(pipe, buf);
+	if (unlikely(ret))
+		return ret;
+
+	/*
+	 * See if we can use the atomic maps, by prefaulting in the
+	 * pages and doing an atomic copy
+	 */
+	if (!fault_in_pages_writeable(sd->u.userptr, sd->len)) {
+		src = buf->ops->map(pipe, buf, 1);
+		ret = __copy_to_user_inatomic(sd->u.userptr, src + buf->offset,
+							sd->len);
+		buf->ops->unmap(pipe, buf, src);
+		if (!ret) {
+			ret = sd->len;
+			goto out;
+		}
+	}
+
+	/*
+	 * No dice, use slow non-atomic map and copy
+ 	 */
+	src = buf->ops->map(pipe, buf, 0);
+
+	ret = sd->len;
+	if (copy_to_user(sd->u.userptr, src + buf->offset, sd->len))
+		ret = -EFAULT;
+
+out:
+	if (ret > 0)
+		sd->u.userptr += ret;
+	buf->ops->unmap(pipe, buf, src);
+	return ret;
+}
+
+/*
+ * For lack of a better implementation, implement vmsplice() to userspace
+ * as a simple copy of the pipes pages to the user iov.
+ */
+static long vmsplice_to_user(struct file *file, const struct iovec __user *iov,
+			     unsigned long nr_segs, unsigned int flags)
+{
+	struct pipe_inode_info *pipe;
+	struct splice_desc sd;
+	ssize_t size;
+	int error;
+	long ret;
+
+	pipe = pipe_info(file->f_path.dentry->d_inode);
+	if (!pipe)
+		return -EBADF;
+
+	if (pipe->inode)
+		mutex_lock(&pipe->inode->i_mutex);
+
+	error = ret = 0;
+	while (nr_segs) {
+		void __user *base;
+		size_t len;
+
+		/*
+		 * Get user address base and length for this iovec.
+		 */
+		error = get_user(base, &iov->iov_base);
+		if (unlikely(error))
+			break;
+		error = get_user(len, &iov->iov_len);
+		if (unlikely(error))
+			break;
+
+		/*
+		 * Sanity check this iovec. 0 read succeeds.
+		 */
+		if (unlikely(!len))
+			break;
+		if (unlikely(!base)) {
+			error = -EFAULT;
+			break;
+		}
+
+		sd.len = 0;
+		sd.total_len = len;
+		sd.flags = flags;
+		sd.u.userptr = base;
+		sd.pos = 0;
+
+		size = __splice_from_pipe(pipe, &sd, pipe_to_user);
+		if (size < 0) {
+			if (!ret)
+				ret = size;
+
+			break;
+		}
+
+		ret += size;
+
+		if (size < len)
+			break;
+
+		nr_segs--;
+		iov++;
+	}
+
+	if (pipe->inode)
+		mutex_unlock(&pipe->inode->i_mutex);
+
+	if (!ret)
+		ret = error;
+
+	return ret;
+}
+
 /*
  * vmsplice splices a user address range into a pipe. It can be thought of
  * as splice-from-memory, where the regular splice is splice-from-file (or
  * to file). In both cases the output is a pipe, naturally.
- *
- * Note that vmsplice only supports splicing _from_ user memory to a pipe,
- * not the other way around. Splicing from user memory is a simple operation
- * that can be supported without any funky alignment restrictions or nasty
- * vm tricks. We simply map in the user memory and fill them into a pipe.
- * The reverse isn't quite as easy, though. There are two possible solutions
- * for that:
- *
- *	- memcpy() the data internally, at which point we might as well just
- *	  do a regular read() on the buffer anyway.
- *	- Lots of nasty vm tricks, that are neither fast nor flexible (it
- *	  has restriction limitations on both ends of the pipe).
- *
- * Alas, it isn't here.
- *
  */
-static long do_vmsplice(struct file *file, const struct iovec __user *iov,
-			unsigned long nr_segs, unsigned int flags)
+static long vmsplice_to_pipe(struct file *file, const struct iovec __user *iov,
+			     unsigned long nr_segs, unsigned int flags)
 {
 	struct pipe_inode_info *pipe;
 	struct page *pages[PIPE_BUFFERS];
@@ -1284,10 +1478,6 @@ static long do_vmsplice(struct file *file, const struct iovec __user *iov,
 	pipe = pipe_info(file->f_path.dentry->d_inode);
 	if (!pipe)
 		return -EBADF;
-	if (unlikely(nr_segs > UIO_MAXIOV))
-		return -EINVAL;
-	else if (unlikely(!nr_segs))
-		return 0;
 
 	spd.nr_pages = get_iovec_page_array(iov, nr_segs, pages, partial,
 					    flags & SPLICE_F_GIFT);
@@ -1297,6 +1487,22 @@ static long do_vmsplice(struct file *file, const struct iovec __user *iov,
 	return splice_to_pipe(pipe, &spd);
 }
 
+/*
+ * Note that vmsplice only really supports true splicing _from_ user memory
+ * to a pipe, not the other way around. Splicing from user memory is a simple
+ * operation that can be supported without any funky alignment restrictions
+ * or nasty vm tricks. We simply map in the user memory and fill them into
+ * a pipe. The reverse isn't quite as easy, though. There are two possible
+ * solutions for that:
+ *
+ *	- memcpy() the data internally, at which point we might as well just
+ *	  do a regular read() on the buffer anyway.
+ *	- Lots of nasty vm tricks, that are neither fast nor flexible (it
+ *	  has restriction limitations on both ends of the pipe).
+ *
+ * Currently we punt and implement it as a normal copy, see pipe_to_user().
+ *
+ */
 asmlinkage long sys_vmsplice(int fd, const struct iovec __user *iov,
 			     unsigned long nr_segs, unsigned int flags)
 {
@@ -1304,11 +1510,18 @@ asmlinkage long sys_vmsplice(int fd, const struct iovec __user *iov,
 	long error;
 	int fput;
 
+	if (unlikely(nr_segs > UIO_MAXIOV))
+		return -EINVAL;
+	else if (unlikely(!nr_segs))
+		return 0;
+
 	error = -EBADF;
 	file = fget_light(fd, &fput);
 	if (file) {
 		if (file->f_mode & FMODE_WRITE)
-			error = do_vmsplice(file, iov, nr_segs, flags);
+			error = vmsplice_to_pipe(file, iov, nr_segs, flags);
+		else if (file->f_mode & FMODE_READ)
+			error = vmsplice_to_user(file, iov, nr_segs, flags);
 
 		fput_light(file, fput);
 	}

fs/sysv/file.c

@@ -27,7 +27,7 @@ const struct file_operations sysv_file_operations = {
 	.aio_write	= generic_file_aio_write,
 	.mmap		= generic_file_mmap,
 	.fsync		= sysv_sync_file,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };
 
 const struct inode_operations sysv_file_inode_operations = {

fs/udf/file.c

@@ -261,7 +261,7 @@ const struct file_operations udf_file_operations = {
 	.aio_write		= udf_file_aio_write,
 	.release		= udf_release_file,
 	.fsync			= udf_fsync_file,
-	.sendfile		= generic_file_sendfile,
+	.splice_read		= generic_file_splice_read,
 };
 
 const struct inode_operations udf_file_inode_operations = {

fs/ufs/file.c

@@ -60,5 +60,5 @@ const struct file_operations ufs_file_operations = {
 	.mmap		= generic_file_mmap,
 	.open           = generic_file_open,
 	.fsync		= ufs_sync_file,
-	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
 };

fs/xfs/linux-2.6/xfs_file.c

@@ -123,30 +123,6 @@ xfs_file_aio_write_invis(
 	return __xfs_file_write(iocb, iov, nr_segs, IO_ISAIO|IO_INVIS, pos);
 }
 
-STATIC ssize_t
-xfs_file_sendfile(
-	struct file		*filp,
-	loff_t			*pos,
-	size_t			count,
-	read_actor_t		actor,
-	void			*target)
-{
-	return bhv_vop_sendfile(vn_from_inode(filp->f_path.dentry->d_inode),
-				filp, pos, 0, count, actor, target, NULL);
-}
-
-STATIC ssize_t
-xfs_file_sendfile_invis(
-	struct file		*filp,
-	loff_t			*pos,
-	size_t			count,
-	read_actor_t		actor,
-	void			*target)
-{
-	return bhv_vop_sendfile(vn_from_inode(filp->f_path.dentry->d_inode),
-				filp, pos, IO_INVIS, count, actor, target, NULL);
-}
-
 STATIC ssize_t
 xfs_file_splice_read(
 	struct file		*infilp,
@@ -452,7 +428,6 @@ const struct file_operations xfs_file_operations = {
 	.write		= do_sync_write,
 	.aio_read	= xfs_file_aio_read,
 	.aio_write	= xfs_file_aio_write,
-	.sendfile	= xfs_file_sendfile,
 	.splice_read	= xfs_file_splice_read,
 	.splice_write	= xfs_file_splice_write,
 	.unlocked_ioctl	= xfs_file_ioctl,
@@ -475,7 +450,6 @@ const struct file_operations xfs_invis_file_operations = {
 	.write		= do_sync_write,
 	.aio_read	= xfs_file_aio_read_invis,
 	.aio_write	= xfs_file_aio_write_invis,
-	.sendfile	= xfs_file_sendfile_invis,
 	.splice_read	= xfs_file_splice_read_invis,
 	.splice_write	= xfs_file_splice_write_invis,
 	.unlocked_ioctl	= xfs_file_ioctl_invis,

fs/xfs/linux-2.6/xfs_linux.h

@@ -101,7 +101,6 @@
  * Feature macros (disable/enable)
  */
 #undef  HAVE_REFCACHE	/* reference cache not needed for NFS in 2.6 */
-#define HAVE_SENDFILE	/* sendfile(2) exists in 2.6, but not in 2.4 */
 #define HAVE_SPLICE	/* a splice(2) exists in 2.6, but not in 2.4 */
 #ifdef CONFIG_SMP
 #define HAVE_PERCPU_SB	/* per cpu superblock counters are a 2.6 feature */

fs/xfs/linux-2.6/xfs_lrw.c

@@ -286,50 +286,6 @@ xfs_read(
 	return ret;
 }
 
-ssize_t
-xfs_sendfile(
-	bhv_desc_t		*bdp,
-	struct file		*filp,
-	loff_t			*offset,
-	int			ioflags,
-	size_t			count,
-	read_actor_t		actor,
-	void			*target,
-	cred_t			*credp)
-{
-	xfs_inode_t		*ip = XFS_BHVTOI(bdp);
-	xfs_mount_t		*mp = ip->i_mount;
-	ssize_t			ret;
-
-	XFS_STATS_INC(xs_read_calls);
-	if (XFS_FORCED_SHUTDOWN(mp))
-		return -EIO;
-
-	xfs_ilock(ip, XFS_IOLOCK_SHARED);
-
-	if (DM_EVENT_ENABLED(BHV_TO_VNODE(bdp)->v_vfsp, ip, DM_EVENT_READ) &&
-	    (!(ioflags & IO_INVIS))) {
-		bhv_vrwlock_t locktype = VRWLOCK_READ;
-		int error;
-
-		error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp),
-				      *offset, count,
-				      FILP_DELAY_FLAG(filp), &locktype);
-		if (error) {
-			xfs_iunlock(ip, XFS_IOLOCK_SHARED);
-			return -error;
-		}
-	}
-	xfs_rw_enter_trace(XFS_SENDFILE_ENTER, &ip->i_iocore,
-		   (void *)(unsigned long)target, count, *offset, ioflags);
-	ret = generic_file_sendfile(filp, offset, count, actor, target);
-	if (ret > 0)
-		XFS_STATS_ADD(xs_read_bytes, ret);
-
-	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
-	return ret;
-}
-
 ssize_t
 xfs_splice_read(
 	bhv_desc_t		*bdp,

fs/xfs/linux-2.6/xfs_lrw.h

@@ -90,9 +90,6 @@ extern ssize_t xfs_read(struct bhv_desc *, struct kiocb *,
 extern ssize_t xfs_write(struct bhv_desc *, struct kiocb *,
 				const struct iovec *, unsigned int,
 				loff_t *, int, struct cred *);
-extern ssize_t xfs_sendfile(struct bhv_desc *, struct file *,
-				loff_t *, int, size_t, read_actor_t,
-				void *, struct cred *);
 extern ssize_t xfs_splice_read(struct bhv_desc *, struct file *, loff_t *,
 				struct pipe_inode_info *, size_t, int, int,
 				struct cred *);

fs/xfs/linux-2.6/xfs_vnode.h

@@ -139,9 +139,6 @@ typedef ssize_t (*vop_read_t)(bhv_desc_t *, struct kiocb *,
 typedef ssize_t (*vop_write_t)(bhv_desc_t *, struct kiocb *,
 				const struct iovec *, unsigned int,
 				loff_t *, int, struct cred *);
-typedef ssize_t (*vop_sendfile_t)(bhv_desc_t *, struct file *,
-				loff_t *, int, size_t, read_actor_t,
-				void *, struct cred *);
 typedef ssize_t (*vop_splice_read_t)(bhv_desc_t *, struct file *, loff_t *,
 				struct pipe_inode_info *, size_t, int, int,
 				struct cred *);
@@ -206,7 +203,6 @@ typedef struct bhv_vnodeops {
 	vop_close_t		vop_close;
 	vop_read_t		vop_read;
 	vop_write_t		vop_write;
-	vop_sendfile_t		vop_sendfile;
 	vop_splice_read_t	vop_splice_read;
 	vop_splice_write_t	vop_splice_write;
 	vop_ioctl_t		vop_ioctl;
@@ -254,8 +250,6 @@ typedef struct bhv_vnodeops {
 		VOP(vop_read, vp)(VNHEAD(vp),file,iov,segs,offset,ioflags,cr)
 #define bhv_vop_write(vp,file,iov,segs,offset,ioflags,cr)		\
 		VOP(vop_write, vp)(VNHEAD(vp),file,iov,segs,offset,ioflags,cr)
-#define bhv_vop_sendfile(vp,f,off,ioflags,cnt,act,targ,cr)		\
-		VOP(vop_sendfile, vp)(VNHEAD(vp),f,off,ioflags,cnt,act,targ,cr)
 #define bhv_vop_splice_read(vp,f,o,pipe,cnt,fl,iofl,cr)			\
 		VOP(vop_splice_read, vp)(VNHEAD(vp),f,o,pipe,cnt,fl,iofl,cr)
 #define bhv_vop_splice_write(vp,f,o,pipe,cnt,fl,iofl,cr)		\

fs/xfs/xfs_vnodeops.c

@@ -4680,9 +4680,6 @@ bhv_vnodeops_t xfs_vnodeops = {
 	.vop_open		= xfs_open,
 	.vop_close		= xfs_close,
 	.vop_read		= xfs_read,
-#ifdef HAVE_SENDFILE
-	.vop_sendfile		= xfs_sendfile,
-#endif
 #ifdef HAVE_SPLICE
 	.vop_splice_read	= xfs_splice_read,
 	.vop_splice_write	= xfs_splice_write,

include/linux/fs.h

@@ -1054,7 +1054,7 @@ struct block_device_operations {
 };
 
 /*
- * "descriptor" for what we're up to with a read for sendfile().
+ * "descriptor" for what we're up to with a read.
  * This allows us to use the same read code yet
  * have multiple different users of the data that
  * we read from a file.
@@ -1105,7 +1105,6 @@ struct file_operations {
 	int (*aio_fsync) (struct kiocb *, int datasync);
 	int (*fasync) (int, struct file *, int);
 	int (*lock) (struct file *, int, struct file_lock *);
-	ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
 	int (*check_flags)(int);
@@ -1762,7 +1761,6 @@ extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
-extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
 extern void do_generic_mapping_read(struct address_space *mapping,
 				    struct file_ra_state *, struct file *,
 				    loff_t *, read_descriptor_t *, read_actor_t);
@@ -1792,9 +1790,6 @@ extern int nonseekable_open(struct inode * inode, struct file * filp);
 #ifdef CONFIG_FS_XIP
 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
 			     loff_t *ppos);
-extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
-				 size_t count, read_actor_t actor,
-				 void *target);
 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
 			      size_t len, loff_t *ppos);

include/linux/pipe_fs_i.h

@@ -9,13 +9,39 @@
 #define PIPE_BUF_FLAG_ATOMIC	0x02	/* was atomically mapped */
 #define PIPE_BUF_FLAG_GIFT	0x04	/* page is a gift */
 
+/**
+ *	struct pipe_buffer - a linux kernel pipe buffer
+ *	@page: the page containing the data for the pipe buffer
+ *	@offset: offset of data inside the @page
+ *	@len: length of data inside the @page
+ *	@ops: operations associated with this buffer. See @pipe_buf_operations.
+ *	@flags: pipe buffer flags. See above.
+ *	@private: private data owned by the ops.
+ **/
 struct pipe_buffer {
 	struct page *page;
 	unsigned int offset, len;
 	const struct pipe_buf_operations *ops;
 	unsigned int flags;
+	unsigned long private;
 };
 
+/**
+ *	struct pipe_inode_info - a linux kernel pipe
+ *	@wait: reader/writer wait point in case of empty/full pipe
+ *	@nrbufs: the number of non-empty pipe buffers in this pipe
+ *	@curbuf: the current pipe buffer entry
+ *	@tmp_page: cached released page
+ *	@readers: number of current readers of this pipe
+ *	@writers: number of current writers of this pipe
+ *	@waiting_writers: number of writers blocked waiting for room
+ *	@r_counter: reader counter
+ *	@w_counter: writer counter
+ *	@fasync_readers: reader side fasync
+ *	@fasync_writers: writer side fasync
+ *	@inode: inode this pipe is attached to
+ *	@bufs: the circular array of pipe buffers
+ **/
 struct pipe_inode_info {
 	wait_queue_head_t wait;
 	unsigned int nrbufs, curbuf;
@@ -34,22 +60,73 @@ struct pipe_inode_info {
 /*
  * Note on the nesting of these functions:
  *
- * ->pin()
+ * ->confirm()
  *	->steal()
  *	...
  *	->map()
  *	...
  *	->unmap()
  *
- * That is, ->map() must be called on a pinned buffer, same goes for ->steal().
+ * That is, ->map() must be called on a confirmed buffer,
+ * same goes for ->steal(). See below for the meaning of each
+ * operation. Also see kerneldoc in fs/pipe.c for the pipe
+ * and generic variants of these hooks.
  */
 struct pipe_buf_operations {
+	/*
+	 * This is set to 1, if the generic pipe read/write may coalesce
+	 * data into an existing buffer. If this is set to 0, a new pipe
+	 * page segment is always used for new data.
+	 */
 	int can_merge;
+
+	/*
+	 * ->map() returns a virtual address mapping of the pipe buffer.
+	 * The last integer flag reflects whether this should be an atomic
+	 * mapping or not. The atomic map is faster, however you can't take
+	 * page faults before calling ->unmap() again. So if you need to eg
+	 * access user data through copy_to/from_user(), then you must get
+	 * a non-atomic map. ->map() uses the KM_USER0 atomic slot for
+	 * atomic maps, so you can't map more than one pipe_buffer at once
+	 * and you have to be careful if mapping another page as source
+	 * or destination for a copy (IOW, it has to use something else
+	 * than KM_USER0).
+	 */
 	void * (*map)(struct pipe_inode_info *, struct pipe_buffer *, int);
+
+	/*
+	 * Undoes ->map(), finishes the virtual mapping of the pipe buffer.
+	 */
 	void (*unmap)(struct pipe_inode_info *, struct pipe_buffer *, void *);
-	int (*pin)(struct pipe_inode_info *, struct pipe_buffer *);
+
+	/*
+	 * ->confirm() verifies that the data in the pipe buffer is there
+	 * and that the contents are good. If the pages in the pipe belong
+	 * to a file system, we may need to wait for IO completion in this
+	 * hook. Returns 0 for good, or a negative error value in case of
+	 * error.
+	 */
+	int (*confirm)(struct pipe_inode_info *, struct pipe_buffer *);
+
+	/*
+	 * When the contents of this pipe buffer has been completely
+	 * consumed by a reader, ->release() is called.
+	 */
 	void (*release)(struct pipe_inode_info *, struct pipe_buffer *);
+
+	/*
+	 * Attempt to take ownership of the pipe buffer and its contents.
+	 * ->steal() returns 0 for success, in which case the contents
+	 * of the pipe (the buf->page) is locked and now completely owned
+	 * by the caller. The page may then be transferred to a different
+	 * mapping, the most often used case is insertion into different
+	 * file address space cache.
+	 */
 	int (*steal)(struct pipe_inode_info *, struct pipe_buffer *);
+
+	/*
+	 * Get a reference to the pipe buffer.
+	 */
 	void (*get)(struct pipe_inode_info *, struct pipe_buffer *);
 };
 
@@ -68,39 +145,7 @@ void __free_pipe_info(struct pipe_inode_info *);
 void *generic_pipe_buf_map(struct pipe_inode_info *, struct pipe_buffer *, int);
 void generic_pipe_buf_unmap(struct pipe_inode_info *, struct pipe_buffer *, void *);
 void generic_pipe_buf_get(struct pipe_inode_info *, struct pipe_buffer *);
-int generic_pipe_buf_pin(struct pipe_inode_info *, struct pipe_buffer *);
+int generic_pipe_buf_confirm(struct pipe_inode_info *, struct pipe_buffer *);
 int generic_pipe_buf_steal(struct pipe_inode_info *, struct pipe_buffer *);
 
-/*
- * splice is tied to pipes as a transport (at least for now), so we'll just
- * add the splice flags here.
- */
-#define SPLICE_F_MOVE	(0x01)	/* move pages instead of copying */
-#define SPLICE_F_NONBLOCK (0x02) /* don't block on the pipe splicing (but */
-				 /* we may still block on the fd we splice */
-				 /* from/to, of course */
-#define SPLICE_F_MORE	(0x04)	/* expect more data */
-#define SPLICE_F_GIFT	(0x08)	/* pages passed in are a gift */
-
-/*
- * Passed to the actors
- */
-struct splice_desc {
-	unsigned int len, total_len;	/* current and remaining length */
-	unsigned int flags;		/* splice flags */
-	struct file *file;		/* file to read/write */
-	loff_t pos;			/* file position */
-};
-
-typedef int (splice_actor)(struct pipe_inode_info *, struct pipe_buffer *,
-			   struct splice_desc *);
-
-extern ssize_t splice_from_pipe(struct pipe_inode_info *, struct file *,
-				loff_t *, size_t, unsigned int,
-				splice_actor *);
-
-extern ssize_t __splice_from_pipe(struct pipe_inode_info *, struct file *,
-				  loff_t *, size_t, unsigned int,
-				  splice_actor *);
-
 #endif

include/linux/splice.h

+/*
+ * Function declerations and data structures related to the splice
+ * implementation.
+ *
+ * Copyright (C) 2007 Jens Axboe <jens.axboe@oracle.com>
+ *
+ */
+#ifndef SPLICE_H
+#define SPLICE_H
+
+#include <linux/pipe_fs_i.h>
+
+/*
+ * splice is tied to pipes as a transport (at least for now), so we'll just
+ * add the splice flags here.
+ */
+#define SPLICE_F_MOVE	(0x01)	/* move pages instead of copying */
+#define SPLICE_F_NONBLOCK (0x02) /* don't block on the pipe splicing (but */
+				 /* we may still block on the fd we splice */
+				 /* from/to, of course */
+#define SPLICE_F_MORE	(0x04)	/* expect more data */
+#define SPLICE_F_GIFT	(0x08)	/* pages passed in are a gift */
+
+/*
+ * Passed to the actors
+ */
+struct splice_desc {
+	unsigned int len, total_len;	/* current and remaining length */
+	unsigned int flags;		/* splice flags */
+	/*
+	 * actor() private data
+	 */
+	union {
+		void __user *userptr;	/* memory to write to */
+		struct file *file;	/* file to read/write */
+		void *data;		/* cookie */
+	} u;
+	loff_t pos;			/* file position */
+};
+
+struct partial_page {
+	unsigned int offset;
+	unsigned int len;
+	unsigned long private;
+};
+
+/*
+ * Passed to splice_to_pipe
+ */
+struct splice_pipe_desc {
+	struct page **pages;		/* page map */
+	struct partial_page *partial;	/* pages[] may not be contig */
+	int nr_pages;			/* number of pages in map */
+	unsigned int flags;		/* splice flags */
+	const struct pipe_buf_operations *ops;/* ops associated with output pipe */
+};
+
+typedef int (splice_actor)(struct pipe_inode_info *, struct pipe_buffer *,
+			   struct splice_desc *);
+typedef int (splice_direct_actor)(struct pipe_inode_info *,
+				  struct splice_desc *);
+
+extern ssize_t splice_from_pipe(struct pipe_inode_info *, struct file *,
+				loff_t *, size_t, unsigned int,
+				splice_actor *);
+extern ssize_t __splice_from_pipe(struct pipe_inode_info *,
+				  struct splice_desc *, splice_actor *);
+extern ssize_t splice_to_pipe(struct pipe_inode_info *,
+			      struct splice_pipe_desc *);
+extern ssize_t splice_direct_to_actor(struct file *, struct splice_desc *,
+				      splice_direct_actor *);
+
+#endif

include/linux/sunrpc/svc.h

@@ -253,7 +253,7 @@ struct svc_rqst {
 						 * determine what device number
 						 * to report (real or virtual)
 						 */
-	int			rq_sendfile_ok; /* turned off in gss privacy
+	int			rq_splice_ok;   /* turned off in gss privacy
 						 * to prevent encrypting page
 						 * cache pages */
 	wait_queue_head_t	rq_wait;	/* synchronization */

kernel/relay.c

@@ -21,6 +21,7 @@
 #include <linux/vmalloc.h>
 #include <linux/mm.h>
 #include <linux/cpu.h>
+#include <linux/splice.h>
 
 /* list of open channels, for cpu hotplug */
 static DEFINE_MUTEX(relay_channels_mutex);
@@ -121,6 +122,7 @@ static void *relay_alloc_buf(struct rchan_buf *buf, size_t *size)
 		buf->page_array[i] = alloc_page(GFP_KERNEL);
 		if (unlikely(!buf->page_array[i]))
 			goto depopulate;
+		set_page_private(buf->page_array[i], (unsigned long)buf);
 	}
 	mem = vmap(buf->page_array, n_pages, VM_MAP, PAGE_KERNEL);
 	if (!mem)
@@ -970,43 +972,6 @@ static int subbuf_read_actor(size_t read_start,
 	return ret;
 }
 
-/*
- *	subbuf_send_actor - send up to one subbuf's worth of data
- */
-static int subbuf_send_actor(size_t read_start,
-			     struct rchan_buf *buf,
-			     size_t avail,
-			     read_descriptor_t *desc,
-			     read_actor_t actor)
-{
-	unsigned long pidx, poff;
-	unsigned int subbuf_pages;
-	int ret = 0;
-
-	subbuf_pages = buf->chan->alloc_size >> PAGE_SHIFT;
-	pidx = (read_start / PAGE_SIZE) % subbuf_pages;
-	poff = read_start & ~PAGE_MASK;
-	while (avail) {
-		struct page *p = buf->page_array[pidx];
-		unsigned int len;
-
-		len = PAGE_SIZE - poff;
-		if (len > avail)
-			len = avail;
-
-		len = actor(desc, p, poff, len);
-		if (desc->error)
-			break;
-
-		avail -= len;
-		ret += len;
-		poff = 0;
-		pidx = (pidx + 1) % subbuf_pages;
-	}
-
-	return ret;
-}
-
 typedef int (*subbuf_actor_t) (size_t read_start,
 			       struct rchan_buf *buf,
 			       size_t avail,
@@ -1067,19 +1032,159 @@ static ssize_t relay_file_read(struct file *filp,
 				       NULL, &desc);
 }
 
-static ssize_t relay_file_sendfile(struct file *filp,
-				   loff_t *ppos,
-				   size_t count,
-				   read_actor_t actor,
-				   void *target)
+static void relay_consume_bytes(struct rchan_buf *rbuf, int bytes_consumed)
 {
-	read_descriptor_t desc;
-	desc.written = 0;
-	desc.count = count;
-	desc.arg.data = target;
-	desc.error = 0;
-	return relay_file_read_subbufs(filp, ppos, subbuf_send_actor,
-				       actor, &desc);
+	rbuf->bytes_consumed += bytes_consumed;
+
+	if (rbuf->bytes_consumed >= rbuf->chan->subbuf_size) {
+		relay_subbufs_consumed(rbuf->chan, rbuf->cpu, 1);
+		rbuf->bytes_consumed %= rbuf->chan->subbuf_size;
+	}
+}
+
+static void relay_pipe_buf_release(struct pipe_inode_info *pipe,
+				   struct pipe_buffer *buf)
+{
+	struct rchan_buf *rbuf;
+
+	rbuf = (struct rchan_buf *)page_private(buf->page);
+	relay_consume_bytes(rbuf, buf->private);
+}
+
+static struct pipe_buf_operations relay_pipe_buf_ops = {
+	.can_merge = 0,
+	.map = generic_pipe_buf_map,
+	.unmap = generic_pipe_buf_unmap,
+	.confirm = generic_pipe_buf_confirm,
+	.release = relay_pipe_buf_release,
+	.steal = generic_pipe_buf_steal,
+	.get = generic_pipe_buf_get,
+};
+
+/**
+ *	subbuf_splice_actor - splice up to one subbuf's worth of data
+ */
+static int subbuf_splice_actor(struct file *in,
+			       loff_t *ppos,
+			       struct pipe_inode_info *pipe,
+			       size_t len,
+			       unsigned int flags,
+			       int *nonpad_ret)
+{
+	unsigned int pidx, poff, total_len, subbuf_pages, ret;
+	struct rchan_buf *rbuf = in->private_data;
+	unsigned int subbuf_size = rbuf->chan->subbuf_size;
+	size_t read_start = ((size_t)*ppos) % rbuf->chan->alloc_size;
+	size_t read_subbuf = read_start / subbuf_size;
+	size_t padding = rbuf->padding[read_subbuf];
+	size_t nonpad_end = read_subbuf * subbuf_size + subbuf_size - padding;
+	struct page *pages[PIPE_BUFFERS];
+	struct partial_page partial[PIPE_BUFFERS];
+	struct splice_pipe_desc spd = {
+		.pages = pages,
+		.nr_pages = 0,
+		.partial = partial,
+		.flags = flags,
+		.ops = &relay_pipe_buf_ops,
+	};
+
+	if (rbuf->subbufs_produced == rbuf->subbufs_consumed)
+		return 0;
+
+	/*
+	 * Adjust read len, if longer than what is available
+	 */
+	if (len > (subbuf_size - read_start % subbuf_size))
+		len = subbuf_size - read_start % subbuf_size;
+
+	subbuf_pages = rbuf->chan->alloc_size >> PAGE_SHIFT;
+	pidx = (read_start / PAGE_SIZE) % subbuf_pages;
+	poff = read_start & ~PAGE_MASK;
+
+	for (total_len = 0; spd.nr_pages < subbuf_pages; spd.nr_pages++) {
+		unsigned int this_len, this_end, private;
+		unsigned int cur_pos = read_start + total_len;
+
+		if (!len)
+			break;
+
+		this_len = min_t(unsigned long, len, PAGE_SIZE - poff);
+		private = this_len;
+
+		spd.pages[spd.nr_pages] = rbuf->page_array[pidx];
+		spd.partial[spd.nr_pages].offset = poff;
+
+		this_end = cur_pos + this_len;
+		if (this_end >= nonpad_end) {
+			this_len = nonpad_end - cur_pos;
+			private = this_len + padding;
+		}
+		spd.partial[spd.nr_pages].len = this_len;
+		spd.partial[spd.nr_pages].private = private;
+
+		len -= this_len;
+		total_len += this_len;
+		poff = 0;
+		pidx = (pidx + 1) % subbuf_pages;
+
+		if (this_end >= nonpad_end) {
+			spd.nr_pages++;
+			break;
+		}
+	}
+
+	if (!spd.nr_pages)
+		return 0;
+
+	ret = *nonpad_ret = splice_to_pipe(pipe, &spd);
+	if (ret < 0 || ret < total_len)
+		return ret;
+
+        if (read_start + ret == nonpad_end)
+                ret += padding;
+
+        return ret;
+}
+
+static ssize_t relay_file_splice_read(struct file *in,
+				      loff_t *ppos,
+				      struct pipe_inode_info *pipe,
+				      size_t len,
+				      unsigned int flags)
+{
+	ssize_t spliced;
+	int ret;
+	int nonpad_ret = 0;
+
+	ret = 0;
+	spliced = 0;
+
+	while (len) {
+		ret = subbuf_splice_actor(in, ppos, pipe, len, flags, &nonpad_ret);
+		if (ret < 0)
+			break;
+		else if (!ret) {
+			if (spliced)
+				break;
+			if (flags & SPLICE_F_NONBLOCK) {
+				ret = -EAGAIN;
+				break;
+			}
+		}
+
+		*ppos += ret;
+		if (ret > len)
+			len = 0;
+		else
+			len -= ret;
+		spliced += nonpad_ret;
+		nonpad_ret = 0;
+	}
+
+	if (spliced)
+		return spliced;
+
+	return ret;
 }
 
 const struct file_operations relay_file_operations = {
@@ -1089,7 +1194,7 @@ const struct file_operations relay_file_operations = {
 	.read		= relay_file_read,
 	.llseek		= no_llseek,
 	.release	= relay_file_release,
-	.sendfile       = relay_file_sendfile,
+	.splice_read	= relay_file_splice_read,
 };
 EXPORT_SYMBOL_GPL(relay_file_operations);
 

mm/filemap.c

@@ -1245,26 +1245,6 @@ int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long o
 	return written;
 }
 
-ssize_t generic_file_sendfile(struct file *in_file, loff_t *ppos,
-			 size_t count, read_actor_t actor, void *target)
-{
-	read_descriptor_t desc;
-
-	if (!count)
-		return 0;
-
-	desc.written = 0;
-	desc.count = count;
-	desc.arg.data = target;
-	desc.error = 0;
-
-	do_generic_file_read(in_file, ppos, &desc, actor);
-	if (desc.written)
-		return desc.written;
-	return desc.error;
-}
-EXPORT_SYMBOL(generic_file_sendfile);
-
 static ssize_t
 do_readahead(struct address_space *mapping, struct file *filp,
 	     unsigned long index, unsigned long nr)

mm/filemap_xip.c

@@ -159,28 +159,6 @@ xip_file_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos)
 }
 EXPORT_SYMBOL_GPL(xip_file_read);
 
-ssize_t
-xip_file_sendfile(struct file *in_file, loff_t *ppos,
-	     size_t count, read_actor_t actor, void *target)
-{
-	read_descriptor_t desc;
-
-	if (!count)
-		return 0;
-
-	desc.written = 0;
-	desc.count = count;
-	desc.arg.data = target;
-	desc.error = 0;
-
-	do_xip_mapping_read(in_file->f_mapping, &in_file->f_ra, in_file,
-			    ppos, &desc, actor);
-	if (desc.written)
-		return desc.written;
-	return desc.error;
-}
-EXPORT_SYMBOL_GPL(xip_file_sendfile);
-
 /*
  * __xip_unmap is invoked from xip_unmap and
  * xip_write

mm/shmem.c

@@ -1100,9 +1100,9 @@ static int shmem_getpage(struct inode *inode, unsigned long idx,
 	 * Normally, filepage is NULL on entry, and either found
 	 * uptodate immediately, or allocated and zeroed, or read
 	 * in under swappage, which is then assigned to filepage.
-	 * But shmem_prepare_write passes in a locked filepage,
-	 * which may be found not uptodate by other callers too,
-	 * and may need to be copied from the swappage read in.
+	 * But shmem_readpage and shmem_prepare_write pass in a locked
+	 * filepage, which may be found not uptodate by other callers
+	 * too, and may need to be copied from the swappage read in.
 	 */
 repeat:
 	if (!filepage)
@@ -1485,9 +1485,18 @@ static const struct inode_operations shmem_symlink_inode_operations;
 static const struct inode_operations shmem_symlink_inline_operations;
 
 /*
- * Normally tmpfs makes no use of shmem_prepare_write, but it
- * lets a tmpfs file be used read-write below the loop driver.
+ * Normally tmpfs avoids the use of shmem_readpage and shmem_prepare_write;
+ * but providing them allows a tmpfs file to be used for splice, sendfile, and
+ * below the loop driver, in the generic fashion that many filesystems support.
  */
+static int shmem_readpage(struct file *file, struct page *page)
+{
+	struct inode *inode = page->mapping->host;
+	int error = shmem_getpage(inode, page->index, &page, SGP_CACHE, NULL);
+	unlock_page(page);
+	return error;
+}
+
 static int
 shmem_prepare_write(struct file *file, struct page *page, unsigned offset, unsigned to)
 {
@@ -1711,25 +1720,6 @@ static ssize_t shmem_file_read(struct file *filp, char __user *buf, size_t count
 	return desc.error;
 }
 
-static ssize_t shmem_file_sendfile(struct file *in_file, loff_t *ppos,
-			 size_t count, read_actor_t actor, void *target)
-{
-	read_descriptor_t desc;
-
-	if (!count)
-		return 0;
-
-	desc.written = 0;
-	desc.count = count;
-	desc.arg.data = target;
-	desc.error = 0;
-
-	do_shmem_file_read(in_file, ppos, &desc, actor);
-	if (desc.written)
-		return desc.written;
-	return desc.error;
-}
-
 static int shmem_statfs(struct dentry *dentry, struct kstatfs *buf)
 {
 	struct shmem_sb_info *sbinfo = SHMEM_SB(dentry->d_sb);
@@ -2386,6 +2376,7 @@ static const struct address_space_operations shmem_aops = {
 	.writepage	= shmem_writepage,
 	.set_page_dirty	= __set_page_dirty_no_writeback,
 #ifdef CONFIG_TMPFS
+	.readpage	= shmem_readpage,
 	.prepare_write	= shmem_prepare_write,
 	.commit_write	= simple_commit_write,
 #endif
@@ -2399,7 +2390,8 @@ static const struct file_operations shmem_file_operations = {
 	.read		= shmem_file_read,
 	.write		= shmem_file_write,
 	.fsync		= simple_sync_file,
-	.sendfile	= shmem_file_sendfile,
+	.splice_read	= generic_file_splice_read,
+	.splice_write	= generic_file_splice_write,
 #endif
 };
 

net/sunrpc/auth_gss/svcauth_gss.c

@@ -853,7 +853,7 @@ unwrap_priv_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct gs
 	u32 priv_len, maj_stat;
 	int pad, saved_len, remaining_len, offset;
 
-	rqstp->rq_sendfile_ok = 0;
+	rqstp->rq_splice_ok = 0;
 
 	priv_len = svc_getnl(&buf->head[0]);
 	if (rqstp->rq_deferred) {

net/sunrpc/svc.c

@@ -814,7 +814,7 @@ svc_process(struct svc_rqst *rqstp)
 	rqstp->rq_res.tail[0].iov_base = NULL;
 	rqstp->rq_res.tail[0].iov_len = 0;
 	/* Will be turned off only in gss privacy case: */
-	rqstp->rq_sendfile_ok = 1;
+	rqstp->rq_splice_ok = 1;
 	/* tcp needs a space for the record length... */
 	if (rqstp->rq_prot == IPPROTO_TCP)
 		svc_putnl(resv, 0);