[PATCH] Introduce sys_splice() system call

This adds support for the sys_splice system call. Using a pipe as a
transport, it can connect to files or sockets (latter as output only).

From the splice.c comments:

   "splice": joining two ropes together by interweaving their strands.

   This is the "extended pipe" functionality, where a pipe is used as
   an arbitrary in-memory buffer. Think of a pipe as a small kernel
   buffer that you can use to transfer data from one end to the other.

   The traditional unix read/write is extended with a "splice()" operation
   that transfers data buffers to or from a pipe buffer.

   Named by Larry McVoy, original implementation from Linus, extended by
   Jens to support splicing to files and fixing the initial implementation
   bugs.

Signed-off-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

arch/i386/kernel/syscall_table.S

@@ -312,3 +312,4 @@ ENTRY(sys_call_table)
 	.long sys_unshare		/* 310 */
 	.long sys_set_robust_list
 	.long sys_get_robust_list
+	.long sys_splice

arch/ia64/kernel/entry.S

@@ -1605,5 +1605,6 @@ sys_call_table:
 	data8 sys_ni_syscall			// reserved for pselect
 	data8 sys_ni_syscall			// 1295 reserved for ppoll
 	data8 sys_unshare
+	data8 sys_splice
 
 	.org sys_call_table + 8*NR_syscalls	// guard against failures to increase NR_syscalls

fs/Makefile

@@ -10,7 +10,7 @@ obj-y :=	open.o read_write.o file_table.o buffer.o  bio.o super.o \
 		ioctl.o readdir.o select.o fifo.o locks.o dcache.o inode.o \
 		attr.o bad_inode.o file.o filesystems.o namespace.o aio.o \
 		seq_file.o xattr.o libfs.o fs-writeback.o mpage.o direct-io.o \
-		ioprio.o pnode.o drop_caches.o
+		ioprio.o pnode.o drop_caches.o splice.o
 
 obj-$(CONFIG_INOTIFY)		+= inotify.o
 obj-$(CONFIG_EPOLL)		+= eventpoll.o

fs/ext2/file.c

@@ -53,6 +53,8 @@ const struct file_operations ext2_file_operations = {
 	.readv		= generic_file_readv,
 	.writev		= generic_file_writev,
 	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
+	.splice_write	= generic_file_splice_write,
 };
 
 #ifdef CONFIG_EXT2_FS_XIP

fs/ext3/file.c

@@ -119,6 +119,8 @@ const struct file_operations ext3_file_operations = {
 	.release	= ext3_release_file,
 	.fsync		= ext3_sync_file,
 	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
+	.splice_write	= generic_file_splice_write,
 };
 
 struct inode_operations ext3_file_inode_operations = {

fs/pipe.c

@@ -15,6 +15,7 @@
 #include <linux/pipe_fs_i.h>
 #include <linux/uio.h>
 #include <linux/highmem.h>
+#include <linux/pagemap.h>
 
 #include <asm/uaccess.h>
 #include <asm/ioctls.h>
@@ -94,11 +95,20 @@ static void anon_pipe_buf_release(struct pipe_inode_info *info, struct pipe_buff
 {
 	struct page *page = buf->page;
 
-	if (info->tmp_page) {
-		__free_page(page);
+	/*
+	 * If nobody else uses this page, and we don't already have a
+	 * temporary page, let's keep track of it as a one-deep
+	 * allocation cache
+	 */
+	if (page_count(page) == 1 && !info->tmp_page) {
+		info->tmp_page = page;
 		return;
 	}
-	info->tmp_page = page;
+
+	/*
+	 * Otherwise just release our reference to it
+	 */
+	page_cache_release(page);
 }
 
 static void *anon_pipe_buf_map(struct file *file, struct pipe_inode_info *info, struct pipe_buffer *buf)
@@ -152,6 +162,11 @@ pipe_readv(struct file *filp, const struct iovec *_iov,
 				chars = total_len;
 
 			addr = ops->map(filp, info, buf);
+			if (IS_ERR(addr)) {
+				if (!ret)
+					ret = PTR_ERR(addr);
+				break;
+			}
 			error = pipe_iov_copy_to_user(iov, addr + buf->offset, chars);
 			ops->unmap(info, buf);
 			if (unlikely(error)) {
@@ -254,8 +269,16 @@ pipe_writev(struct file *filp, const struct iovec *_iov,
 		struct pipe_buf_operations *ops = buf->ops;
 		int offset = buf->offset + buf->len;
 		if (ops->can_merge && offset + chars <= PAGE_SIZE) {
-			void *addr = ops->map(filp, info, buf);
-			int error = pipe_iov_copy_from_user(offset + addr, iov, chars);
+			void *addr;
+			int error;
+
+			addr = ops->map(filp, info, buf);
+			if (IS_ERR(addr)) {
+				error = PTR_ERR(addr);
+				goto out;
+			}
+			error = pipe_iov_copy_from_user(offset + addr, iov,
+							chars);
 			ops->unmap(info, buf);
 			ret = error;
 			do_wakeup = 1;

fs/reiserfs/file.c

@@ -1576,6 +1576,8 @@ const struct file_operations reiserfs_file_operations = {
 	.sendfile = generic_file_sendfile,
 	.aio_read = generic_file_aio_read,
 	.aio_write = reiserfs_aio_write,
+	.splice_read = generic_file_splice_read,
+	.splice_write = generic_file_splice_write,
 };
 
 struct inode_operations reiserfs_file_inode_operations = {

include/asm-i386/unistd.h

@@ -318,8 +318,9 @@
 #define __NR_unshare		310
 #define __NR_set_robust_list	311
 #define __NR_get_robust_list	312
+#define __NR_sys_splice		313
 
-#define NR_syscalls 313
+#define NR_syscalls 314
 
 /*
  * user-visible error numbers are in the range -1 - -128: see

include/asm-ia64/unistd.h

@@ -285,12 +285,13 @@
 #define __NR_faccessat			1293
 /* 1294, 1295 reserved for pselect/ppoll */
 #define __NR_unshare			1296
+#define __NR_splice			1297
 
 #ifdef __KERNEL__
 
 #include <linux/config.h>
 
-#define NR_syscalls			273 /* length of syscall table */
+#define NR_syscalls			274 /* length of syscall table */
 
 #define __ARCH_WANT_SYS_RT_SIGACTION
 

include/asm-powerpc/unistd.h

@@ -301,8 +301,9 @@
 #define __NR_pselect6		280
 #define __NR_ppoll		281
 #define __NR_unshare		282
+#define __NR_splice		283
 
-#define __NR_syscalls		283
+#define __NR_syscalls		284
 
 #ifdef __KERNEL__
 #define __NR__exit __NR_exit

include/asm-x86_64/unistd.h

@@ -609,8 +609,10 @@ __SYSCALL(__NR_unshare,	sys_unshare)
 __SYSCALL(__NR_set_robust_list, sys_set_robust_list)
 #define __NR_get_robust_list	274
 __SYSCALL(__NR_get_robust_list, sys_get_robust_list)
+#define __NR_splice		275
+__SYSCALL(__NR_splice, sys_splice)
 
-#define __NR_syscall_max __NR_get_robust_list
+#define __NR_syscall_max __NR_splice
 
 #ifndef __NO_STUBS
 

include/linux/fs.h

@@ -1032,6 +1032,8 @@ struct file_operations {
 	int (*check_flags)(int);
 	int (*dir_notify)(struct file *filp, unsigned long arg);
 	int (*flock) (struct file *, int, struct file_lock *);
+	ssize_t (*splice_write)(struct inode *, struct file *, size_t, unsigned int);
+	ssize_t (*splice_read)(struct file *, struct inode *, size_t, unsigned int);
 };
 
 struct inode_operations {
@@ -1609,6 +1611,8 @@ extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor
 extern void do_generic_mapping_read(struct address_space *mapping,
 				    struct file_ra_state *, struct file *,
 				    loff_t *, read_descriptor_t *, read_actor_t);
+extern ssize_t generic_file_splice_read(struct file *, struct inode *, size_t, unsigned int);
+extern ssize_t generic_file_splice_write(struct inode *, struct file *, size_t, unsigned int);
 extern void
 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
 extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov, 

include/linux/syscalls.h

@@ -569,5 +569,7 @@ asmlinkage long compat_sys_newfstatat(unsigned int dfd, char __user * filename,
 asmlinkage long compat_sys_openat(unsigned int dfd, const char __user *filename,
 				   int flags, int mode);
 asmlinkage long sys_unshare(unsigned long unshare_flags);
+asmlinkage long sys_splice(int fdin, int fdout, size_t len,
+				unsigned int flags);
 
 #endif

net/socket.c

@@ -119,6 +119,9 @@ static ssize_t sock_writev(struct file *file, const struct iovec *vector,
 static ssize_t sock_sendpage(struct file *file, struct page *page,
 			     int offset, size_t size, loff_t *ppos, int more);
 
+extern ssize_t generic_splice_sendpage(struct inode *inode, struct file *out,
+				size_t len, unsigned int flags);
+
 
 /*
  *	Socket files have a set of 'special' operations as well as the generic file ones. These don't appear
@@ -141,7 +144,8 @@ static struct file_operations socket_file_ops = {
 	.fasync =	sock_fasync,
 	.readv =	sock_readv,
 	.writev =	sock_writev,
-	.sendpage =	sock_sendpage
+	.sendpage =	sock_sendpage,
+	.splice_write = generic_splice_sendpage,
 };
 
 /*