[TCP]: Splice receive support.

Support for network splice receive.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 389 insertions, 0 deletions

diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index b628377..98420f9 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -52,6 +52,7 @@
 #endif
 #include <linux/string.h>
 #include <linux/skbuff.h>
+#include <linux/splice.h>
 #include <linux/cache.h>
 #include <linux/rtnetlink.h>
 #include <linux/init.h>
@@ -71,6 +72,40 @@
 static struct kmem_cache *skbuff_head_cache __read_mostly;
 static struct kmem_cache *skbuff_fclone_cache __read_mostly;
 
+static void sock_pipe_buf_release(struct pipe_inode_info *pipe,
+				  struct pipe_buffer *buf)
+{
+	struct sk_buff *skb = (struct sk_buff *) buf->private;
+
+	kfree_skb(skb);
+}
+
+static void sock_pipe_buf_get(struct pipe_inode_info *pipe,
+				struct pipe_buffer *buf)
+{
+	struct sk_buff *skb = (struct sk_buff *) buf->private;
+
+	skb_get(skb);
+}
+
+static int sock_pipe_buf_steal(struct pipe_inode_info *pipe,
+			       struct pipe_buffer *buf)
+{
+	return 1;
+}
+
+
+/* Pipe buffer operations for a socket. */
+static struct pipe_buf_operations sock_pipe_buf_ops = {
+	.can_merge = 0,
+	.map = generic_pipe_buf_map,
+	.unmap = generic_pipe_buf_unmap,
+	.confirm = generic_pipe_buf_confirm,
+	.release = sock_pipe_buf_release,
+	.steal = sock_pipe_buf_steal,
+	.get = sock_pipe_buf_get,
+};
+
 /*
  *	Keep out-of-line to prevent kernel bloat.
  *	__builtin_return_address is not used because it is not always
@@ -1122,6 +1157,217 @@ fault:
 	return -EFAULT;
 }
 
+/*
+ * Callback from splice_to_pipe(), if we need to release some pages
+ * at the end of the spd in case we error'ed out in filling the pipe.
+ */
+static void sock_spd_release(struct splice_pipe_desc *spd, unsigned int i)
+{
+	struct sk_buff *skb = (struct sk_buff *) spd->partial[i].private;
+
+	kfree_skb(skb);
+}
+
+/*
+ * Fill page/offset/length into spd, if it can hold more pages.
+ */
+static inline int spd_fill_page(struct splice_pipe_desc *spd, struct page *page,
+				unsigned int len, unsigned int offset,
+				struct sk_buff *skb)
+{
+	if (unlikely(spd->nr_pages == PIPE_BUFFERS))
+		return 1;
+
+	spd->pages[spd->nr_pages] = page;
+	spd->partial[spd->nr_pages].len = len;
+	spd->partial[spd->nr_pages].offset = offset;
+	spd->partial[spd->nr_pages].private = (unsigned long) skb_get(skb);
+	spd->nr_pages++;
+	return 0;
+}
+
+/*
+ * Map linear and fragment data from the skb to spd. Returns number of
+ * pages mapped.
+ */
+static int __skb_splice_bits(struct sk_buff *skb, unsigned int *offset,
+			     unsigned int *total_len,
+			     struct splice_pipe_desc *spd)
+{
+	unsigned int nr_pages = spd->nr_pages;
+	unsigned int poff, plen, len, toff, tlen;
+	int headlen, seg;
+
+	toff = *offset;
+	tlen = *total_len;
+	if (!tlen)
+		goto err;
+
+	/*
+	 * if the offset is greater than the linear part, go directly to
+	 * the fragments.
+	 */
+	headlen = skb_headlen(skb);
+	if (toff >= headlen) {
+		toff -= headlen;
+		goto map_frag;
+	}
+
+	/*
+	 * first map the linear region into the pages/partial map, skipping
+	 * any potential initial offset.
+	 */
+	len = 0;
+	while (len < headlen) {
+		void *p = skb->data + len;
+
+		poff = (unsigned long) p & (PAGE_SIZE - 1);
+		plen = min_t(unsigned int, headlen - len, PAGE_SIZE - poff);
+		len += plen;
+
+		if (toff) {
+			if (plen <= toff) {
+				toff -= plen;
+				continue;
+			}
+			plen -= toff;
+			poff += toff;
+			toff = 0;
+		}
+
+		plen = min(plen, tlen);
+		if (!plen)
+			break;
+
+		/*
+		 * just jump directly to update and return, no point
+		 * in going over fragments when the output is full.
+		 */
+		if (spd_fill_page(spd, virt_to_page(p), plen, poff, skb))
+			goto done;
+
+		tlen -= plen;
+	}
+
+	/*
+	 * then map the fragments
+	 */
+map_frag:
+	for (seg = 0; seg < skb_shinfo(skb)->nr_frags; seg++) {
+		const skb_frag_t *f = &skb_shinfo(skb)->frags[seg];
+
+		plen = f->size;
+		poff = f->page_offset;
+
+		if (toff) {
+			if (plen <= toff) {
+				toff -= plen;
+				continue;
+			}
+			plen -= toff;
+			poff += toff;
+			toff = 0;
+		}
+
+		plen = min(plen, tlen);
+		if (!plen)
+			break;
+
+		if (spd_fill_page(spd, f->page, plen, poff, skb))
+			break;
+
+		tlen -= plen;
+	}
+
+done:
+	if (spd->nr_pages - nr_pages) {
+		*offset = 0;
+		*total_len = tlen;
+		return 0;
+	}
+err:
+	return 1;
+}
+
+/*
+ * Map data from the skb to a pipe. Should handle both the linear part,
+ * the fragments, and the frag list. It does NOT handle frag lists within
+ * the frag list, if such a thing exists. We'd probably need to recurse to
+ * handle that cleanly.
+ */
+int skb_splice_bits(struct sk_buff *__skb, unsigned int offset,
+		    struct pipe_inode_info *pipe, unsigned int tlen,
+		    unsigned int flags)
+{
+	struct partial_page partial[PIPE_BUFFERS];
+	struct page *pages[PIPE_BUFFERS];
+	struct splice_pipe_desc spd = {
+		.pages = pages,
+		.partial = partial,
+		.flags = flags,
+		.ops = &sock_pipe_buf_ops,
+		.spd_release = sock_spd_release,
+	};
+	struct sk_buff *skb;
+
+	/*
+	 * I'd love to avoid the clone here, but tcp_read_sock()
+	 * ignores reference counts and unconditonally kills the sk_buff
+	 * on return from the actor.
+	 */
+	skb = skb_clone(__skb, GFP_KERNEL);
+	if (unlikely(!skb))
+		return -ENOMEM;
+
+	/*
+	 * __skb_splice_bits() only fails if the output has no room left,
+	 * so no point in going over the frag_list for the error case.
+	 */
+	if (__skb_splice_bits(skb, &offset, &tlen, &spd))
+		goto done;
+	else if (!tlen)
+		goto done;
+
+	/*
+	 * now see if we have a frag_list to map
+	 */
+	if (skb_shinfo(skb)->frag_list) {
+		struct sk_buff *list = skb_shinfo(skb)->frag_list;
+
+		for (; list && tlen; list = list->next) {
+			if (__skb_splice_bits(list, &offset, &tlen, &spd))
+				break;
+		}
+	}
+
+done:
+	/*
+	 * drop our reference to the clone, the pipe consumption will
+	 * drop the rest.
+	 */
+	kfree_skb(skb);
+
+	if (spd.nr_pages) {
+		int ret;
+
+		/*
+		 * Drop the socket lock, otherwise we have reverse
+		 * locking dependencies between sk_lock and i_mutex
+		 * here as compared to sendfile(). We enter here
+		 * with the socket lock held, and splice_to_pipe() will
+		 * grab the pipe inode lock. For sendfile() emulation,
+		 * we call into ->sendpage() with the i_mutex lock held
+		 * and networking will grab the socket lock.
+		 */
+		release_sock(__skb->sk);
+		ret = splice_to_pipe(pipe, &spd);
+		lock_sock(__skb->sk);
+		return ret;
+	}
+
+	return 0;
+}
+
 /**
  *	skb_store_bits - store bits from kernel buffer to skb
  *	@skb: destination buffer
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index d2f22e7..c75f20b 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -838,6 +838,7 @@ const struct proto_ops inet_stream_ops = {
 	.recvmsg	   = sock_common_recvmsg,
 	.mmap		   = sock_no_mmap,
 	.sendpage	   = tcp_sendpage,
+	.splice_read	   = tcp_splice_read,
 #ifdef CONFIG_COMPAT
 	.compat_setsockopt = compat_sock_common_setsockopt,
 	.compat_getsockopt = compat_sock_common_getsockopt,
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 8e65182..56ed407 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -254,6 +254,10 @@
 #include <linux/poll.h>
 #include <linux/init.h>
 #include <linux/fs.h>
+#include <linux/skbuff.h>
+#include <linux/splice.h>
+#include <linux/net.h>
+#include <linux/socket.h>
 #include <linux/random.h>
 #include <linux/bootmem.h>
 #include <linux/cache.h>
@@ -265,6 +269,7 @@
 #include <net/xfrm.h>
 #include <net/ip.h>
 #include <net/netdma.h>
+#include <net/sock.h>
 
 #include <asm/uaccess.h>
 #include <asm/ioctls.h>
@@ -292,6 +297,15 @@ EXPORT_SYMBOL(tcp_memory_allocated);
 EXPORT_SYMBOL(tcp_sockets_allocated);
 
 /*
+ * TCP splice context
+ */
+struct tcp_splice_state {
+	struct pipe_inode_info *pipe;
+	size_t len;
+	unsigned int flags;
+};
+
+/*
  * Pressure flag: try to collapse.
  * Technical note: it is used by multiple contexts non atomically.
  * All the sk_stream_mem_schedule() is of this nature: accounting
@@ -501,6 +515,120 @@ static inline void tcp_push(struct sock *sk, int flags, int mss_now,
 	}
 }
 
+int tcp_splice_data_recv(read_descriptor_t *rd_desc, struct sk_buff *skb,
+			 unsigned int offset, size_t len)
+{
+	struct tcp_splice_state *tss = rd_desc->arg.data;
+
+	return skb_splice_bits(skb, offset, tss->pipe, tss->len, tss->flags);
+}
+
+static int __tcp_splice_read(struct sock *sk, struct tcp_splice_state *tss)
+{
+	/* Store TCP splice context information in read_descriptor_t. */
+	read_descriptor_t rd_desc = {
+		.arg.data = tss,
+	};
+
+	return tcp_read_sock(sk, &rd_desc, tcp_splice_data_recv);
+}
+
+/**
+ *  tcp_splice_read - splice data from TCP socket to a pipe
+ * @sock:	socket to splice from
+ * @ppos:	position (not valid)
+ * @pipe:	pipe to splice to
+ * @len:	number of bytes to splice
+ * @flags:	splice modifier flags
+ *
+ * Description:
+ *    Will read pages from given socket and fill them into a pipe.
+ *
+ **/
+ssize_t tcp_splice_read(struct socket *sock, loff_t *ppos,
+			struct pipe_inode_info *pipe, size_t len,
+			unsigned int flags)
+{
+	struct sock *sk = sock->sk;
+	struct tcp_splice_state tss = {
+		.pipe = pipe,
+		.len = len,
+		.flags = flags,
+	};
+	long timeo;
+	ssize_t spliced;
+	int ret;
+
+	/*
+	 * We can't seek on a socket input
+	 */
+	if (unlikely(*ppos))
+		return -ESPIPE;
+
+	ret = spliced = 0;
+
+	lock_sock(sk);
+
+	timeo = sock_rcvtimeo(sk, flags & SPLICE_F_NONBLOCK);
+	while (tss.len) {
+		ret = __tcp_splice_read(sk, &tss);
+		if (ret < 0)
+			break;
+		else if (!ret) {
+			if (spliced)
+				break;
+			if (flags & SPLICE_F_NONBLOCK) {
+				ret = -EAGAIN;
+				break;
+			}
+			if (sock_flag(sk, SOCK_DONE))
+				break;
+			if (sk->sk_err) {
+				ret = sock_error(sk);
+				break;
+			}
+			if (sk->sk_shutdown & RCV_SHUTDOWN)
+				break;
+			if (sk->sk_state == TCP_CLOSE) {
+				/*
+				 * This occurs when user tries to read
+				 * from never connected socket.
+				 */
+				if (!sock_flag(sk, SOCK_DONE))
+					ret = -ENOTCONN;
+				break;
+			}
+			if (!timeo) {
+				ret = -EAGAIN;
+				break;
+			}
+			sk_wait_data(sk, &timeo);
+			if (signal_pending(current)) {
+				ret = sock_intr_errno(timeo);
+				break;
+			}
+			continue;
+		}
+		tss.len -= ret;
+		spliced += ret;
+
+		release_sock(sk);
+		lock_sock(sk);
+
+		if (sk->sk_err || sk->sk_state == TCP_CLOSE ||
+		    (sk->sk_shutdown & RCV_SHUTDOWN) || !timeo ||
+		    signal_pending(current))
+			break;
+	}
+
+	release_sock(sk);
+
+	if (spliced)
+		return spliced;
+
+	return ret;
+}
+
 static ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset,
 			 size_t psize, int flags)
 {
@@ -2532,6 +2660,7 @@ EXPORT_SYMBOL(tcp_poll);
 EXPORT_SYMBOL(tcp_read_sock);
 EXPORT_SYMBOL(tcp_recvmsg);
 EXPORT_SYMBOL(tcp_sendmsg);
+EXPORT_SYMBOL(tcp_splice_read);
 EXPORT_SYMBOL(tcp_sendpage);
 EXPORT_SYMBOL(tcp_setsockopt);
 EXPORT_SYMBOL(tcp_shutdown);
diff --git a/net/socket.c b/net/socket.c
index 74784df..92fab9e 100644
--- a/net/socket.c
+++ b/net/socket.c
@@ -112,6 +112,9 @@ static long compat_sock_ioctl(struct file *file,
 static int sock_fasync(int fd, struct file *filp, int on);
 static ssize_t sock_sendpage(struct file *file, struct page *page,
 			     int offset, size_t size, loff_t *ppos, int more);
+static ssize_t sock_splice_read(struct file *file, loff_t *ppos,
+			        struct pipe_inode_info *pipe, 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
@@ -134,6 +137,7 @@ static const struct file_operations socket_file_ops = {
 	.fasync =	sock_fasync,
 	.sendpage =	sock_sendpage,
 	.splice_write = generic_splice_sendpage,
+	.splice_read =	sock_splice_read,
 };
 
 /*
@@ -691,6 +695,15 @@ static ssize_t sock_sendpage(struct file *file, struct page *page,
 	return sock->ops->sendpage(sock, page, offset, size, flags);
 }
 
+static ssize_t sock_splice_read(struct file *file, loff_t *ppos,
+			        struct pipe_inode_info *pipe, size_t len,
+				unsigned int flags)
+{
+	struct socket *sock = file->private_data;
+
+	return sock->ops->splice_read(sock, ppos, pipe, len, flags);
+}
+
 static struct sock_iocb *alloc_sock_iocb(struct kiocb *iocb,
 					 struct sock_iocb *siocb)
 {