From 49a6fe055739a77ef910c283efb99928423c97a0 Mon Sep 17 00:00:00 2001
From: Eyal Birger <eyal.birger@gmail.com>
Date: Sun, 1 Mar 2015 14:58:25 +0200
Subject: net: bluetooth: compact struct bt_skb_cb by inlining struct
 hci_req_ctrl

struct hci_req_ctrl is never used outside of struct bt_skb_cb;
Inlining it frees 8 bytes on a 64 bit system in skb->cb[] allowing
the addition of more ancillary data.

Signed-off-by: Eyal Birger <eyal.birger@gmail.com>
Reviewed-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/net/bluetooth/bluetooth.h b/include/net/bluetooth/bluetooth.h
index e00455a..0989366 100644
--- a/include/net/bluetooth/bluetooth.h
+++ b/include/net/bluetooth/bluetooth.h
@@ -275,21 +275,17 @@ struct hci_dev;
 
 typedef void (*hci_req_complete_t)(struct hci_dev *hdev, u8 status, u16 opcode);
 
-struct hci_req_ctrl {
-	bool			start;
-	u8			event;
-	hci_req_complete_t	complete;
-};
-
 struct bt_skb_cb {
 	__u8 pkt_type;
 	__u8 incoming;
 	__u16 opcode;
 	__u16 expect;
 	__u8 force_active;
+	bool req_start;
+	u8 req_event;
+	hci_req_complete_t req_complete;
 	struct l2cap_chan *chan;
 	struct l2cap_ctrl control;
-	struct hci_req_ctrl req;
 	bdaddr_t bdaddr;
 	__le16 psm;
 };
diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
index 3322d3f..85a0655 100644
--- a/net/bluetooth/hci_core.c
+++ b/net/bluetooth/hci_core.c
@@ -3517,7 +3517,7 @@ int hci_send_cmd(struct hci_dev *hdev, __u16 opcode, __u32 plen,
 	/* Stand-alone HCI commands must be flagged as
 	 * single-command requests.
 	 */
-	bt_cb(skb)->req.start = true;
+	bt_cb(skb)->req_start = true;
 
 	skb_queue_tail(&hdev->cmd_q, skb);
 	queue_work(hdev->workqueue, &hdev->cmd_work);
@@ -4195,7 +4195,7 @@ static bool hci_req_is_complete(struct hci_dev *hdev)
 	if (!skb)
 		return true;
 
-	return bt_cb(skb)->req.start;
+	return bt_cb(skb)->req_start;
 }
 
 static void hci_resend_last(struct hci_dev *hdev)
@@ -4255,14 +4255,14 @@ void hci_req_cmd_complete(struct hci_dev *hdev, u16 opcode, u8 status)
 	 * command queue (hdev->cmd_q).
 	 */
 	if (hdev->sent_cmd) {
-		req_complete = bt_cb(hdev->sent_cmd)->req.complete;
+		req_complete = bt_cb(hdev->sent_cmd)->req_complete;
 
 		if (req_complete) {
 			/* We must set the complete callback to NULL to
 			 * avoid calling the callback more than once if
 			 * this function gets called again.
 			 */
-			bt_cb(hdev->sent_cmd)->req.complete = NULL;
+			bt_cb(hdev->sent_cmd)->req_complete = NULL;
 
 			goto call_complete;
 		}
@@ -4271,12 +4271,12 @@ void hci_req_cmd_complete(struct hci_dev *hdev, u16 opcode, u8 status)
 	/* Remove all pending commands belonging to this request */
 	spin_lock_irqsave(&hdev->cmd_q.lock, flags);
 	while ((skb = __skb_dequeue(&hdev->cmd_q))) {
-		if (bt_cb(skb)->req.start) {
+		if (bt_cb(skb)->req_start) {
 			__skb_queue_head(&hdev->cmd_q, skb);
 			break;
 		}
 
-		req_complete = bt_cb(skb)->req.complete;
+		req_complete = bt_cb(skb)->req_complete;
 		kfree_skb(skb);
 	}
 	spin_unlock_irqrestore(&hdev->cmd_q.lock, flags);
diff --git a/net/bluetooth/hci_event.c b/net/bluetooth/hci_event.c
index a3fb094..8e8c433 100644
--- a/net/bluetooth/hci_event.c
+++ b/net/bluetooth/hci_event.c
@@ -3106,7 +3106,7 @@ static void hci_cmd_status_evt(struct hci_dev *hdev, struct sk_buff *skb)
 		cancel_delayed_work(&hdev->cmd_timer);
 
 	if (ev->status ||
-	    (hdev->sent_cmd && !bt_cb(hdev->sent_cmd)->req.event))
+	    (hdev->sent_cmd && !bt_cb(hdev->sent_cmd)->req_event))
 		hci_req_cmd_complete(hdev, opcode, ev->status);
 
 	if (ev->ncmd && !test_bit(HCI_RESET, &hdev->flags)) {
@@ -5039,7 +5039,7 @@ void hci_event_packet(struct hci_dev *hdev, struct sk_buff *skb)
 
 	skb_pull(skb, HCI_EVENT_HDR_SIZE);
 
-	if (hdev->sent_cmd && bt_cb(hdev->sent_cmd)->req.event == event) {
+	if (hdev->sent_cmd && bt_cb(hdev->sent_cmd)->req_event == event) {
 		struct hci_command_hdr *cmd_hdr = (void *) hdev->sent_cmd->data;
 		u16 opcode = __le16_to_cpu(cmd_hdr->opcode);
 
diff --git a/net/bluetooth/hci_request.c b/net/bluetooth/hci_request.c
index b59f92c..db2f45a 100644
--- a/net/bluetooth/hci_request.c
+++ b/net/bluetooth/hci_request.c
@@ -55,7 +55,7 @@ int hci_req_run(struct hci_request *req, hci_req_complete_t complete)
 		return -ENODATA;
 
 	skb = skb_peek_tail(&req->cmd_q);
-	bt_cb(skb)->req.complete = complete;
+	bt_cb(skb)->req_complete = complete;
 
 	spin_lock_irqsave(&hdev->cmd_q.lock, flags);
 	skb_queue_splice_tail(&req->cmd_q, &hdev->cmd_q);
@@ -116,9 +116,9 @@ void hci_req_add_ev(struct hci_request *req, u16 opcode, u32 plen,
 	}
 
 	if (skb_queue_empty(&req->cmd_q))
-		bt_cb(skb)->req.start = true;
+		bt_cb(skb)->req_start = true;
 
-	bt_cb(skb)->req.event = event;
+	bt_cb(skb)->req_event = event;
 
 	skb_queue_tail(&req->cmd_q, skb);
 }
diff --git a/net/bluetooth/hci_sock.c b/net/bluetooth/hci_sock.c
index 1d65c5b..f003818 100644
--- a/net/bluetooth/hci_sock.c
+++ b/net/bluetooth/hci_sock.c
@@ -965,7 +965,7 @@ static int hci_sock_sendmsg(struct kiocb *iocb, struct socket *sock,
 			/* Stand-alone HCI commands must be flagged as
 			 * single-command requests.
 			 */
-			bt_cb(skb)->req.start = true;
+			bt_cb(skb)->req_start = true;
 
 			skb_queue_tail(&hdev->cmd_q, skb);
 			queue_work(hdev->workqueue, &hdev->cmd_work);
-- 
cgit v0.10.2


From 6368c235777456ddaeaa61360186d8d03a61cb18 Mon Sep 17 00:00:00 2001
From: Eyal Birger <eyal.birger@gmail.com>
Date: Sun, 1 Mar 2015 14:58:26 +0200
Subject: net: bluetooth: compact struct bt_skb_cb by converting boolean fields
 to bit fields

Convert boolean fields incoming and req_start to bit fields and move
force_active in order save space in bt_skb_cb in an effort to use
a portion of skb->cb[] for storing skb->dropcount.

Signed-off-by: Eyal Birger <eyal.birger@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/net/bluetooth/bluetooth.h b/include/net/bluetooth/bluetooth.h
index 0989366..4500bf8 100644
--- a/include/net/bluetooth/bluetooth.h
+++ b/include/net/bluetooth/bluetooth.h
@@ -277,11 +277,11 @@ typedef void (*hci_req_complete_t)(struct hci_dev *hdev, u8 status, u16 opcode);
 
 struct bt_skb_cb {
 	__u8 pkt_type;
-	__u8 incoming;
+	__u8 force_active;
 	__u16 opcode;
 	__u16 expect;
-	__u8 force_active;
-	bool req_start;
+	__u8 incoming:1;
+	__u8 req_start:1;
 	u8 req_event;
 	hci_req_complete_t req_complete;
 	struct l2cap_chan *chan;
diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
index 85a0655..80f40e8 100644
--- a/net/bluetooth/hci_core.c
+++ b/net/bluetooth/hci_core.c
@@ -3517,7 +3517,7 @@ int hci_send_cmd(struct hci_dev *hdev, __u16 opcode, __u32 plen,
 	/* Stand-alone HCI commands must be flagged as
 	 * single-command requests.
 	 */
-	bt_cb(skb)->req_start = true;
+	bt_cb(skb)->req_start = 1;
 
 	skb_queue_tail(&hdev->cmd_q, skb);
 	queue_work(hdev->workqueue, &hdev->cmd_work);
diff --git a/net/bluetooth/hci_request.c b/net/bluetooth/hci_request.c
index db2f45a..f857e76 100644
--- a/net/bluetooth/hci_request.c
+++ b/net/bluetooth/hci_request.c
@@ -116,7 +116,7 @@ void hci_req_add_ev(struct hci_request *req, u16 opcode, u32 plen,
 	}
 
 	if (skb_queue_empty(&req->cmd_q))
-		bt_cb(skb)->req_start = true;
+		bt_cb(skb)->req_start = 1;
 
 	bt_cb(skb)->req_event = event;
 
diff --git a/net/bluetooth/hci_sock.c b/net/bluetooth/hci_sock.c
index f003818..37198fb 100644
--- a/net/bluetooth/hci_sock.c
+++ b/net/bluetooth/hci_sock.c
@@ -965,7 +965,7 @@ static int hci_sock_sendmsg(struct kiocb *iocb, struct socket *sock,
 			/* Stand-alone HCI commands must be flagged as
 			 * single-command requests.
 			 */
-			bt_cb(skb)->req_start = true;
+			bt_cb(skb)->req_start = 1;
 
 			skb_queue_tail(&hdev->cmd_q, skb);
 			queue_work(hdev->workqueue, &hdev->cmd_work);
-- 
cgit v0.10.2


From 2cfdf9fcb85bf6c0bb4556384ba800b9f443f601 Mon Sep 17 00:00:00 2001
From: Eyal Birger <eyal.birger@gmail.com>
Date: Sun, 1 Mar 2015 14:58:27 +0200
Subject: net: rxrpc: change call to sock_recv_ts_and_drops() on rxrpc recvmsg
 to sock_recv_timestamp()

Commit 3b885787ea4112 ("net: Generalize socket rx gap / receive queue overflow cmsg")
allowed receiving packet dropcount information as a socket level option.
RXRPC sockets recvmsg function was changed to support this by calling
sock_recv_ts_and_drops() instead of sock_recv_timestamp().

However, protocol families wishing to receive dropcount should call
sock_queue_rcv_skb() or set the dropcount specifically (as done
in packet_rcv()). This was not done for rxrpc and thus this feature
never worked on these sockets.

Formalizing this by not calling sock_recv_ts_and_drops() in rxrpc as
part of an effort to move skb->dropcount into skb->cb[]

Signed-off-by: Eyal Birger <eyal.birger@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/rxrpc/ar-recvmsg.c b/net/rxrpc/ar-recvmsg.c
index 4575485..d58ba70 100644
--- a/net/rxrpc/ar-recvmsg.c
+++ b/net/rxrpc/ar-recvmsg.c
@@ -150,7 +150,7 @@ int rxrpc_recvmsg(struct kiocb *iocb, struct socket *sock,
 				       &call->conn->trans->peer->srx, len);
 				msg->msg_namelen = len;
 			}
-			sock_recv_ts_and_drops(msg, &rx->sk, skb);
+			sock_recv_timestamp(msg, &rx->sk, skb);
 		}
 
 		/* receive the message */
-- 
cgit v0.10.2


From 2472d7613bd3bae40a7dc75539c0858d5b9e795a Mon Sep 17 00:00:00 2001
From: Eyal Birger <eyal.birger@gmail.com>
Date: Sun, 1 Mar 2015 14:58:28 +0200
Subject: net: packet: use sockaddr_ll fields as storage for skb original
 length in recvmsg path

As part of an effort to move skb->dropcount to skb->cb[], 4 bytes
of additional room are needed in skb->cb[] in packet sockets.

Store the skb original length in the first two fields of sockaddr_ll
(sll_family and sll_protocol) as they can be derived from the skb when
needed.

Signed-off-by: Eyal Birger <eyal.birger@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 9c28cec..7eea30b 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -216,10 +216,16 @@ static void prb_fill_vlan_info(struct tpacket_kbdq_core *,
 static void packet_flush_mclist(struct sock *sk);
 
 struct packet_skb_cb {
-	unsigned int origlen;
 	union {
 		struct sockaddr_pkt pkt;
-		struct sockaddr_ll ll;
+		union {
+			/* Trick: alias skb original length with
+			 * ll.sll_family and ll.protocol in order
+			 * to save room.
+			 */
+			unsigned int origlen;
+			struct sockaddr_ll ll;
+		};
 	} sa;
 };
 
@@ -1814,9 +1820,7 @@ static int packet_rcv(struct sk_buff *skb, struct net_device *dev,
 		     sizeof(skb->cb));
 
 	sll = &PACKET_SKB_CB(skb)->sa.ll;
-	sll->sll_family = AF_PACKET;
 	sll->sll_hatype = dev->type;
-	sll->sll_protocol = skb->protocol;
 	sll->sll_pkttype = skb->pkt_type;
 	if (unlikely(po->origdev))
 		sll->sll_ifindex = orig_dev->ifindex;
@@ -1825,7 +1829,10 @@ static int packet_rcv(struct sk_buff *skb, struct net_device *dev,
 
 	sll->sll_halen = dev_parse_header(skb, sll->sll_addr);
 
-	PACKET_SKB_CB(skb)->origlen = skb->len;
+	/* sll->sll_family and sll->sll_protocol are set in packet_recvmsg().
+	 * Use their space for storing the original skb length.
+	 */
+	PACKET_SKB_CB(skb)->sa.origlen = skb->len;
 
 	if (pskb_trim(skb, snaplen))
 		goto drop_n_acct;
@@ -2883,6 +2890,7 @@ static int packet_recvmsg(struct kiocb *iocb, struct socket *sock,
 	struct sk_buff *skb;
 	int copied, err;
 	int vnet_hdr_len = 0;
+	unsigned int origlen = 0;
 
 	err = -EINVAL;
 	if (flags & ~(MSG_PEEK|MSG_DONTWAIT|MSG_TRUNC|MSG_CMSG_COMPAT|MSG_ERRQUEUE))
@@ -2982,6 +2990,15 @@ static int packet_recvmsg(struct kiocb *iocb, struct socket *sock,
 	if (err)
 		goto out_free;
 
+	if (sock->type != SOCK_PACKET) {
+		struct sockaddr_ll *sll = &PACKET_SKB_CB(skb)->sa.ll;
+
+		/* Original length was stored in sockaddr_ll fields */
+		origlen = PACKET_SKB_CB(skb)->sa.origlen;
+		sll->sll_family = AF_PACKET;
+		sll->sll_protocol = skb->protocol;
+	}
+
 	sock_recv_ts_and_drops(msg, sk, skb);
 
 	if (msg->msg_name) {
@@ -2993,6 +3010,7 @@ static int packet_recvmsg(struct kiocb *iocb, struct socket *sock,
 			msg->msg_namelen = sizeof(struct sockaddr_pkt);
 		} else {
 			struct sockaddr_ll *sll = &PACKET_SKB_CB(skb)->sa.ll;
+
 			msg->msg_namelen = sll->sll_halen +
 				offsetof(struct sockaddr_ll, sll_addr);
 		}
@@ -3006,7 +3024,7 @@ static int packet_recvmsg(struct kiocb *iocb, struct socket *sock,
 		aux.tp_status = TP_STATUS_USER;
 		if (skb->ip_summed == CHECKSUM_PARTIAL)
 			aux.tp_status |= TP_STATUS_CSUMNOTREADY;
-		aux.tp_len = PACKET_SKB_CB(skb)->origlen;
+		aux.tp_len = origlen;
 		aux.tp_snaplen = skb->len;
 		aux.tp_mac = 0;
 		aux.tp_net = skb_network_offset(skb);
-- 
cgit v0.10.2


From b4772ef879a8f7d8c56118c2ae5a296fcf6f81d2 Mon Sep 17 00:00:00 2001
From: Eyal Birger <eyal.birger@gmail.com>
Date: Sun, 1 Mar 2015 14:58:29 +0200
Subject: net: use common macro for assering skb->cb[] available size in
 protocol families

As part of an effort to move skb->dropcount to skb->cb[] use a common
macro in protocol families using skb->cb[] for ancillary data to
validate available room in skb->cb[].

Signed-off-by: Eyal Birger <eyal.birger@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/net/sock.h b/include/net/sock.h
index ab186b1..a2502d2 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -2078,6 +2078,9 @@ static inline int sock_intr_errno(long timeo)
 	return timeo == MAX_SCHEDULE_TIMEOUT ? -ERESTARTSYS : -EINTR;
 }
 
+#define sock_skb_cb_check_size(size) \
+	BUILD_BUG_ON((size) > FIELD_SIZEOF(struct sk_buff, cb))
+
 void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk,
 			   struct sk_buff *skb);
 void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk,
diff --git a/net/bluetooth/af_bluetooth.c b/net/bluetooth/af_bluetooth.c
index ce22e0c..4b904c9 100644
--- a/net/bluetooth/af_bluetooth.c
+++ b/net/bluetooth/af_bluetooth.c
@@ -711,10 +711,9 @@ EXPORT_SYMBOL_GPL(bt_debugfs);
 
 static int __init bt_init(void)
 {
-	struct sk_buff *skb;
 	int err;
 
-	BUILD_BUG_ON(sizeof(struct bt_skb_cb) > sizeof(skb->cb));
+	sock_skb_cb_check_size(sizeof(struct bt_skb_cb));
 
 	BT_INFO("Core ver %s", VERSION);
 
diff --git a/net/can/bcm.c b/net/can/bcm.c
index ee9ffd9..d559f92 100644
--- a/net/can/bcm.c
+++ b/net/can/bcm.c
@@ -328,7 +328,7 @@ static void bcm_send_to_user(struct bcm_op *op, struct bcm_msg_head *head,
 	 *  containing the interface index.
 	 */
 
-	BUILD_BUG_ON(sizeof(skb->cb) < sizeof(struct sockaddr_can));
+	sock_skb_cb_check_size(sizeof(struct sockaddr_can));
 	addr = (struct sockaddr_can *)skb->cb;
 	memset(addr, 0, sizeof(*addr));
 	addr->can_family  = AF_CAN;
diff --git a/net/can/raw.c b/net/can/raw.c
index 00c13ef..94601b7 100644
--- a/net/can/raw.c
+++ b/net/can/raw.c
@@ -95,8 +95,8 @@ struct raw_sock {
  */
 static inline unsigned int *raw_flags(struct sk_buff *skb)
 {
-	BUILD_BUG_ON(sizeof(skb->cb) <= (sizeof(struct sockaddr_can) +
-					 sizeof(unsigned int)));
+	sock_skb_cb_check_size(sizeof(struct sockaddr_can) +
+			       sizeof(unsigned int));
 
 	/* return pointer after struct sockaddr_can */
 	return (unsigned int *)(&((struct sockaddr_can *)skb->cb)[1]);
@@ -135,7 +135,7 @@ static void raw_rcv(struct sk_buff *oskb, void *data)
 	 *  containing the interface index.
 	 */
 
-	BUILD_BUG_ON(sizeof(skb->cb) < sizeof(struct sockaddr_can));
+	sock_skb_cb_check_size(sizeof(struct sockaddr_can));
 	addr = (struct sockaddr_can *)skb->cb;
 	memset(addr, 0, sizeof(*addr));
 	addr->can_family  = AF_CAN;
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index d2e49ba..4ce954c 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1675,7 +1675,7 @@ static int __init inet_init(void)
 	struct list_head *r;
 	int rc = -EINVAL;
 
-	BUILD_BUG_ON(sizeof(struct inet_skb_parm) > FIELD_SIZEOF(struct sk_buff, cb));
+	sock_skb_cb_check_size(sizeof(struct inet_skb_parm));
 
 	rc = proto_register(&tcp_prot, 1);
 	if (rc)
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 9d72a0f..4b57ea8 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -3005,12 +3005,11 @@ static void __init tcp_init_mem(void)
 
 void __init tcp_init(void)
 {
-	struct sk_buff *skb = NULL;
 	unsigned long limit;
 	int max_rshare, max_wshare, cnt;
 	unsigned int i;
 
-	BUILD_BUG_ON(sizeof(struct tcp_skb_cb) > sizeof(skb->cb));
+	sock_skb_cb_check_size(sizeof(struct tcp_skb_cb));
 
 	percpu_counter_init(&tcp_sockets_allocated, 0, GFP_KERNEL);
 	percpu_counter_init(&tcp_orphan_count, 0, GFP_KERNEL);
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index e8c4400..6bafcc2 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -824,7 +824,7 @@ static int __init inet6_init(void)
 	struct list_head *r;
 	int err = 0;
 
-	BUILD_BUG_ON(sizeof(struct inet6_skb_parm) > FIELD_SIZEOF(struct sk_buff, cb));
+	sock_skb_cb_check_size(sizeof(struct inet6_skb_parm));
 
 	/* Register the socket-side information for inet6_create.  */
 	for (r = &inetsw6[0]; r < &inetsw6[SOCK_MAX]; ++r)
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 7eea30b..9cdb414 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -1816,8 +1816,7 @@ static int packet_rcv(struct sk_buff *skb, struct net_device *dev,
 		skb = nskb;
 	}
 
-	BUILD_BUG_ON(sizeof(*PACKET_SKB_CB(skb)) + MAX_ADDR_LEN - 8 >
-		     sizeof(skb->cb));
+	sock_skb_cb_check_size(sizeof(*PACKET_SKB_CB(skb)) + MAX_ADDR_LEN - 8);
 
 	sll = &PACKET_SKB_CB(skb)->sa.ll;
 	sll->sll_hatype = dev->type;
diff --git a/net/sctp/protocol.c b/net/sctp/protocol.c
index 8f34b27..53b7acd 100644
--- a/net/sctp/protocol.c
+++ b/net/sctp/protocol.c
@@ -1322,8 +1322,7 @@ static __init int sctp_init(void)
 	int max_share;
 	int order;
 
-	BUILD_BUG_ON(sizeof(struct sctp_ulpevent) >
-		     sizeof(((struct sk_buff *) 0)->cb));
+	sock_skb_cb_check_size(sizeof(struct sctp_ulpevent));
 
 	/* Allocate bind_bucket and chunk caches. */
 	status = -ENOBUFS;
-- 
cgit v0.10.2


From 3bc3b96f3b455bd14a8ccd83ffffc85625aba641 Mon Sep 17 00:00:00 2001
From: Eyal Birger <eyal.birger@gmail.com>
Date: Sun, 1 Mar 2015 14:58:30 +0200
Subject: net: add common accessor for setting dropcount on packets

As part of an effort to move skb->dropcount to skb->cb[], use
a common function in order to set dropcount in struct sk_buff.

Signed-off-by: Eyal Birger <eyal.birger@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/net/sock.h b/include/net/sock.h
index a2502d2..0996fe4 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -2081,6 +2081,12 @@ static inline int sock_intr_errno(long timeo)
 #define sock_skb_cb_check_size(size) \
 	BUILD_BUG_ON((size) > FIELD_SIZEOF(struct sk_buff, cb))
 
+static inline void
+sock_skb_set_dropcount(const struct sock *sk, struct sk_buff *skb)
+{
+	skb->dropcount = atomic_read(&sk->sk_drops);
+}
+
 void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk,
 			   struct sk_buff *skb);
 void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk,
diff --git a/net/core/sock.c b/net/core/sock.c
index 93c8b20..9c74fc8 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -466,7 +466,7 @@ int sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
 	skb_dst_force(skb);
 
 	spin_lock_irqsave(&list->lock, flags);
-	skb->dropcount = atomic_read(&sk->sk_drops);
+	sock_skb_set_dropcount(sk, skb);
 	__skb_queue_tail(list, skb);
 	spin_unlock_irqrestore(&list->lock, flags);
 
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 9cdb414..9db8369 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -1845,7 +1845,7 @@ static int packet_rcv(struct sk_buff *skb, struct net_device *dev,
 
 	spin_lock(&sk->sk_receive_queue.lock);
 	po->stats.stats1.tp_packets++;
-	skb->dropcount = atomic_read(&sk->sk_drops);
+	sock_skb_set_dropcount(sk, skb);
 	__skb_queue_tail(&sk->sk_receive_queue, skb);
 	spin_unlock(&sk->sk_receive_queue.lock);
 	sk->sk_data_ready(sk);
-- 
cgit v0.10.2


From 744d5a3e9fe2690dd85d9991dbb078301694658b Mon Sep 17 00:00:00 2001
From: Eyal Birger <eyal.birger@gmail.com>
Date: Sun, 1 Mar 2015 14:58:31 +0200
Subject: net: move skb->dropcount to skb->cb[]

Commit 977750076d98 ("af_packet: add interframe drop cmsg (v6)")
unionized skb->mark and skb->dropcount in order to allow recording
of the socket drop count while maintaining struct sk_buff size.

skb->dropcount was introduced since there was no available room
in skb->cb[] in packet sockets. However, its introduction led to
the inability to export skb->mark, or any other aliased field to
userspace if so desired.

Moving the dropcount metric to skb->cb[] eliminates this problem
at the expense of 4 bytes less in skb->cb[] for protocol families
using it.

Signed-off-by: Eyal Birger <eyal.birger@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index d898b32..bba1330 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -492,7 +492,6 @@ static inline u32 skb_mstamp_us_delta(const struct skb_mstamp *t1,
   *	@napi_id: id of the NAPI struct this skb came from
  *	@secmark: security marking
  *	@mark: Generic packet mark
- *	@dropcount: total number of sk_receive_queue overflows
  *	@vlan_proto: vlan encapsulation protocol
  *	@vlan_tci: vlan tag control information
  *	@inner_protocol: Protocol (encapsulation)
@@ -641,7 +640,6 @@ struct sk_buff {
 #endif
 	union {
 		__u32		mark;
-		__u32		dropcount;
 		__u32		reserved_tailroom;
 	};
 
diff --git a/include/net/sock.h b/include/net/sock.h
index 0996fe4..38369d3 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -2078,13 +2078,27 @@ static inline int sock_intr_errno(long timeo)
 	return timeo == MAX_SCHEDULE_TIMEOUT ? -ERESTARTSYS : -EINTR;
 }
 
+struct sock_skb_cb {
+	u32 dropcount;
+};
+
+/* Store sock_skb_cb at the end of skb->cb[] so protocol families
+ * using skb->cb[] would keep using it directly and utilize its
+ * alignement guarantee.
+ */
+#define SOCK_SKB_CB_OFFSET ((FIELD_SIZEOF(struct sk_buff, cb) - \
+			    sizeof(struct sock_skb_cb)))
+
+#define SOCK_SKB_CB(__skb) ((struct sock_skb_cb *)((__skb)->cb + \
+			    SOCK_SKB_CB_OFFSET))
+
 #define sock_skb_cb_check_size(size) \
-	BUILD_BUG_ON((size) > FIELD_SIZEOF(struct sk_buff, cb))
+	BUILD_BUG_ON((size) > SOCK_SKB_CB_OFFSET)
 
 static inline void
 sock_skb_set_dropcount(const struct sock *sk, struct sk_buff *skb)
 {
-	skb->dropcount = atomic_read(&sk->sk_drops);
+	SOCK_SKB_CB(skb)->dropcount = atomic_read(&sk->sk_drops);
 }
 
 void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk,
diff --git a/net/socket.c b/net/socket.c
index bbedbfc..b78cf60 100644
--- a/net/socket.c
+++ b/net/socket.c
@@ -731,9 +731,9 @@ EXPORT_SYMBOL_GPL(__sock_recv_wifi_status);
 static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk,
 				   struct sk_buff *skb)
 {
-	if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && skb->dropcount)
+	if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount)
 		put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL,
-			sizeof(__u32), &skb->dropcount);
+			sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount);
 }
 
 void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk,
-- 
cgit v0.10.2