netdev - [PATCH v4 net-next 17/19] inet: frags: get rid of ipfrag_skb_cb/FRAG

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20180331195900.183604-18-edumazet@google.com>
Date:   Sat, 31 Mar 2018 12:58:58 -0700
From:   Eric Dumazet <edumazet@...gle.com>
To:     "David S . Miller" <davem@...emloft.net>
Cc:     netdev <netdev@...r.kernel.org>, Florian Westphal <fw@...len.de>,
        Herbert Xu <herbert@...dor.apana.org.au>,
        Thomas Graf <tgraf@...g.ch>,
        Jesper Dangaard Brouer <brouer@...hat.com>,
        Alexander Aring <alex.aring@...il.com>,
        Stefan Schmidt <stefan@....samsung.com>,
        Kirill Tkhai <ktkhai@...tuozzo.com>,
        Eric Dumazet <edumazet@...gle.com>,
        Eric Dumazet <eric.dumazet@...il.com>
Subject: [PATCH v4 net-next 17/19] inet: frags: get rid of ipfrag_skb_cb/FRAG_CB

ip_defrag uses skb->cb[] to store the fragment offset, and unfortunately
this integer is currently in a different cache line than skb->next,
meaning that we use two cache lines per skb when finding the insertion point.

By aliasing skb->ip_defrag_offset and skb->dev, we pack all the fields
in a single cache line and save precious memory bandwidth.

Note that after the fast path added by Changli Gao in commit
d6bebca92c66 ("fragment: add fast path for in-order fragments")
this change wont help the fast path, since we still need
to access prev->len (2nd cache line), but will show great
benefits when slow path is entered, since we perform
a linear scan of a potentially long list.

Also, note that this potential long list is an attack vector,
we might consider also using an rb-tree there eventually.

Signed-off-by: Eric Dumazet <edumazet@...gle.com>
---
 include/linux/skbuff.h |  1 +
 net/ipv4/ip_fragment.c | 35 ++++++++++++++---------------------
 2 files changed, 15 insertions(+), 21 deletions(-)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 47082f54ec1f645bed95a37d8c101efd160c60dc..9065477ed255a48f7e01b8a28ea6321cce9127f5 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -672,6 +672,7 @@ struct sk_buff {
 				 * UDP receive path is one user.
 				 */
 				unsigned long		dev_scratch;
+				int			ip_defrag_offset;
 			};
 		};
 		struct rb_node	rbnode; /* used in netem & tcp stack */
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index fb185d9a5cc771778fb650695e2d9894724efc23..994fa70a910f472ebecc336ddd62d1442014eaba 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -57,14 +57,6 @@
  */
 static const char ip_frag_cache_name[] = "ip4-frags";
 
-struct ipfrag_skb_cb
-{
-	struct inet_skb_parm	h;
-	int			offset;
-};
-
-#define FRAG_CB(skb)	((struct ipfrag_skb_cb *)((skb)->cb))
-
 /* Describe an entry in the "incomplete datagrams" queue. */
 struct ipq {
 	struct inet_frag_queue q;
@@ -353,13 +345,13 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
 	 * this fragment, right?
 	 */
 	prev = qp->q.fragments_tail;
-	if (!prev || FRAG_CB(prev)->offset < offset) {
+	if (!prev || prev->ip_defrag_offset < offset) {
 		next = NULL;
 		goto found;
 	}
 	prev = NULL;
 	for (next = qp->q.fragments; next != NULL; next = next->next) {
-		if (FRAG_CB(next)->offset >= offset)
+		if (next->ip_defrag_offset >= offset)
 			break;	/* bingo! */
 		prev = next;
 	}
@@ -370,7 +362,7 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
 	 * any overlaps are eliminated.
 	 */
 	if (prev) {
-		int i = (FRAG_CB(prev)->offset + prev->len) - offset;
+		int i = (prev->ip_defrag_offset + prev->len) - offset;
 
 		if (i > 0) {
 			offset += i;
@@ -387,8 +379,8 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
 
 	err = -ENOMEM;
 
-	while (next && FRAG_CB(next)->offset < end) {
-		int i = end - FRAG_CB(next)->offset; /* overlap is 'i' bytes */
+	while (next && next->ip_defrag_offset < end) {
+		int i = end - next->ip_defrag_offset; /* overlap is 'i' bytes */
 
 		if (i < next->len) {
 			/* Eat head of the next overlapped fragment
@@ -396,7 +388,7 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
 			 */
 			if (!pskb_pull(next, i))
 				goto err;
-			FRAG_CB(next)->offset += i;
+			next->ip_defrag_offset += i;
 			qp->q.meat -= i;
 			if (next->ip_summed != CHECKSUM_UNNECESSARY)
 				next->ip_summed = CHECKSUM_NONE;
@@ -420,7 +412,13 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
 		}
 	}
 
-	FRAG_CB(skb)->offset = offset;
+	/* Note : skb->ip_defrag_offset and skb->dev share the same location */
+	dev = skb->dev;
+	if (dev)
+		qp->iif = dev->ifindex;
+	/* Makes sure compiler wont do silly aliasing games */
+	barrier();
+	skb->ip_defrag_offset = offset;
 
 	/* Insert this fragment in the chain of fragments. */
 	skb->next = next;
@@ -431,11 +429,6 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
 	else
 		qp->q.fragments = skb;
 
-	dev = skb->dev;
-	if (dev) {
-		qp->iif = dev->ifindex;
-		skb->dev = NULL;
-	}
 	qp->q.stamp = skb->tstamp;
 	qp->q.meat += skb->len;
 	qp->ecn |= ecn;
@@ -511,7 +504,7 @@ static int ip_frag_reasm(struct ipq *qp, struct sk_buff *prev,
 	}
 
 	WARN_ON(!head);
-	WARN_ON(FRAG_CB(head)->offset != 0);
+	WARN_ON(head->ip_defrag_offset != 0);
 
 	/* Allocate a new buffer for the datagram. */
 	ihlen = ip_hdrlen(head);
-- 
2.17.0.rc1.321.gba9d0f2565-goog