[PATCH net-next 1/2] ipv6: remove unnecessary codes in tcp_ipv6.c

From: RongQing.Li <redacted>

opt always equals np->opts, so it is meaningless to define opt, and
check if opt does not equal np->opts and then try to free opt.

Signed-off-by: RongQing.Li <redacted>
---
 net/ipv6/tcp_ipv6.c |   16 +++-------------
 1 files changed, 3 insertions(+), 13 deletions(-)

diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 9c06eaf..6cc67ed 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c

@@ -486,7 +486,6 @@ static int tcp_v6_send_synack(struct sock *sk, struct dst_entry *dst,
 	struct inet6_request_sock *treq = inet6_rsk(req);
 	struct ipv6_pinfo *np = inet6_sk(sk);
 	struct sk_buff * skb;
-	struct ipv6_txoptions *opt = np->opt;
 	int err = -ENOMEM;
 
 	/* First, grab a route. */

@@ -500,13 +499,11 @@ static int tcp_v6_send_synack(struct sock *sk, struct dst_entry *dst,
 
 		fl6->daddr = treq->rmt_addr;
 		skb_set_queue_mapping(skb, queue_mapping);
-		err = ip6_xmit(sk, skb, fl6, opt, np->tclass);
+		err = ip6_xmit(sk, skb, fl6, np->opt, np->tclass);
 		err = net_xmit_eval(err);
 	}
 
 done:
-	if (opt && opt != np->opt)
-		sock_kfree_s(sk, opt, opt->tot_len);
 	return err;
 }
 

@@ -1229,7 +1226,6 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 	struct inet_sock *newinet;
 	struct tcp_sock *newtp;
 	struct sock *newsk;
-	struct ipv6_txoptions *opt;
 #ifdef CONFIG_TCP_MD5SIG
 	struct tcp_md5sig_key *key;
 #endif

@@ -1290,7 +1286,6 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 	}
 
 	treq = inet6_rsk(req);
-	opt = np->opt;
 
 	if (sk_acceptq_is_full(sk))
 		goto out_overflow;

@@ -1359,11 +1354,8 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 	   but we make one more one thing there: reattach optmem
 	   to newsk.
 	 */
-	if (opt) {
-		newnp->opt = ipv6_dup_options(newsk, opt);
-		if (opt != np->opt)
-			sock_kfree_s(sk, opt, opt->tot_len);
-	}
+	if (np->opt)
+		newnp->opt = ipv6_dup_options(newsk, np->opt);
 
 	inet_csk(newsk)->icsk_ext_hdr_len = 0;
 	if (newnp->opt)

@@ -1410,8 +1402,6 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 out_overflow:
 	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
 out_nonewsk:
-	if (opt && opt != np->opt)
-		sock_kfree_s(sk, opt, opt->tot_len);
 	dst_release(dst);
 out:
 	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);

-- 
1.7.1

From: RongQing.Li <redacted>

opt always equals np->opts, so it is meaningless to define opt, and
check if opt does not equal np->opts and then try to free opt.

Signed-off-by: RongQing.Li <redacted>
---
 net/dccp/ipv6.c |   21 +++++----------------
 1 files changed, 5 insertions(+), 16 deletions(-)

diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index 9991be0..02162cf 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c

@@ -239,7 +239,6 @@ static int dccp_v6_send_response(struct sock *sk, struct request_sock *req,
 	struct inet6_request_sock *ireq6 = inet6_rsk(req);
 	struct ipv6_pinfo *np = inet6_sk(sk);
 	struct sk_buff *skb;
-	struct ipv6_txoptions *opt = NULL;
 	struct in6_addr *final_p, final;
 	struct flowi6 fl6;
 	int err = -1;

@@ -255,9 +254,8 @@ static int dccp_v6_send_response(struct sock *sk, struct request_sock *req,
 	fl6.fl6_sport = inet_rsk(req)->loc_port;
 	security_req_classify_flow(req, flowi6_to_flowi(&fl6));
 
-	opt = np->opt;
 
-	final_p = fl6_update_dst(&fl6, opt, &final);
+	final_p = fl6_update_dst(&fl6, np->opt, &final);
 
 	dst = ip6_dst_lookup_flow(sk, &fl6, final_p, false);
 	if (IS_ERR(dst)) {

@@ -274,13 +272,11 @@ static int dccp_v6_send_response(struct sock *sk, struct request_sock *req,
 							 &ireq6->loc_addr,
 							 &ireq6->rmt_addr);
 		fl6.daddr = ireq6->rmt_addr;
-		err = ip6_xmit(sk, skb, &fl6, opt, np->tclass);
+		err = ip6_xmit(sk, skb, &fl6, np->opt, np->tclass);
 		err = net_xmit_eval(err);
 	}
 
 done:
-	if (opt != NULL && opt != np->opt)
-		sock_kfree_s(sk, opt, opt->tot_len);
 	dst_release(dst);
 	return err;
 }

@@ -475,7 +471,6 @@ static struct sock *dccp_v6_request_recv_sock(struct sock *sk,
 	struct inet_sock *newinet;
 	struct dccp6_sock *newdp6;
 	struct sock *newsk;
-	struct ipv6_txoptions *opt;
 
 	if (skb->protocol == htons(ETH_P_IP)) {
 		/*

@@ -520,7 +515,6 @@ static struct sock *dccp_v6_request_recv_sock(struct sock *sk,
 		return newsk;
 	}
 
-	opt = np->opt;
 
 	if (sk_acceptq_is_full(sk))
 		goto out_overflow;

@@ -532,7 +526,7 @@ static struct sock *dccp_v6_request_recv_sock(struct sock *sk,
 		memset(&fl6, 0, sizeof(fl6));
 		fl6.flowi6_proto = IPPROTO_DCCP;
 		fl6.daddr = ireq6->rmt_addr;
-		final_p = fl6_update_dst(&fl6, opt, &final);
+		final_p = fl6_update_dst(&fl6, np->opt, &final);
 		fl6.saddr = ireq6->loc_addr;
 		fl6.flowi6_oif = sk->sk_bound_dev_if;
 		fl6.fl6_dport = inet_rsk(req)->rmt_port;

@@ -597,11 +591,8 @@ static struct sock *dccp_v6_request_recv_sock(struct sock *sk,
 	 * Yes, keeping reference count would be much more clever, but we make
 	 * one more one thing there: reattach optmem to newsk.
 	 */
-	if (opt != NULL) {
-		newnp->opt = ipv6_dup_options(newsk, opt);
-		if (opt != np->opt)
-			sock_kfree_s(sk, opt, opt->tot_len);
-	}
+	if (np->opt != NULL)
+		newnp->opt = ipv6_dup_options(newsk, np->opt);
 
 	inet_csk(newsk)->icsk_ext_hdr_len = 0;
 	if (newnp->opt != NULL)

@@ -627,8 +618,6 @@ out_nonewsk:
 	dst_release(dst);
 out:
 	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENDROPS);
-	if (opt != NULL && opt != np->opt)
-		sock_kfree_s(sk, opt, opt->tot_len);
 	return NULL;
 }
 

-- 
1.7.1

From: roy.qing.li@gmail.com
Date: Mon,  2 Jul 2012 11:18:59 +0800

-	if (opt) {
-		newnp->opt = ipv6_dup_options(newsk, opt);
-		if (opt != np->opt)
-			sock_kfree_s(sk, opt, opt->tot_len);

This is bogus, if we copy the options into a new copy in
ipv6_dup_options() we have to free the old one or else we
leak it.

From: roy.qing.li@gmail.com
Date: Mon,  2 Jul 2012 11:19:00 +0800

-	if (opt != NULL) {
-		newnp->opt = ipv6_dup_options(newsk, opt);
-		if (opt != np->opt)
-			sock_kfree_s(sk, opt, opt->tot_len);
-	}

Same bug as your other patch, this leaks options.

2012/7/2 David Miller [off-list ref]:

From: roy.qing.li@gmail.com
Date: Mon,  2 Jul 2012 11:18:59 +0800

quoted

-     if (opt) {
-             newnp->opt = ipv6_dup_options(newsk, opt);
-             if (opt != np->opt)
-                     sock_kfree_s(sk, opt, opt->tot_len);

This is bogus, if we copy the options into a new copy in
ipv6_dup_options() we have to free the old one or else we
leak it.

Do you mean I should free newnp->opt firstly ?

If I understand it right, I think we do not need to free it. the
process is below:

newsk = tcp_v4_syn_recv_sock(sk, skb, req, dst);
..
newnp = inet6_sk(newsk);
..
memcpy(newnp, np, sizeof(struct ipv6_pinfo));
..
newnp->opt         = NULL;

So newnp->opt is not a effective memory.


Thanks.
-Roy

From: RongQing Li <redacted>
Date: Mon, 2 Jul 2012 13:23:09 +0800

2012/7/2 David Miller [off-list ref]:

quoted

From: roy.qing.li@gmail.com
Date: Mon,  2 Jul 2012 11:18:59 +0800

quoted

-     if (opt) {
-             newnp->opt = ipv6_dup_options(newsk, opt);
-             if (opt != np->opt)
-                     sock_kfree_s(sk, opt, opt->tot_len);

This is bogus, if we copy the options into a new copy in
ipv6_dup_options() we have to free the old one or else we
leak it.

Do you mean I should free newnp->opt firstly ?

If I understand it right, I think we do not need to free it. the
process is below:

newsk = tcp_v4_syn_recv_sock(sk, skb, req, dst);
..
newnp = inet6_sk(newsk);
..
memcpy(newnp, np, sizeof(struct ipv6_pinfo));
..
newnp->opt         = NULL;

So newnp->opt is not a effective memory.

ipv6_dup_options() allocates new memory for the options and this call
statement assigns that new pointer to np->opt.

If you do not free the old (before ipv6_dup_options()) np->opt memory
here, it is lost forever.

On Sun, 2012-07-01 at 20:26 -0700, David Miller wrote:

From: roy.qing.li@gmail.com
Date: Mon,  2 Jul 2012 11:18:59 +0800

quoted

-	if (opt) {
-		newnp->opt = ipv6_dup_options(newsk, opt);
-		if (opt != np->opt)
-			sock_kfree_s(sk, opt, opt->tot_len);

This is bogus, if we copy the options into a new copy in
ipv6_dup_options() we have to free the old one or else we
leak it.

Note that the old one is the np->opt of the listener, not the child.

I dont understand how np->opt could change under us, since we have
the listener socket locked.

If np->opt can change under us, we are doomed and need to add refcounts.

On Mon, 2012-07-02 at 10:13 +0200, Eric Dumazet wrote:

Note that the old one is the np->opt of the listener, not the child.

I dont understand how np->opt could change under us, since we have
the listener socket locked.

If np->opt can change under us, we are doomed and need to add refcounts.

Hmm, it seems net/ipv6/udp.c uses np->opt outside of the
lock_sock()/release_sock(sk) boundary.

On Mon, 2012-07-02 at 11:18 +0800, roy.qing.li@gmail.com wrote:

From: RongQing.Li <redacted>

opt always equals np->opts, so it is meaningless to define opt, and
check if opt does not equal np->opts and then try to free opt.

Signed-off-by: RongQing.Li <redacted>
---
 net/ipv6/tcp_ipv6.c |   16 +++-------------
 1 files changed, 3 insertions(+), 13 deletions(-)

Acked-by: Eric Dumazet <edumazet@google.com>

On Mon, 2012-07-02 at 11:19 +0800, roy.qing.li@gmail.com wrote:

From: RongQing.Li <redacted>

opt always equals np->opts, so it is meaningless to define opt, and
check if opt does not equal np->opts and then try to free opt.

Signed-off-by: RongQing.Li <redacted>
---

Acked-by: Eric Dumazet <edumazet@google.com>

From: Eric Dumazet <redacted>
Date: Mon, 02 Jul 2012 11:07:47 +0200

On Mon, 2012-07-02 at 11:18 +0800, roy.qing.li@gmail.com wrote:

quoted

From: RongQing.Li <redacted>

opt always equals np->opts, so it is meaningless to define opt, and
check if opt does not equal np->opts and then try to free opt.

Signed-off-by: RongQing.Li <redacted>
---
 net/ipv6/tcp_ipv6.c |   16 +++-------------
 1 files changed, 3 insertions(+), 13 deletions(-)

Acked-by: Eric Dumazet <edumazet@google.com>

Ok I now understand better why these changes are correct,
applied.

From: Eric Dumazet <redacted>
Date: Mon, 02 Jul 2012 11:08:50 +0200

On Mon, 2012-07-02 at 11:19 +0800, roy.qing.li@gmail.com wrote:

quoted

From: RongQing.Li <redacted>

opt always equals np->opts, so it is meaningless to define opt, and
check if opt does not equal np->opts and then try to free opt.

Signed-off-by: RongQing.Li <redacted>
---

Acked-by: Eric Dumazet <edumazet@google.com>

Also applied, thank you.