Skip to contentSkip to conversation

tcp: bugs and cleanup for 2.6.33

PATCH In review William Allen Simpson 3 participants 13 messages Feb 25, 2010 – Feb 26, 2010

Subsystems (4)
Networking Network drivers TCP Networking IPsec
lore
Share & tools
lore

William Allen Simpson

author
#0 lore raw

cover tcp: bugs and cleanup for 2.6.33

I'd have thought that there would be greater interest about patching crashing bugs, signed versus unsigned (underflow) bugs, TCP DoS bugs, TCP data corruption, and TCP performance problems....

Combination of patches reported in October, November, December, January, and February, for 2.6.32 and now 2.6.33.

This code has had previous review and several months of limited testing.

Some portions were removed during the various TCPCT part 1 patch splits, then were cut off by the sudden unexpected end of that merge window. [03 Dec 2009] I've restarted the sub-numbering (again).

Of particular interest are the TCPCT header extensions that already appear in the next phase of testing with other platforms. These patches allow correct reception without data corruption.

The remainder of the original TCPCT part 2 will be merged with part 3.

[Updated to 2010 Feb 24 2.6.33 release.]

author
#1 lore raw

Redefine two TCP header functions to accept TCP header pointer. When subtracting, return signed int to allow error checking.

These functions will be used in subsequent patches that implement additional features.

Signed-off-by [email protected]
---
 include/linux/tcp.h |   12 ++++++++++++
 1 files changed, 12 insertions(+), 0 deletions(-)
include/linux/tcp.h +12 -0
index 7fee8a4..d0133cf 100644
@@ -223,6 +223,18 @@ static inline unsigned int tcp_optlen(const struct sk_buff *skb)
 	return (tcp_hdr(skb)->doff - 5) * 4;
 }
 
+/* Length of fixed header plus standard options. */
+static inline unsigned int tcp_header_len_th(const struct tcphdr *th)
+{
+	return th->doff * 4;
+}
+
+/* Length of standard options only.  This could be negative. */
+static inline int tcp_option_len_th(const struct tcphdr *th)
+{
+	return (int)(th->doff * 4) - sizeof(*th);
+}
+
 /* This defines a selective acknowledgement block. */
 struct tcp_sack_block_wire {
 	__be32	start_seq;
Signature · 1 line
1.6.3.3
#2 lore raw
Quoted 2 lines · William Allen Simpson

Redefine two TCP header functions to accept TCP header pointer. When subtracting, return signed int to allow error checking.

This patch only adds functions, so why do you say 'redefine'?

Quoted 2 lines

These functions will be used in subsequent patches that implement additional features.

[...]

include/linux/tcp.h static inline unsigned int tcp_optlen(const struct sk_buff *skb) · line 236

Quoted 5 lines
include/linux/tcp.h +5 -0
+/* Length of standard options only.  This could be negative. */
+static inline int tcp_option_len_th(const struct tcphdr *th)
+{
+       return (int)(th->doff * 4) - sizeof(*th);
+}

I don't see the point of this cast; the left operand of the subtraction will in any case be promoted to size_t to match the right operand.

Did you mean

	return (int)(th->doff * 4 - sizeof(*th));

?

Ben.

Signature · 3 lines
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
author
#4 lore raw

PATCH v3 1/7 net: tcp_header_len_th and tcp_option_len_th

In reply to William Allen Simpson

Redefine two TCP header functions to accept TCP header pointer. When subtracting, return signed int to allow error checking.

These functions will be used in subsequent patches that implement additional features.

Signed-off-by [email protected]
---
 include/linux/tcp.h |   12 ++++++++++++
 1 files changed, 12 insertions(+), 0 deletions(-)
include/linux/tcp.h +12 -0
index 7fee8a4..d0133cf 100644
@@ -223,6 +223,18 @@ static inline unsigned int tcp_optlen(const struct sk_buff *skb)
 	return (tcp_hdr(skb)->doff - 5) * 4;
 }
 
+/* Length of fixed header plus standard options. */
+static inline unsigned int tcp_header_len_th(const struct tcphdr *th)
+{
+	return th->doff * 4;
+}
+
+/* Length of standard options only.  This could be negative. */
+static inline int tcp_option_len_th(const struct tcphdr *th)
+{
+	return (int)(th->doff * 4) - sizeof(*th);
+}
+
 /* This defines a selective acknowledgement block. */
 struct tcp_sack_block_wire {
 	__be32	start_seq;
Signature · 1 line
1.6.3.3
author
#5 lore raw

PATCH v7 7/7 TCPCT part 2g: parse cookie pair and 64-bit timestamp

Every bit is sacred. Use as few bits as possible in tcp_options_received. Group related timestamp flag bits for cache line memory efficiency.

Fix #define spacing for TCP options.

Define and parse 64-bit timestamp extended option (and minor cleanup). However, only 32-bits are used at this time (permitted by specification).

Parse cookie pair extended option (previously defined).

Handle header extension.

Fix initialization in tcp_minisocks.

Requires:
net: tcp_header_len_th and tcp_option_len_th
TCPCT part 2f: cleanup tcp_parse_options

Signed-off-by [email protected]
---
 include/linux/tcp.h      |   12 ++++-
 include/net/tcp.h        |   45 +++++++++--------
 net/ipv4/tcp_input.c     |  123 ++++++++++++++++++++++++++++++++++++++++++----
 net/ipv4/tcp_minisocks.c |    6 ++
 4 files changed, 152 insertions(+), 34 deletions(-)
include/linux/tcp.h +10 -2
index 2987ee8..05fa9b2 100644
@@ -247,7 +247,7 @@ struct tcp_options_received {
 	u32	ts_recent;	/* Time stamp to echo next		*/
 	u32	rcv_tsval;	/* Time stamp value             	*/
 	u32	rcv_tsecr;	/* Time stamp echo reply        	*/
-	u16 	saw_tstamp : 1,	/* Saw TIMESTAMP on last packet		*/
+	u16 	tstamp64_ok:1,	/* Verified with cookie pair		*/
 		tstamp_ok : 1,	/* TIMESTAMP seen on SYN packet		*/
 		dsack : 1,	/* D-SACK is scheduled			*/
 		wscale_ok : 1,	/* Wscale seen on SYN packet		*/

@@ -260,13 +260,21 @@ struct tcp_options_received {
 	u8	num_sacks;	/* Number of SACK blocks		*/
 	u16	user_mss;	/* mss requested by user in ioctl	*/
 	u16	mss_clamp;	/* Maximal mss, negotiated at connection setup */
+
+	/* When the options are extended beyond the standard 40 bytes,
+	 * this holds the additional data offset (up to 1,020 bytes).
+	 */
+	u8	extended;	/* in 32-bit words			*/
+	u8	saw_tstamp64:1,	/* 64-bit TIMESTAMP seen on last packet	*/
+		saw_tstamp:1,	/* TIMESTAMP seen on last packet	*/
+		__unused:6;
 };
 
 static inline void tcp_clear_options(struct tcp_options_received *rx_opt)
 {
+	rx_opt->tstamp64_ok = 0;
 	rx_opt->tstamp_ok = rx_opt->sack_ok = 0;
 	rx_opt->wscale_ok = rx_opt->snd_wscale = 0;
-	rx_opt->cookie_plus = 0;
 }
 
 /* This is the max number of SACKS that we'll generate and process. It's safe
include/net/tcp.h +23 -22

62 lines not on this page · raw message

net/ipv4/tcp_input.c +113 -10

202 lines not on this page · raw message

net/ipv4/tcp_minisocks.c +6 -0
index 0f1b409..2240012 100644
@@ -109,6 +109,9 @@ tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb,
 		}
 	} else {
 		/* otherwise initialized by tcp_parse_options() */
+		tmp_opt.cookie_plus = 0;
+		tmp_opt.extended = 0;
+		tmp_opt.saw_tstamp64 = 0; /* false */
 		tmp_opt.saw_tstamp = 0; /* false */
 	}
 

@@ -550,6 +553,9 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
 		}
 	} else {
 		/* otherwise initialized by tcp_parse_options() */
+		tmp_opt.cookie_plus = 0;
+		tmp_opt.extended = 0;
+		tmp_opt.saw_tstamp64 = 0; /* false */
 		tmp_opt.saw_tstamp = 0; /* false */
 	}
Signature · 1 line
1.6.3.3
author
#6 lore raw

PATCH v3 2/7 net: remove old tcp_optlen function

In reply to William Allen Simpson

The tcp_optlen() function returns a potential *negative* unsigned.

In the only two existing files using the old tcp_optlen() function, clean up confusing and inconsistent mixing of both byte and word offsets, and other coding style issues. Document assumptions.

Quoth David Miller:
This is transmit, and the packets can only come from the Linux TCP stack, not some external entity.

You're being way too anal here, and adding these checks to drivers would be just a lot of rediculious bloat. [sic]

Therefore, there are *no* checks for bad TCP and IP header sizes, nor any semantic changes. The drivers should function exactly as existing.

No response from testers in 19+ weeks.

Requires:
net: tcp_header_len_th and tcp_option_len_th

Signed-off-by [email protected]
CC Michael Chan <[email protected]>
---
 drivers/net/bnx2.c  |   29 +++++++++++++-----------
 drivers/net/tg3.c   |   60 +++++++++++++++++++++++---------------------------
 include/linux/tcp.h |    5 ----
 3 files changed, 44 insertions(+), 50 deletions(-)
drivers/net/bnx2.c +16 -13
index 65df1de..45452c5 100644
@@ -6352,6 +6352,8 @@ bnx2_vlan_rx_register(struct net_device *dev, struct vlan_group *vlgrp)
 /* Called with netif_tx_lock.
  * bnx2_tx_int() runs without netif_tx_lock unless it needs to call
  * netif_wake_queue().
+ *
+ * No TCP or IP length checking, per David Miller (see commit log).
  */
 static netdev_tx_t
 bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev)

@@ -6396,19 +6398,19 @@ bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev)
 			(TX_BD_FLAGS_VLAN_TAG | (vlan_tx_tag_get(skb) << 16));
 	}
 #endif
-	if ((mss = skb_shinfo(skb)->gso_size)) {
-		u32 tcp_opt_len;
-		struct iphdr *iph;
+	mss = skb_shinfo(skb)->gso_size;
+	if (mss != 0) {
+		struct tcphdr *th = tcp_hdr(skb);
+		int tcp_opt_words = th->doff - (sizeof(*th) >> 2);
+		/* assumes positive tcp_opt_words without checking */
 
 		vlan_tag_flags |= TX_BD_FLAGS_SW_LSO;
 
-		tcp_opt_len = tcp_optlen(skb);
-
 		if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6) {
 			u32 tcp_off = skb_transport_offset(skb) -
 				      sizeof(struct ipv6hdr) - ETH_HLEN;
 
-			vlan_tag_flags |= ((tcp_opt_len >> 2) << 8) |
+			vlan_tag_flags |= (tcp_opt_words << 8) |
 					  TX_BD_FLAGS_SW_FLAGS;
 			if (likely(tcp_off == 0))
 				vlan_tag_flags &= ~TX_BD_FLAGS_TCP6_OFF0_MSK;

@@ -6421,14 +6423,15 @@ bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev)
 				mss |= (tcp_off & 0xc) << TX_BD_TCP6_OFF2_SHL;
 			}
 		} else {
-			iph = ip_hdr(skb);
-			if (tcp_opt_len || (iph->ihl > 5)) {
-				vlan_tag_flags |= ((iph->ihl - 5) +
-						   (tcp_opt_len >> 2)) << 8;
-			}
+			struct iphdr *iph = ip_hdr(skb);
+			int ip_opt_words = iph->ihl - (sizeof(*iph) >> 2);
+			/* assumes positive ip_opt_words without checking */
+			int opt_words = ip_opt_words + tcp_opt_words;
+
+			if (opt_words > 0)
+				vlan_tag_flags |= opt_words << 8;
 		}
-	} else
-		mss = 0;
+	}
 
 	mapping = pci_map_single(bp->pdev, skb->data, len, PCI_DMA_TODEVICE);
 	if (pci_dma_mapping_error(bp->pdev, mapping)) {
drivers/net/tg3.c +28 -32

126 lines not on this page · raw message

include/linux/tcp.h +0 -5
index d0133cf..74728f7 100644
@@ -218,11 +218,6 @@ static inline unsigned int tcp_hdrlen(const struct sk_buff *skb)
 	return tcp_hdr(skb)->doff * 4;
 }
 
-static inline unsigned int tcp_optlen(const struct sk_buff *skb)
-{
-	return (tcp_hdr(skb)->doff - 5) * 4;
-}
-
 /* Length of fixed header plus standard options. */
 static inline unsigned int tcp_header_len_th(const struct tcphdr *th)
 {
Signature · 1 line
1.6.3.3
#7 lore raw

Re: [PATCH v3 2/7] net: remove old tcp_optlen function

Quoted 10 lines · William Allen Simpson

The tcp_optlen() function returns a potential *negative* unsigned.

In the only two existing files using the old tcp_optlen() function, clean up confusing and inconsistent mixing of both byte and word offsets, and other coding style issues. Document assumptions.

Quoth David Miller:
This is transmit, and the packets can only come from the Linux TCP stack, not some external entity.

Quoted 7 lines

You're being way too anal here, and adding these checks to drivers would be just a lot of rediculious bloat. [sic]

Therefore, there are *no* checks for bad TCP and IP header sizes, nor any semantic changes. The drivers should function exactly as existing.

No response from testers in 19+ weeks.

There's no need for these passive-aggressive comments and they certainly won't get your patches merged any faster.

Quoted 26 lines

Requires:
net: tcp_header_len_th and tcp_option_len_th

Signed-off-by [email protected]
CC Michael Chan <[email protected]>
---
  drivers/net/bnx2.c  |   29 +++++++++++++-----------
  drivers/net/tg3.c   |   60 +++++++++++++++++++++++---------------------------
  include/linux/tcp.h |    5 ----
  3 files changed, 44 insertions(+), 50 deletions(-)
drivers/net/bnx2.c +2 -0
index 65df1de..45452c5 100644
@@ -6352,6 +6352,8 @@ bnx2_vlan_rx_register(struct net_device *dev, struct vlan_group *vlgrp)
 /* Called with netif_tx_lock.
  * bnx2_tx_int() runs without netif_tx_lock unless it needs to call
  * netif_wake_queue().
+ *
+ * No TCP or IP length checking, per David Miller (see commit log).
  */
 static netdev_tx_t
 bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev)

@@ -6396,19 +6398,19 @@ bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev)
 			(TX_BD_FLAGS_VLAN_TAG | (vlan_tx_tag_get(skb) << 16));

drivers/net/bnx2.c bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev) · line 6405

Quoted 10 lines
drivers/net/bnx2.c +5 -3
 	}
 #endif
-	if ((mss = skb_shinfo(skb)->gso_size)) {
-		u32 tcp_opt_len;
-		struct iphdr *iph;
+	mss = skb_shinfo(skb)->gso_size;
+	if (mss != 0) {
+		struct tcphdr *th = tcp_hdr(skb);
+		int tcp_opt_words = th->doff - (sizeof(*th) >> 2);
+		/* assumes positive tcp_opt_words without checking */

It would be far more helpful to add a comment to the top of this block that notes that skbs with gso_size set are known to have valid headers.

drivers/net/bnx2.c bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev) · line 6406

Quoted 4 lines
drivers/net/bnx2.c +0 -2
 		vlan_tag_flags |= TX_BD_FLAGS_SW_LSO;
 
-		tcp_opt_len = tcp_optlen(skb);
-

How is this an improvement?

Quoted 15 lines
drivers/net/bnx2.c +1 -3
 		if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6) {
 			u32 tcp_off = skb_transport_offset(skb) -
 				      sizeof(struct ipv6hdr) - ETH_HLEN;
 
-			vlan_tag_flags |= ((tcp_opt_len >> 2) << 8) |
+			vlan_tag_flags |= (tcp_opt_words << 8) |
 					  TX_BD_FLAGS_SW_FLAGS;
 			if (likely(tcp_off == 0))
 				vlan_tag_flags &= ~TX_BD_FLAGS_TCP6_OFF0_MSK;

@@ -6421,14 +6423,15 @@ bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev)
 				mss |= (tcp_off & 0xc) << TX_BD_TCP6_OFF2_SHL;
 			}
 		} else {
-			iph = ip_hdr(skb);
-			if (tcp_opt_len || (iph->ihl > 5)) {

drivers/net/bnx2.c bnx2_start_xmit(struct sk_buff *skb, struct net_device *dev) · line 6432

Quoted 10 lines
drivers/net/bnx2.c +7 -3
-				vlan_tag_flags |= ((iph->ihl - 5) +
-						   (tcp_opt_len >> 2)) << 8;
-			}
+			struct iphdr *iph = ip_hdr(skb);
+			int ip_opt_words = iph->ihl - (sizeof(*iph) >> 2);
+			/* assumes positive ip_opt_words without checking */
+			int opt_words = ip_opt_words + tcp_opt_words;
+
+			if (opt_words > 0)
+				vlan_tag_flags |= opt_words << 8;

[...]

That *is* clearer.

Ben.

Signature · 3 lines
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
author
#8 lore raw

PATCH v5 3/7 tcp: harmonize tcp_vx_rcv header length assumptions

In reply to William Allen Simpson

Harmonize tcp_v4_rcv() and tcp_v6_rcv() -- better document tcp doff and header length assumptions, and carefully compare implementations.

Reduces multiply/shifts, marginally improving speed.

Removes redundant tcp header length checks before checksumming.

Instead, assumes (and documents) that any backlog processing and transform policies will carefully preserve the header, and will ensure the socket buffer length remains >= the header size.

Stand-alone patch, originally developed for TCPCT.

Signed-off-by [email protected]
Reviewed-by Andi Kleen <[email protected]>
---
 include/net/xfrm.h  |    7 ++++++
 net/ipv4/tcp_ipv4.c |   45 +++++++++++++++++++++-----------------
 net/ipv6/tcp_ipv6.c |   59 ++++++++++++++++++++++++++++----------------------
 3 files changed, 65 insertions(+), 46 deletions(-)
include/net/xfrm.h +7 -0
index 60c2770..81492a1 100644
@@ -975,6 +975,13 @@ xfrm_state_addr_cmp(struct xfrm_tmpl *tmpl, struct xfrm_state *x, unsigned short
 }
 
 #ifdef CONFIG_XFRM
+/*
+ * For transport, the policy is checked before the presumed more expensive
+ * checksum. The transport header has already been checked for size, and is
+ * guaranteed to be contiguous. These policies must not alter the header or
+ * its position in the buffer, and should not shorten the buffer length
+ * without ensuring the length remains >= the header size.
+ */
 extern int __xfrm_policy_check(struct sock *, int dir, struct sk_buff *skb, unsigned short family);
 
 static inline int __xfrm_policy_check2(struct sock *sk, int dir,
net/ipv4/tcp_ipv4.c +25 -20

109 lines not on this page · raw message

net/ipv6/tcp_ipv6.c +33 -26

126 lines not on this page · raw message

Signature · 1 line
1.6.3.3
author
#9 lore raw

PATCH v5 4/7 tcp: input header length, prediction, and timestamp bugs

In reply to William Allen Simpson

Fix incorrect header prediction flags documentation.

Relieve register pressure in (the i386) fast path by accessing skb->len directly, instead of carrying a rarely used len parameter.

Eliminate unused len parameters in two other functions.

Don't use output calculated tp->tcp_header_len for input decisions. While the output header is usually the same as the input (same options in both directions), that's a poor assumption. In particular, Sack will be different. Newer options are not guaranteed.

Moreover, in the fast path, that only saved a shift or two. The other efficiencies in this patch more than make up the difference.

Instead, use tp->rx_opt.tstamp_ok to accurately predict header length.

Likewise, use tp->rx_opt.tstamp_ok for received MSS calculations.

Don't use "sizeof(struct tcphdr) + TCPOLEN_TSTAMP_ALIGNED" to guess that the timestamp is present. This may have been OK in the days with fewer possible options, but various combinations of newer options may yield the same header length. (This bug is in 3 places.)

Instead, use tp->rx_opt.saw_tstamp to determine a timestamp is present.

There's no need to test buffer length against header length, already checked by tcp_v[4,6]_rcv(). Straighten code for minor efficiency gain.

Stand-alone patch, originally developed for TCPCT.

Requires:
net: tcp_header_len_th and tcp_option_len_th
tcp: harmonize tcp_vx_rcv header length assumptions

Signed-off-by [email protected]
Reviewed-by Andi Kleen <[email protected]>
---
 include/linux/tcp.h      |    6 ++-
 include/net/tcp.h        |   18 ++++++--
 net/ipv4/tcp_input.c     |   96 +++++++++++++++++++---------------------------
 net/ipv4/tcp_ipv4.c      |    4 +-
 net/ipv4/tcp_minisocks.c |    3 +-
 net/ipv4/tcp_probe.c     |    2 +-
 net/ipv6/tcp_ipv6.c      |    4 +-
 7 files changed, 63 insertions(+), 70 deletions(-)
include/linux/tcp.h +5 -1
index 74728f7..2987ee8 100644
@@ -301,7 +301,11 @@ struct tcp_sock {
 
 /*
  *	Header prediction flags
- *	0x5?10 << 16 + snd_wnd in net byte order
+ *	S << 28 + TCP_FLAG_ACK + snd_wnd, in net byte order
+ *		(PSH flag is ignored)
+ *		S is 5 (no options), or 8 (timestamp aligned)
+ *	otherwise, 0 to turn it off -- for instance, when there are
+ *	holes in receive space.
  */
 	__be32	pred_flags;
include/net/tcp.h +13 -5
index 34f5cc2..52dd185 100644
@@ -310,13 +310,11 @@ extern int			tcp_ioctl(struct sock *sk,
 
 extern int			tcp_rcv_state_process(struct sock *sk, 
 						      struct sk_buff *skb,
-						      struct tcphdr *th,
-						      unsigned len);
+						      struct tcphdr *th);
 
 extern int			tcp_rcv_established(struct sock *sk, 
 						    struct sk_buff *skb,
-						    struct tcphdr *th, 
-						    unsigned len);
+						    struct tcphdr *th);
 
 extern void			tcp_rcv_space_adjust(struct sock *sk);
 

@@ -533,9 +531,19 @@ static inline u32 __tcp_set_rto(const struct tcp_sock *tp)
 	return (tp->srtt >> 3) + tp->rttvar;
 }
 
+static inline u8 __tcp_fast_path_header_length(const struct tcp_sock *tp)
+{
+	return tp->rx_opt.tstamp_ok
+		? sizeof(struct tcphdr) + TCPOLEN_TSTAMP_ALIGNED
+		: sizeof(struct tcphdr);
+}
+
 static inline void __tcp_fast_path_on(struct tcp_sock *tp, u32 snd_wnd)
 {
-	tp->pred_flags = htonl((tp->tcp_header_len << 26) |
+	/* See linux/tcp.h for pred_flags details.
+	 * 'S' (doff) is 32-bit words, convert from bytes: 26 = 28 - 2.
+	 */
+	tp->pred_flags = htonl((__tcp_fast_path_header_length(tp) << 26) |
 			       ntohl(TCP_FLAG_ACK) |
 			       snd_wnd);
 }
net/ipv4/tcp_input.c +39 -57

169 lines not on this page · raw message

net/ipv4/tcp_ipv4.c +2 -2
index 0a76e41..f999e06 100644
@@ -1551,7 +1551,7 @@ int tcp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
 
 	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
 		TCP_CHECK_TIMER(sk);
-		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len)) {
+		if (tcp_rcv_established(sk, skb, tcp_hdr(skb))) {
 			rsk = sk;
 			goto reset;
 		}

@@ -1578,7 +1578,7 @@ int tcp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
 	}
 
 	TCP_CHECK_TIMER(sk);
-	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len)) {
+	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb))) {
 		rsk = sk;
 		goto reset;
 	}
net/ipv4/tcp_minisocks.c +1 -2
index f206ee5..37b7536 100644
@@ -718,8 +718,7 @@ int tcp_child_process(struct sock *parent, struct sock *child,
 	int state = child->sk_state;
 
 	if (!sock_owned_by_user(child)) {
-		ret = tcp_rcv_state_process(child, skb, tcp_hdr(skb),
-					    skb->len);
+		ret = tcp_rcv_state_process(child, skb, tcp_hdr(skb));
 		/* Wakeup parent, send SIGIO */
 		if (state == TCP_SYN_RECV && child->sk_state != state)
 			parent->sk_data_ready(parent, 0);
net/ipv4/tcp_probe.c +1 -1
index 9bc805d..de2a32e 100644
@@ -88,7 +88,7 @@ static inline int tcp_probe_avail(void)
  * Note: arguments must match tcp_rcv_established()!
  */
 static int jtcp_rcv_established(struct sock *sk, struct sk_buff *skb,
-			       struct tcphdr *th, unsigned len)
+			       struct tcphdr *th)
 {
 	const struct tcp_sock *tp = tcp_sk(sk);
 	const struct inet_sock *inet = inet_sk(sk);
net/ipv6/tcp_ipv6.c +2 -2
index b76939a..3d08a4d 100644
@@ -1586,7 +1586,7 @@ static int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
 
 	if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
 		TCP_CHECK_TIMER(sk);
-		if (tcp_rcv_established(sk, skb, tcp_hdr(skb), skb->len))
+		if (tcp_rcv_established(sk, skb, tcp_hdr(skb)))
 			goto reset;
 		TCP_CHECK_TIMER(sk);
 		if (opt_skb)

@@ -1618,7 +1618,7 @@ static int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
 	}
 
 	TCP_CHECK_TIMER(sk);
-	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb), skb->len))
+	if (tcp_rcv_state_process(sk, skb, tcp_hdr(skb)))
 		goto reset;
 	TCP_CHECK_TIMER(sk);
 	if (opt_skb)
Signature · 1 line
1.6.3.3
author
#10 lore raw

PATCH v3 5/7 TCPCT part 2e: accept SYNACK data

In reply to William Allen Simpson

When accompanied by cookie option, Initiator (client) queues incoming SYNACK transaction data.

This is a straightforward re-implementation of an earlier (year-old) patch that no longer applies cleanly, with permission of the original author (Adam Langley). The patch was previously reviewed:

http://thread.gmane.org/gmane.linux.network/102586

This function will also be used in subsequent patches that implement additional features.

Requires:
TCPCT part 1g: Responder Cookie => Initiator
net: tcp_header_len_th and tcp_option_len_th

Signed-off-by [email protected]
---
 net/ipv4/tcp_input.c |   26 +++++++++++++++++++++++++-
 1 files changed, 25 insertions(+), 1 deletions(-)
net/ipv4/tcp_input.c +25 -1
index e8f9e3b..08a1bb4 100644
@@ -5397,6 +5397,12 @@ discard:
 	return 0;
 }
 
+/*
+ * Returns:
+ *	+1 on reset,
+ *	0 success and/or SYNACK data,
+ *	-1 on discard.
+ */
 static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 					 struct tcphdr *th)
 {

@@ -5405,6 +5411,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 	struct tcp_sock *tp = tcp_sk(sk);
 	struct tcp_cookie_values *cvp = tp->cookie_values;
 	int saved_clamp = tp->rx_opt.mss_clamp;
+	int queued = 0;
 
 	tcp_parse_options(skb, &tp->rx_opt, &hash_location, 0);
 

@@ -5511,6 +5518,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 					- TCPOLEN_COOKIE_BASE;
 			int cookie_pair_size = cookie_size
 					     + cvp->cookie_desired;
+			int tcp_header_len = tcp_header_len_th(th);
 
 			/* A cookie extension option was sent and returned.
 			 * Note that each incoming SYNACK replaces the

@@ -5526,6 +5534,19 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 				       hash_location, cookie_size);
 				cvp->cookie_pair_size = cookie_pair_size;
 			}
+
+			queued = skb->len - tcp_header_len;
+			if (queued > 0) {
+				/* Queue incoming transaction data. */
+				__skb_pull(skb, tcp_header_len);
+				__skb_queue_tail(&sk->sk_receive_queue, skb);
+				skb_set_owner_r(skb, sk);
+				sk->sk_data_ready(sk, 0);
+				cvp->s_data_in = 1; /* true */
+				tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
+				tp->rcv_wup = TCP_SKB_CB(skb)->end_seq;
+				tp->copied_seq = TCP_SKB_CB(skb)->seq + 1;
+			}
 		}
 
 		smp_mb();

@@ -5579,11 +5600,14 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 						  TCP_DELACK_MAX, TCP_RTO_MAX);
 
 discard:
-			__kfree_skb(skb);
+			if (queued <= 0)
+				__kfree_skb(skb);
 			return 0;
 		} else {
 			tcp_send_ack(sk);
 		}
+		if (queued > 0)
+			return 0;
 		return -1;
 	}
Signature · 1 line
1.6.3.3
author
#11 lore raw

PATCH v5 6/7 TCPCT part 2f: cleanup tcp_parse_options

In reply to William Allen Simpson

Split switch, shift cases to the left, fix most lines beyond column 80.

Prepare (future) error return.

Harmonize initialization in syncookies.
Fix initialization in tcp_minisocks.

Repair net/ipv4/tcp_ipv4.c errors with goto targets, overlooked by David Miller in commit bb5b7c11263dbbe78253cd05945a6bf8f55add8e

Requires:
TCPCT part 1g: Responder Cookie => Initiator
net: tcp_header_len_th and tcp_option_len_th

Signed-off-by [email protected]
---
 include/net/tcp.h        |    3 +-
 net/ipv4/syncookies.c    |    9 ++-
 net/ipv4/tcp_input.c     |  223 ++++++++++++++++++++++++++--------------------
 net/ipv4/tcp_ipv4.c      |   10 ++-
 net/ipv4/tcp_minisocks.c |   26 ++++--
 net/ipv6/syncookies.c    |    9 ++-
 net/ipv6/tcp_ipv6.c      |    6 +-
 7 files changed, 172 insertions(+), 114 deletions(-)
include/net/tcp.h +2 -1
index 52dd185..4aed0c8 100644
@@ -403,7 +403,8 @@ extern int			tcp_recvmsg(struct kiocb *iocb, struct sock *sk,
 					    size_t len, int nonblock, 
 					    int flags, int *addr_len);
 
-extern void			tcp_parse_options(struct sk_buff *skb,
+extern int			tcp_parse_options(struct sk_buff *skb,
+						  const struct tcphdr *th,
 						  struct tcp_options_received *opt_rx,
 						  u8 **hvpp,
 						  int estab);
net/ipv4/syncookies.c +6 -3
index 66fd80e..d1f45ad 100644
@@ -255,15 +255,16 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
 {
 	struct tcp_options_received tcp_opt;
 	u8 *hash_location;
+	struct rtable *rt;
+	struct request_sock *req;
 	struct inet_request_sock *ireq;
 	struct tcp_request_sock *treq;
 	struct tcp_sock *tp = tcp_sk(sk);
 	const struct tcphdr *th = tcp_hdr(skb);
 	__u32 cookie = ntohl(th->ack_seq) - 1;
 	struct sock *ret = sk;
-	struct request_sock *req;
 	int mss;
-	struct rtable *rt;
+	int parsed;
 	__u8 rcv_wscale;
 
 	if (!sysctl_tcp_syncookies || !th->ack)

@@ -279,7 +280,9 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
 
 	/* check for timestamp cookie support */
 	memset(&tcp_opt, 0, sizeof(tcp_opt));
-	tcp_parse_options(skb, &tcp_opt, &hash_location, 0);
+	parsed = tcp_parse_options(skb, th, &tcp_opt, &hash_location, 0);
+	if (parsed < 0)
+		goto out;
 
 	if (tcp_opt.saw_tstamp)
 		cookie_check_timestamp(&tcp_opt);
net/ipv4/tcp_input.c +127 -96

275 lines not on this page · raw message

net/ipv4/tcp_ipv4.c +7 -3
index f999e06..3f0813f 100644
@@ -1215,6 +1215,7 @@ int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 {
 	struct tcp_extend_values tmp_ext;
 	struct tcp_options_received tmp_opt;
+	int parsed;
 	u8 *hash_location;
 	struct request_sock *req;
 	struct inet_request_sock *ireq;

@@ -1265,7 +1266,10 @@ int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 	tcp_clear_options(&tmp_opt);
 	tmp_opt.mss_clamp = TCP_MSS_DEFAULT;
 	tmp_opt.user_mss  = tp->rx_opt.user_mss;
-	tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
+	parsed = tcp_parse_options(skb, tcp_hdr(skb), &tmp_opt, &hash_location,
+				   0);
+	if (parsed < 0)
+		goto drop_and_free;
 
 	if (tmp_opt.cookie_plus > 0 &&
 	    tmp_opt.saw_tstamp &&

@@ -1278,7 +1282,7 @@ int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 		int l = tmp_opt.cookie_plus - TCPOLEN_COOKIE_BASE;
 
 		if (tcp_cookie_generator(&tmp_ext.cookie_bakery[0]) != 0)
-			goto drop_and_release;
+			goto drop_and_free;
 
 		/* Secret recipe starts with IP addresses */
 		*mess++ ^= daddr;

@@ -1299,7 +1303,7 @@ int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 		tmp_ext.cookie_out_never = 1; /* true */
 		tmp_ext.cookie_plus = 0;
 	} else {
-		goto drop_and_release;
+		goto drop_and_free;
 	}
 	tmp_ext.cookie_in_always = tp->rx_opt.cookie_in_always;
net/ipv4/tcp_minisocks.c +19 -7
index 37b7536..0f1b409 100644
@@ -95,15 +95,21 @@ tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb,
 	struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
 	int paws_reject = 0;
 
-	tmp_opt.saw_tstamp = 0;
+	/* Fast check for options, compare doff directly to constant value. */
 	if (th->doff > (sizeof(*th) >> 2) && tcptw->tw_ts_recent_stamp) {
-		tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
+		int parsed = tcp_parse_options(skb, th, &tmp_opt,
+					       &hash_location, 0);
 
-		if (tmp_opt.saw_tstamp) {
+		if (parsed < 0) {
+			paws_reject = 1; /* true */
+		} else if (tmp_opt.saw_tstamp) {
 			tmp_opt.ts_recent	= tcptw->tw_ts_recent;
 			tmp_opt.ts_recent_stamp	= tcptw->tw_ts_recent_stamp;
 			paws_reject = tcp_paws_reject(&tmp_opt, th->rst);
 		}
+	} else {
+		/* otherwise initialized by tcp_parse_options() */
+		tmp_opt.saw_tstamp = 0; /* false */
 	}
 
 	if (tw->tw_substate == TCP_FIN_WAIT2) {

@@ -526,11 +532,14 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
 	__be32 flg = tcp_flag_word(th) & (TCP_FLAG_RST|TCP_FLAG_SYN|TCP_FLAG_ACK);
 	int paws_reject = 0;
 
-	tmp_opt.saw_tstamp = 0;
-	if (th->doff > (sizeof(struct tcphdr)>>2)) {
-		tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
+	/* Fast check for options, compare doff directly to constant value. */
+	if (th->doff > (sizeof(*th) >> 2)) {
+		int parsed = tcp_parse_options(skb, th, &tmp_opt,
+					       &hash_location, 0);
 
-		if (tmp_opt.saw_tstamp) {
+		if (parsed < 0) {
+			paws_reject = 1; /* true */
+		} else if (tmp_opt.saw_tstamp) {
 			tmp_opt.ts_recent = req->ts_recent;
 			/* We do not store true stamp, but it is not required,
 			 * it can be estimated (approximately)

@@ -539,6 +548,9 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
 			tmp_opt.ts_recent_stamp = get_seconds() - ((TCP_TIMEOUT_INIT/HZ)<<req->retrans);
 			paws_reject = tcp_paws_reject(&tmp_opt, th->rst);
 		}
+	} else {
+		/* otherwise initialized by tcp_parse_options() */
+		tmp_opt.saw_tstamp = 0; /* false */
 	}
 
 	/* Check for pure retransmitted SYN. */
net/ipv6/syncookies.c +6 -3
index 7208a06..a0905b4 100644
@@ -161,6 +161,8 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
 {
 	struct tcp_options_received tcp_opt;
 	u8 *hash_location;
+	struct dst_entry *dst;
+	struct request_sock *req;
 	struct inet_request_sock *ireq;
 	struct inet6_request_sock *ireq6;
 	struct tcp_request_sock *treq;

@@ -169,9 +171,8 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
 	const struct tcphdr *th = tcp_hdr(skb);
 	__u32 cookie = ntohl(th->ack_seq) - 1;
 	struct sock *ret = sk;
-	struct request_sock *req;
 	int mss;
-	struct dst_entry *dst;
+	int parsed;
 	__u8 rcv_wscale;
 
 	if (!sysctl_tcp_syncookies || !th->ack)

@@ -187,7 +188,9 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
 
 	/* check for timestamp cookie support */
 	memset(&tcp_opt, 0, sizeof(tcp_opt));
-	tcp_parse_options(skb, &tcp_opt, &hash_location, 0);
+	parsed = tcp_parse_options(skb, th, &tcp_opt, &hash_location, 0);
+	if (parsed < 0)
+		goto out;
 
 	if (tcp_opt.saw_tstamp)
 		cookie_check_timestamp(&tcp_opt);
net/ipv6/tcp_ipv6.c +5 -1
index 3d08a4d..e15e4f6 100644
@@ -1164,6 +1164,7 @@ static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
 {
 	struct tcp_extend_values tmp_ext;
 	struct tcp_options_received tmp_opt;
+	int parsed;
 	u8 *hash_location;
 	struct request_sock *req;
 	struct inet6_request_sock *treq;

@@ -1207,7 +1208,10 @@ static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
 	tcp_clear_options(&tmp_opt);
 	tmp_opt.mss_clamp = IPV6_MIN_MTU - sizeof(struct tcphdr) - sizeof(struct ipv6hdr);
 	tmp_opt.user_mss = tp->rx_opt.user_mss;
-	tcp_parse_options(skb, &tmp_opt, &hash_location, 0);
+	parsed = tcp_parse_options(skb, tcp_hdr(skb), &tmp_opt, &hash_location,
+				   0);
+	if (parsed < 0)
+		goto drop_and_free;
 
 	if (tmp_opt.cookie_plus > 0 &&
 	    tmp_opt.saw_tstamp &&
Signature · 1 line
1.6.3.3
maintainer
#12 lore raw
In reply to William Allen Simpson

From: William Allen Simpson <[email protected]>
Date: Thu, 25 Feb 2010 15:30:03 -0500

Quoted 3 lines

I'd have thought that there would be greater interest about patching crashing bugs, signed versus unsigned (underflow) bugs, TCP DoS bugs, TCP data corruption, and TCP performance problems....

Your patches add as many bugs and problems as they claim to solve.

You also attack, in your commit messages and code coments, the very people you want to look at your changes and integrate them.

How you hope to make forward progress in these circumstances is beyond me.

Just remember William: Whilst people have a right to say whatever they want, they must earn the privilege to being listened to.

And currently many people have you set strictly to ignore.