Commit 33ad798c authored by Adam Langley's avatar Adam Langley Committed by David S. Miller

tcp: options clean up

This should fix the following bugs:
  * Connections with MD5 signatures produce invalid packets whenever SACK
    options are included
  * MD5 signatures are counted twice in the MSS calculations

Behaviour changes:
  * A SYN with MD5 + SACK + TS elicits a SYNACK with MD5 + SACK

    This is because we can't fit any SACK blocks in a packet with MD5 + TS
    options. There was discussion about disabling SACK rather than TS in
    order to fit in better with old, buggy kernels, but that was deemed to
    be unnecessary.

  * SYNs with MD5 don't include a TS option

    See above.

Additionally, it removes a bunch of duplicated logic for calculating options,
which should help avoid these sort of issues in the future.
Signed-off-by: default avatarAdam Langley <agl@imperialviolet.org>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent 49a72dfb
...@@ -50,6 +50,7 @@ extern atomic_t tcp_orphan_count; ...@@ -50,6 +50,7 @@ extern atomic_t tcp_orphan_count;
extern void tcp_time_wait(struct sock *sk, int state, int timeo); extern void tcp_time_wait(struct sock *sk, int state, int timeo);
#define MAX_TCP_HEADER (128 + MAX_HEADER) #define MAX_TCP_HEADER (128 + MAX_HEADER)
#define MAX_TCP_OPTION_SPACE 40
/* /*
* Never offer a window over 32767 without using window scaling. Some * Never offer a window over 32767 without using window scaling. Some
...@@ -184,6 +185,7 @@ extern void tcp_time_wait(struct sock *sk, int state, int timeo); ...@@ -184,6 +185,7 @@ extern void tcp_time_wait(struct sock *sk, int state, int timeo);
#define TCPOLEN_SACK_BASE_ALIGNED 4 #define TCPOLEN_SACK_BASE_ALIGNED 4
#define TCPOLEN_SACK_PERBLOCK 8 #define TCPOLEN_SACK_PERBLOCK 8
#define TCPOLEN_MD5SIG_ALIGNED 20 #define TCPOLEN_MD5SIG_ALIGNED 20
#define TCPOLEN_MSS_ALIGNED 4
/* Flags in tp->nonagle */ /* Flags in tp->nonagle */
#define TCP_NAGLE_OFF 1 /* Nagle's algo is disabled */ #define TCP_NAGLE_OFF 1 /* Nagle's algo is disabled */
......
...@@ -345,28 +345,82 @@ static void tcp_init_nondata_skb(struct sk_buff *skb, u32 seq, u8 flags) ...@@ -345,28 +345,82 @@ static void tcp_init_nondata_skb(struct sk_buff *skb, u32 seq, u8 flags)
TCP_SKB_CB(skb)->end_seq = seq; TCP_SKB_CB(skb)->end_seq = seq;
} }
static void tcp_build_and_update_options(__be32 *ptr, struct tcp_sock *tp, #define OPTION_SACK_ADVERTISE (1 << 0)
__u32 tstamp, __u8 **md5_hash) #define OPTION_TS (1 << 1)
{ #define OPTION_MD5 (1 << 2)
if (tp->rx_opt.tstamp_ok) {
struct tcp_out_options {
u8 options; /* bit field of OPTION_* */
u8 ws; /* window scale, 0 to disable */
u8 num_sack_blocks; /* number of SACK blocks to include */
u16 mss; /* 0 to disable */
__u32 tsval, tsecr; /* need to include OPTION_TS */
};
static void tcp_options_write(__be32 *ptr, struct tcp_sock *tp,
const struct tcp_out_options *opts,
__u8 **md5_hash) {
if (unlikely(OPTION_MD5 & opts->options)) {
*ptr++ = htonl((TCPOPT_NOP << 24) | *ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_NOP << 16) | (TCPOPT_NOP << 16) |
(TCPOPT_TIMESTAMP << 8) | (TCPOPT_MD5SIG << 8) |
TCPOLEN_TIMESTAMP); TCPOLEN_MD5SIG);
*ptr++ = htonl(tstamp); *md5_hash = (__u8 *)ptr;
*ptr++ = htonl(tp->rx_opt.ts_recent); ptr += 4;
} else {
*md5_hash = NULL;
} }
if (tp->rx_opt.eff_sacks) {
struct tcp_sack_block *sp = tp->rx_opt.dsack ? tp->duplicate_sack : tp->selective_acks; if (likely(OPTION_TS & opts->options)) {
if (unlikely(OPTION_SACK_ADVERTISE & opts->options)) {
*ptr++ = htonl((TCPOPT_SACK_PERM << 24) |
(TCPOLEN_SACK_PERM << 16) |
(TCPOPT_TIMESTAMP << 8) |
TCPOLEN_TIMESTAMP);
} else {
*ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_NOP << 16) |
(TCPOPT_TIMESTAMP << 8) |
TCPOLEN_TIMESTAMP);
}
*ptr++ = htonl(opts->tsval);
*ptr++ = htonl(opts->tsecr);
}
if (unlikely(opts->mss)) {
*ptr++ = htonl((TCPOPT_MSS << 24) |
(TCPOLEN_MSS << 16) |
opts->mss);
}
if (unlikely(OPTION_SACK_ADVERTISE & opts->options &&
!(OPTION_TS & opts->options))) {
*ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_NOP << 16) |
(TCPOPT_SACK_PERM << 8) |
TCPOLEN_SACK_PERM);
}
if (unlikely(opts->ws)) {
*ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_WINDOW << 16) |
(TCPOLEN_WINDOW << 8) |
opts->ws);
}
if (unlikely(opts->num_sack_blocks)) {
struct tcp_sack_block *sp = tp->rx_opt.dsack ?
tp->duplicate_sack : tp->selective_acks;
int this_sack; int this_sack;
*ptr++ = htonl((TCPOPT_NOP << 24) | *ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_NOP << 16) | (TCPOPT_NOP << 16) |
(TCPOPT_SACK << 8) | (TCPOPT_SACK << 8) |
(TCPOLEN_SACK_BASE + (tp->rx_opt.eff_sacks * (TCPOLEN_SACK_BASE + (opts->num_sack_blocks *
TCPOLEN_SACK_PERBLOCK))); TCPOLEN_SACK_PERBLOCK)));
for (this_sack = 0; this_sack < tp->rx_opt.eff_sacks; this_sack++) { for (this_sack = 0; this_sack < opts->num_sack_blocks;
++this_sack) {
*ptr++ = htonl(sp[this_sack].start_seq); *ptr++ = htonl(sp[this_sack].start_seq);
*ptr++ = htonl(sp[this_sack].end_seq); *ptr++ = htonl(sp[this_sack].end_seq);
} }
...@@ -376,81 +430,137 @@ static void tcp_build_and_update_options(__be32 *ptr, struct tcp_sock *tp, ...@@ -376,81 +430,137 @@ static void tcp_build_and_update_options(__be32 *ptr, struct tcp_sock *tp,
tp->rx_opt.eff_sacks--; tp->rx_opt.eff_sacks--;
} }
} }
}
static unsigned tcp_syn_options(struct sock *sk, struct sk_buff *skb,
struct tcp_out_options *opts,
struct tcp_md5sig_key **md5) {
struct tcp_sock *tp = tcp_sk(sk);
unsigned size = 0;
#ifdef CONFIG_TCP_MD5SIG #ifdef CONFIG_TCP_MD5SIG
if (md5_hash) { *md5 = tp->af_specific->md5_lookup(sk, sk);
*ptr++ = htonl((TCPOPT_NOP << 24) | if (*md5) {
(TCPOPT_NOP << 16) | opts->options |= OPTION_MD5;
(TCPOPT_MD5SIG << 8) | size += TCPOLEN_MD5SIG_ALIGNED;
TCPOLEN_MD5SIG);
*md5_hash = (__u8 *)ptr;
} }
#else
*md5 = NULL;
#endif #endif
/* We always get an MSS option. The option bytes which will be seen in
* normal data packets should timestamps be used, must be in the MSS
* advertised. But we subtract them from tp->mss_cache so that
* calculations in tcp_sendmsg are simpler etc. So account for this
* fact here if necessary. If we don't do this correctly, as a
* receiver we won't recognize data packets as being full sized when we
* should, and thus we won't abide by the delayed ACK rules correctly.
* SACKs don't matter, we never delay an ACK when we have any of those
* going out. */
opts->mss = tcp_advertise_mss(sk);
size += TCPOLEN_MSS_ALIGNED;
if (likely(sysctl_tcp_timestamps && *md5 == NULL)) {
opts->options |= OPTION_TS;
opts->tsval = TCP_SKB_CB(skb)->when;
opts->tsecr = tp->rx_opt.ts_recent;
size += TCPOLEN_TSTAMP_ALIGNED;
}
if (likely(sysctl_tcp_window_scaling)) {
opts->ws = tp->rx_opt.rcv_wscale;
size += TCPOLEN_WSCALE_ALIGNED;
}
if (likely(sysctl_tcp_sack)) {
opts->options |= OPTION_SACK_ADVERTISE;
if (unlikely(!OPTION_TS & opts->options))
size += TCPOLEN_SACKPERM_ALIGNED;
}
return size;
} }
/* Construct a tcp options header for a SYN or SYN_ACK packet. static unsigned tcp_synack_options(struct sock *sk,
* If this is every changed make sure to change the definition of struct request_sock *req,
* MAX_SYN_SIZE to match the new maximum number of options that you unsigned mss, struct sk_buff *skb,
* can generate. struct tcp_out_options *opts,
* struct tcp_md5sig_key **md5) {
* Note - that with the RFC2385 TCP option, we make room for the unsigned size = 0;
* 16 byte MD5 hash. This will be filled in later, so the pointer for the struct inet_request_sock *ireq = inet_rsk(req);
* location to be filled is passed back up. char doing_ts;
*/
static void tcp_syn_build_options(__be32 *ptr, int mss, int ts, int sack,
int offer_wscale, int wscale, __u32 tstamp,
__u32 ts_recent, __u8 **md5_hash)
{
/* We always get an MSS option.
* The option bytes which will be seen in normal data
* packets should timestamps be used, must be in the MSS
* advertised. But we subtract them from tp->mss_cache so
* that calculations in tcp_sendmsg are simpler etc.
* So account for this fact here if necessary. If we
* don't do this correctly, as a receiver we won't
* recognize data packets as being full sized when we
* should, and thus we won't abide by the delayed ACK
* rules correctly.
* SACKs don't matter, we never delay an ACK when we
* have any of those going out.
*/
*ptr++ = htonl((TCPOPT_MSS << 24) | (TCPOLEN_MSS << 16) | mss);
if (ts) {
if (sack)
*ptr++ = htonl((TCPOPT_SACK_PERM << 24) |
(TCPOLEN_SACK_PERM << 16) |
(TCPOPT_TIMESTAMP << 8) |
TCPOLEN_TIMESTAMP);
else
*ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_NOP << 16) |
(TCPOPT_TIMESTAMP << 8) |
TCPOLEN_TIMESTAMP);
*ptr++ = htonl(tstamp); /* TSVAL */
*ptr++ = htonl(ts_recent); /* TSECR */
} else if (sack)
*ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_NOP << 16) |
(TCPOPT_SACK_PERM << 8) |
TCPOLEN_SACK_PERM);
if (offer_wscale)
*ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_WINDOW << 16) |
(TCPOLEN_WINDOW << 8) |
(wscale));
#ifdef CONFIG_TCP_MD5SIG #ifdef CONFIG_TCP_MD5SIG
/* *md5 = tcp_rsk(req)->af_specific->md5_lookup(sk, req);
* If MD5 is enabled, then we set the option, and include the size if (*md5) {
* (always 18). The actual MD5 hash is added just before the opts->options |= OPTION_MD5;
* packet is sent. size += TCPOLEN_MD5SIG_ALIGNED;
*/
if (md5_hash) {
*ptr++ = htonl((TCPOPT_NOP << 24) |
(TCPOPT_NOP << 16) |
(TCPOPT_MD5SIG << 8) |
TCPOLEN_MD5SIG);
*md5_hash = (__u8 *)ptr;
} }
#else
*md5 = NULL;
#endif #endif
/* we can't fit any SACK blocks in a packet with MD5 + TS
options. There was discussion about disabling SACK rather than TS in
order to fit in better with old, buggy kernels, but that was deemed
to be unnecessary. */
doing_ts = ireq->tstamp_ok && !(*md5 && ireq->sack_ok);
opts->mss = mss;
size += TCPOLEN_MSS_ALIGNED;
if (likely(ireq->wscale_ok)) {
opts->ws = ireq->rcv_wscale;
size += TCPOLEN_WSCALE_ALIGNED;
}
if (likely(doing_ts)) {
opts->options |= OPTION_TS;
opts->tsval = TCP_SKB_CB(skb)->when;
opts->tsecr = req->ts_recent;
size += TCPOLEN_TSTAMP_ALIGNED;
}
if (likely(ireq->sack_ok)) {
opts->options |= OPTION_SACK_ADVERTISE;
if (unlikely(!doing_ts))
size += TCPOLEN_SACKPERM_ALIGNED;
}
return size;
}
static unsigned tcp_established_options(struct sock *sk, struct sk_buff *skb,
struct tcp_out_options *opts,
struct tcp_md5sig_key **md5) {
struct tcp_skb_cb *tcb = skb ? TCP_SKB_CB(skb) : NULL;
struct tcp_sock *tp = tcp_sk(sk);
unsigned size = 0;
#ifdef CONFIG_TCP_MD5SIG
*md5 = tp->af_specific->md5_lookup(sk, sk);
if (unlikely(*md5)) {
opts->options |= OPTION_MD5;
size += TCPOLEN_MD5SIG_ALIGNED;
}
#else
*md5 = NULL;
#endif
if (likely(tp->rx_opt.tstamp_ok)) {
opts->options |= OPTION_TS;
opts->tsval = tcb ? tcb->when : 0;
opts->tsecr = tp->rx_opt.ts_recent;
size += TCPOLEN_TSTAMP_ALIGNED;
}
if (unlikely(tp->rx_opt.eff_sacks)) {
const unsigned remaining = MAX_TCP_OPTION_SPACE - size;
opts->num_sack_blocks =
min_t(unsigned, tp->rx_opt.eff_sacks,
(remaining - TCPOLEN_SACK_BASE_ALIGNED) /
TCPOLEN_SACK_PERBLOCK);
size += TCPOLEN_SACK_BASE_ALIGNED +
opts->num_sack_blocks * TCPOLEN_SACK_PERBLOCK;
}
return size;
} }
/* This routine actually transmits TCP packets queued in by /* This routine actually transmits TCP packets queued in by
...@@ -471,13 +581,11 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, ...@@ -471,13 +581,11 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
struct inet_sock *inet; struct inet_sock *inet;
struct tcp_sock *tp; struct tcp_sock *tp;
struct tcp_skb_cb *tcb; struct tcp_skb_cb *tcb;
int tcp_header_size; struct tcp_out_options opts;
#ifdef CONFIG_TCP_MD5SIG unsigned tcp_options_size, tcp_header_size;
struct tcp_md5sig_key *md5; struct tcp_md5sig_key *md5;
__u8 *md5_hash_location; __u8 *md5_hash_location;
#endif
struct tcphdr *th; struct tcphdr *th;
int sysctl_flags;
int err; int err;
BUG_ON(!skb || !tcp_skb_pcount(skb)); BUG_ON(!skb || !tcp_skb_pcount(skb));
...@@ -500,52 +608,18 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, ...@@ -500,52 +608,18 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
inet = inet_sk(sk); inet = inet_sk(sk);
tp = tcp_sk(sk); tp = tcp_sk(sk);
tcb = TCP_SKB_CB(skb); tcb = TCP_SKB_CB(skb);
tcp_header_size = tp->tcp_header_len; memset(&opts, 0, sizeof(opts));
#define SYSCTL_FLAG_TSTAMPS 0x1
#define SYSCTL_FLAG_WSCALE 0x2
#define SYSCTL_FLAG_SACK 0x4
sysctl_flags = 0; if (unlikely(tcb->flags & TCPCB_FLAG_SYN))
if (unlikely(tcb->flags & TCPCB_FLAG_SYN)) { tcp_options_size = tcp_syn_options(sk, skb, &opts, &md5);
tcp_header_size = sizeof(struct tcphdr) + TCPOLEN_MSS; else
if (sysctl_tcp_timestamps) { tcp_options_size = tcp_established_options(sk, skb, &opts,
tcp_header_size += TCPOLEN_TSTAMP_ALIGNED; &md5);
sysctl_flags |= SYSCTL_FLAG_TSTAMPS; tcp_header_size = tcp_options_size + sizeof(struct tcphdr);
}
if (sysctl_tcp_window_scaling) {
tcp_header_size += TCPOLEN_WSCALE_ALIGNED;
sysctl_flags |= SYSCTL_FLAG_WSCALE;
}
if (sysctl_tcp_sack) {
sysctl_flags |= SYSCTL_FLAG_SACK;
if (!(sysctl_flags & SYSCTL_FLAG_TSTAMPS))
tcp_header_size += TCPOLEN_SACKPERM_ALIGNED;
}
} else if (unlikely(tp->rx_opt.eff_sacks)) {
/* A SACK is 2 pad bytes, a 2 byte header, plus
* 2 32-bit sequence numbers for each SACK block.
*/
tcp_header_size += (TCPOLEN_SACK_BASE_ALIGNED +
(tp->rx_opt.eff_sacks *
TCPOLEN_SACK_PERBLOCK));
}
if (tcp_packets_in_flight(tp) == 0) if (tcp_packets_in_flight(tp) == 0)
tcp_ca_event(sk, CA_EVENT_TX_START); tcp_ca_event(sk, CA_EVENT_TX_START);
#ifdef CONFIG_TCP_MD5SIG
/*
* Are we doing MD5 on this segment? If so - make
* room for it.
*/
md5 = tp->af_specific->md5_lookup(sk, sk);
if (md5) {
tcp_header_size += TCPOLEN_MD5SIG_ALIGNED;
sk->sk_route_caps &= ~NETIF_F_GSO_MASK;
}
#endif
skb_push(skb, tcp_header_size); skb_push(skb, tcp_header_size);
skb_reset_transport_header(skb); skb_reset_transport_header(skb);
skb_set_owner_w(skb, sk); skb_set_owner_w(skb, sk);
...@@ -576,33 +650,14 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, ...@@ -576,33 +650,14 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
th->urg = 1; th->urg = 1;
} }
if (unlikely(tcb->flags & TCPCB_FLAG_SYN)) { tcp_options_write((__be32 *)(th + 1), tp, &opts, &md5_hash_location);
tcp_syn_build_options((__be32 *)(th + 1), if (likely((tcb->flags & TCPCB_FLAG_SYN) == 0))
tcp_advertise_mss(sk),
(sysctl_flags & SYSCTL_FLAG_TSTAMPS),
(sysctl_flags & SYSCTL_FLAG_SACK),
(sysctl_flags & SYSCTL_FLAG_WSCALE),
tp->rx_opt.rcv_wscale,
tcb->when,
tp->rx_opt.ts_recent,
#ifdef CONFIG_TCP_MD5SIG
md5 ? &md5_hash_location :
#endif
NULL);
} else {
tcp_build_and_update_options((__be32 *)(th + 1),
tp, tcb->when,
#ifdef CONFIG_TCP_MD5SIG
md5 ? &md5_hash_location :
#endif
NULL);
TCP_ECN_send(sk, skb, tcp_header_size); TCP_ECN_send(sk, skb, tcp_header_size);
}
#ifdef CONFIG_TCP_MD5SIG #ifdef CONFIG_TCP_MD5SIG
/* Calculate the MD5 hash, as we have all we need now */ /* Calculate the MD5 hash, as we have all we need now */
if (md5) { if (md5) {
sk->sk_route_caps &= ~NETIF_F_GSO_MASK;
tp->af_specific->calc_md5_hash(md5_hash_location, tp->af_specific->calc_md5_hash(md5_hash_location,
md5, sk, NULL, skb); md5, sk, NULL, skb);
} }
...@@ -626,10 +681,6 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, ...@@ -626,10 +681,6 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
tcp_enter_cwr(sk, 1); tcp_enter_cwr(sk, 1);
return net_xmit_eval(err); return net_xmit_eval(err);
#undef SYSCTL_FLAG_TSTAMPS
#undef SYSCTL_FLAG_WSCALE
#undef SYSCTL_FLAG_SACK
} }
/* This routine just queue's the buffer /* This routine just queue's the buffer
...@@ -970,6 +1021,9 @@ unsigned int tcp_current_mss(struct sock *sk, int large_allowed) ...@@ -970,6 +1021,9 @@ unsigned int tcp_current_mss(struct sock *sk, int large_allowed)
u32 mss_now; u32 mss_now;
u16 xmit_size_goal; u16 xmit_size_goal;
int doing_tso = 0; int doing_tso = 0;
unsigned header_len;
struct tcp_out_options opts;
struct tcp_md5sig_key *md5;
mss_now = tp->mss_cache; mss_now = tp->mss_cache;
...@@ -982,14 +1036,16 @@ unsigned int tcp_current_mss(struct sock *sk, int large_allowed) ...@@ -982,14 +1036,16 @@ unsigned int tcp_current_mss(struct sock *sk, int large_allowed)
mss_now = tcp_sync_mss(sk, mtu); mss_now = tcp_sync_mss(sk, mtu);
} }
if (tp->rx_opt.eff_sacks) header_len = tcp_established_options(sk, NULL, &opts, &md5) +
mss_now -= (TCPOLEN_SACK_BASE_ALIGNED + sizeof(struct tcphdr);
(tp->rx_opt.eff_sacks * TCPOLEN_SACK_PERBLOCK)); /* The mss_cache is sized based on tp->tcp_header_len, which assumes
* some common options. If this is an odd packet (because we have SACK
#ifdef CONFIG_TCP_MD5SIG * blocks etc) then our calculated header_len will be different, and
if (tp->af_specific->md5_lookup(sk, sk)) * we have to adjust mss_now correspondingly */
mss_now -= TCPOLEN_MD5SIG_ALIGNED; if (header_len != tp->tcp_header_len) {
#endif int delta = (int) header_len - tp->tcp_header_len;
mss_now -= delta;
}
xmit_size_goal = mss_now; xmit_size_goal = mss_now;
...@@ -2179,11 +2235,10 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst, ...@@ -2179,11 +2235,10 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst,
struct tcp_sock *tp = tcp_sk(sk); struct tcp_sock *tp = tcp_sk(sk);
struct tcphdr *th; struct tcphdr *th;
int tcp_header_size; int tcp_header_size;
struct tcp_out_options opts;
struct sk_buff *skb; struct sk_buff *skb;
#ifdef CONFIG_TCP_MD5SIG
struct tcp_md5sig_key *md5; struct tcp_md5sig_key *md5;
__u8 *md5_hash_location; __u8 *md5_hash_location;
#endif
skb = sock_wmalloc(sk, MAX_TCP_HEADER + 15, 1, GFP_ATOMIC); skb = sock_wmalloc(sk, MAX_TCP_HEADER + 15, 1, GFP_ATOMIC);
if (skb == NULL) if (skb == NULL)
...@@ -2194,18 +2249,27 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst, ...@@ -2194,18 +2249,27 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst,
skb->dst = dst_clone(dst); skb->dst = dst_clone(dst);
tcp_header_size = (sizeof(struct tcphdr) + TCPOLEN_MSS + if (req->rcv_wnd == 0) { /* ignored for retransmitted syns */
(ireq->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0) + __u8 rcv_wscale;
(ireq->wscale_ok ? TCPOLEN_WSCALE_ALIGNED : 0) + /* Set this up on the first call only */
/* SACK_PERM is in the place of NOP NOP of TS */ req->window_clamp = tp->window_clamp ? : dst_metric(dst, RTAX_WINDOW);
((ireq->sack_ok && !ireq->tstamp_ok) ? TCPOLEN_SACKPERM_ALIGNED : 0)); /* tcp_full_space because it is guaranteed to be the first packet */
tcp_select_initial_window(tcp_full_space(sk),
dst_metric(dst, RTAX_ADVMSS) - (ireq->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0),
&req->rcv_wnd,
&req->window_clamp,
ireq->wscale_ok,
&rcv_wscale);
ireq->rcv_wscale = rcv_wscale;
}
memset(&opts, 0, sizeof(opts));
TCP_SKB_CB(skb)->when = tcp_time_stamp;
tcp_header_size = tcp_synack_options(sk, req,
dst_metric(dst, RTAX_ADVMSS),
skb, &opts, &md5) +
sizeof(struct tcphdr);
#ifdef CONFIG_TCP_MD5SIG
/* Are we doing MD5 on this segment? If so - make room for it */
md5 = tcp_rsk(req)->af_specific->md5_lookup(sk, req);
if (md5)
tcp_header_size += TCPOLEN_MD5SIG_ALIGNED;
#endif
skb_push(skb, tcp_header_size); skb_push(skb, tcp_header_size);
skb_reset_transport_header(skb); skb_reset_transport_header(skb);
...@@ -2223,19 +2287,6 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst, ...@@ -2223,19 +2287,6 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst,
TCPCB_FLAG_SYN | TCPCB_FLAG_ACK); TCPCB_FLAG_SYN | TCPCB_FLAG_ACK);
th->seq = htonl(TCP_SKB_CB(skb)->seq); th->seq = htonl(TCP_SKB_CB(skb)->seq);
th->ack_seq = htonl(tcp_rsk(req)->rcv_isn + 1); th->ack_seq = htonl(tcp_rsk(req)->rcv_isn + 1);
if (req->rcv_wnd == 0) { /* ignored for retransmitted syns */
__u8 rcv_wscale;
/* Set this up on the first call only */
req->window_clamp = tp->window_clamp ? : dst_metric(dst, RTAX_WINDOW);
/* tcp_full_space because it is guaranteed to be the first packet */
tcp_select_initial_window(tcp_full_space(sk),
dst_metric(dst, RTAX_ADVMSS) - (ireq->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0),
&req->rcv_wnd,
&req->window_clamp,
ireq->wscale_ok,
&rcv_wscale);
ireq->rcv_wscale = rcv_wscale;
}
/* RFC1323: The window in SYN & SYN/ACK segments is never scaled. */ /* RFC1323: The window in SYN & SYN/ACK segments is never scaled. */
th->window = htons(min(req->rcv_wnd, 65535U)); th->window = htons(min(req->rcv_wnd, 65535U));
...@@ -2244,18 +2295,7 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst, ...@@ -2244,18 +2295,7 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst,
TCP_SKB_CB(skb)->when = cookie_init_timestamp(req); TCP_SKB_CB(skb)->when = cookie_init_timestamp(req);
else else
#endif #endif
TCP_SKB_CB(skb)->when = tcp_time_stamp; tcp_options_write((__be32 *)(th + 1), tp, &opts, &md5_hash_location);
tcp_syn_build_options((__be32 *)(th + 1), dst_metric(dst, RTAX_ADVMSS), ireq->tstamp_ok,
ireq->sack_ok, ireq->wscale_ok, ireq->rcv_wscale,
TCP_SKB_CB(skb)->when,
req->ts_recent,
(
#ifdef CONFIG_TCP_MD5SIG
md5 ? &md5_hash_location :
#endif
NULL)
);
th->doff = (tcp_header_size >> 2); th->doff = (tcp_header_size >> 2);
TCP_INC_STATS(sock_net(sk), TCP_MIB_OUTSEGS); TCP_INC_STATS(sock_net(sk), TCP_MIB_OUTSEGS);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment