#include <linux/cache.h>
#include <linux/percpu.h>
#include <net/checksum.h>
+#include <net/request_sock.h>
#include <net/sock.h>
#include <net/snmp.h>
#include <net/ip.h>
#else
# define TCP_TW_RECYCLE_TICK (12+2-TCP_TW_RECYCLE_SLOTS_LOG)
#endif
-
-#define BICTCP_BETA_SCALE 1024 /* Scale factor beta calculation
- * max_cwnd = snd_cwnd * beta
- */
-#define BICTCP_MAX_INCREMENT 32 /*
- * Limit on the amount of
- * increment allowed during
- * binary search.
- */
-#define BICTCP_FUNC_OF_MIN_INCR 11 /*
- * log(B/Smin)/log(B/(B-1))+1,
- * Smin:min increment
- * B:log factor
- */
-#define BICTCP_B 4 /*
- * In binary search,
- * go to point (max+min)/N
- */
-
/*
* TCP option
*/
#define TCP_NAGLE_PUSH 4 /* Cork is overriden for already queued data */
/* sysctl variables for tcp */
-extern int sysctl_max_syn_backlog;
extern int sysctl_tcp_timestamps;
extern int sysctl_tcp_window_scaling;
extern int sysctl_tcp_sack;
extern int sysctl_tcp_tw_reuse;
extern int sysctl_tcp_frto;
extern int sysctl_tcp_low_latency;
-extern int sysctl_tcp_westwood;
-extern int sysctl_tcp_vegas_cong_avoid;
-extern int sysctl_tcp_vegas_alpha;
-extern int sysctl_tcp_vegas_beta;
-extern int sysctl_tcp_vegas_gamma;
extern int sysctl_tcp_nometrics_save;
-extern int sysctl_tcp_bic;
-extern int sysctl_tcp_bic_fast_convergence;
-extern int sysctl_tcp_bic_low_window;
-extern int sysctl_tcp_bic_beta;
extern int sysctl_tcp_moderate_rcvbuf;
extern int sysctl_tcp_tso_win_divisor;
extern atomic_t tcp_sockets_allocated;
extern int tcp_memory_pressure;
-struct open_request;
-
-struct or_calltable {
- int family;
- int (*rtx_syn_ack) (struct sock *sk, struct open_request *req, struct dst_entry*);
- void (*send_ack) (struct sk_buff *skb, struct open_request *req);
- void (*destructor) (struct open_request *req);
- void (*send_reset) (struct sk_buff *skb);
-};
-
-struct tcp_v4_open_req {
- __u32 loc_addr;
- __u32 rmt_addr;
- struct ip_options *opt;
-};
-
-#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
-struct tcp_v6_open_req {
- struct in6_addr loc_addr;
- struct in6_addr rmt_addr;
- struct sk_buff *pktopts;
- int iif;
-};
-#endif
-
-/* this structure is too big */
-struct open_request {
- struct open_request *dl_next; /* Must be first member! */
- __u32 rcv_isn;
- __u32 snt_isn;
- __u16 rmt_port;
- __u16 mss;
- __u8 retrans;
- __u8 __pad;
- __u16 snd_wscale : 4,
- rcv_wscale : 4,
- tstamp_ok : 1,
- sack_ok : 1,
- wscale_ok : 1,
- ecn_ok : 1,
- acked : 1;
- /* The following two fields can be easily recomputed I think -AK */
- __u32 window_clamp; /* window clamp at creation time */
- __u32 rcv_wnd; /* rcv_wnd offered first time */
- __u32 ts_recent;
- unsigned long expires;
- struct or_calltable *class;
- struct sock *sk;
- union {
- struct tcp_v4_open_req v4_req;
-#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
- struct tcp_v6_open_req v6_req;
-#endif
- } af;
-};
-
-/* SLAB cache for open requests. */
-extern kmem_cache_t *tcp_openreq_cachep;
-
-#define tcp_openreq_alloc() kmem_cache_alloc(tcp_openreq_cachep, SLAB_ATOMIC)
-#define tcp_openreq_fastfree(req) kmem_cache_free(tcp_openreq_cachep, req)
-
-static inline void tcp_openreq_free(struct open_request *req)
-{
- req->class->destructor(req);
- tcp_openreq_fastfree(req);
-}
-
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
#define TCP_INET_FAMILY(fam) ((fam) == AF_INET)
#else
struct sock * (*syn_recv_sock) (struct sock *sk,
struct sk_buff *skb,
- struct open_request *req,
+ struct request_sock *req,
struct dst_entry *dst);
int (*remember_stamp) (struct sock *sk);
unsigned len);
extern struct sock * tcp_check_req(struct sock *sk,struct sk_buff *skb,
- struct open_request *req,
- struct open_request **prev);
+ struct request_sock *req,
+ struct request_sock **prev);
extern int tcp_child_process(struct sock *parent,
struct sock *child,
struct sk_buff *skb);
struct sk_buff *skb);
extern struct sock * tcp_create_openreq_child(struct sock *sk,
- struct open_request *req,
+ struct request_sock *req,
struct sk_buff *skb);
extern struct sock * tcp_v4_syn_recv_sock(struct sock *sk,
struct sk_buff *skb,
- struct open_request *req,
+ struct request_sock *req,
struct dst_entry *dst);
extern int tcp_v4_do_rcv(struct sock *sk,
extern struct sk_buff * tcp_make_synack(struct sock *sk,
struct dst_entry *dst,
- struct open_request *req);
+ struct request_sock *req);
extern int tcp_disconnect(struct sock *sk, int flags);
tp->packets_out -= tcp_skb_pcount(skb);
}
+/* Events passed to congestion control interface */
+enum tcp_ca_event {
+ CA_EVENT_TX_START, /* first transmit when no packets in flight */
+ CA_EVENT_CWND_RESTART, /* congestion window restart */
+ CA_EVENT_COMPLETE_CWR, /* end of congestion recovery */
+ CA_EVENT_FRTO, /* fast recovery timeout */
+ CA_EVENT_LOSS, /* loss timeout */
+ CA_EVENT_FAST_ACK, /* in sequence ack */
+ CA_EVENT_SLOW_ACK, /* other ack */
+};
+
+/*
+ * Interface for adding new TCP congestion control handlers
+ */
+#define TCP_CA_NAME_MAX 16
+struct tcp_congestion_ops {
+ struct list_head list;
+
+ /* initialize private data (optional) */
+ void (*init)(struct tcp_sock *tp);
+ /* cleanup private data (optional) */
+ void (*release)(struct tcp_sock *tp);
+
+ /* return slow start threshold (required) */
+ u32 (*ssthresh)(struct tcp_sock *tp);
+ /* lower bound for congestion window (optional) */
+ u32 (*min_cwnd)(struct tcp_sock *tp);
+ /* do new cwnd calculation (required) */
+ void (*cong_avoid)(struct tcp_sock *tp, u32 ack,
+ u32 rtt, u32 in_flight, int good_ack);
+ /* round trip time sample per acked packet (optional) */
+ void (*rtt_sample)(struct tcp_sock *tp, u32 usrtt);
+ /* call before changing ca_state (optional) */
+ void (*set_state)(struct tcp_sock *tp, u8 new_state);
+ /* call when cwnd event occurs (optional) */
+ void (*cwnd_event)(struct tcp_sock *tp, enum tcp_ca_event ev);
+ /* new value of cwnd after loss (optional) */
+ u32 (*undo_cwnd)(struct tcp_sock *tp);
+ /* hook for packet ack accounting (optional) */
+ void (*pkts_acked)(struct tcp_sock *tp, u32 num_acked);
+ /* get info for tcp_diag (optional) */
+ void (*get_info)(struct tcp_sock *tp, u32 ext, struct sk_buff *skb);
+
+ char name[TCP_CA_NAME_MAX];
+ struct module *owner;
+};
+
+extern int tcp_register_congestion_control(struct tcp_congestion_ops *type);
+extern void tcp_unregister_congestion_control(struct tcp_congestion_ops *type);
+
+extern void tcp_init_congestion_control(struct tcp_sock *tp);
+extern void tcp_cleanup_congestion_control(struct tcp_sock *tp);
+extern int tcp_set_default_congestion_control(const char *name);
+extern void tcp_get_default_congestion_control(char *name);
+
+extern struct tcp_congestion_ops tcp_reno;
+extern u32 tcp_reno_ssthresh(struct tcp_sock *tp);
+extern void tcp_reno_cong_avoid(struct tcp_sock *tp, u32 ack,
+ u32 rtt, u32 in_flight, int flag);
+extern u32 tcp_reno_min_cwnd(struct tcp_sock *tp);
+
+static inline void tcp_set_ca_state(struct tcp_sock *tp, u8 ca_state)
+{
+ if (tp->ca_ops->set_state)
+ tp->ca_ops->set_state(tp, ca_state);
+ tp->ca_state = ca_state;
+}
+
+static inline void tcp_ca_event(struct tcp_sock *tp, enum tcp_ca_event event)
+{
+ if (tp->ca_ops->cwnd_event)
+ tp->ca_ops->cwnd_event(tp, event);
+}
+
/* This determines how many packets are "in the network" to the best
* of our knowledge. In many cases it is conservative, but where
* detailed information is available from the receiver (via SACK
return (tp->packets_out - tp->left_out + tp->retrans_out);
}
-/*
- * Which congestion algorithim is in use on the connection.
- */
-#define tcp_is_vegas(__tp) ((__tp)->adv_cong == TCP_VEGAS)
-#define tcp_is_westwood(__tp) ((__tp)->adv_cong == TCP_WESTWOOD)
-#define tcp_is_bic(__tp) ((__tp)->adv_cong == TCP_BIC)
-
-/* Recalculate snd_ssthresh, we want to set it to:
- *
- * Reno:
- * one half the current congestion window, but no
- * less than two segments
- *
- * BIC:
- * behave like Reno until low_window is reached,
- * then increase congestion window slowly
- */
-static inline __u32 tcp_recalc_ssthresh(struct tcp_sock *tp)
-{
- if (tcp_is_bic(tp)) {
- if (sysctl_tcp_bic_fast_convergence &&
- tp->snd_cwnd < tp->bictcp.last_max_cwnd)
- tp->bictcp.last_max_cwnd = (tp->snd_cwnd *
- (BICTCP_BETA_SCALE
- + sysctl_tcp_bic_beta))
- / (2 * BICTCP_BETA_SCALE);
- else
- tp->bictcp.last_max_cwnd = tp->snd_cwnd;
-
- if (tp->snd_cwnd > sysctl_tcp_bic_low_window)
- return max((tp->snd_cwnd * sysctl_tcp_bic_beta)
- / BICTCP_BETA_SCALE, 2U);
- }
-
- return max(tp->snd_cwnd >> 1U, 2U);
-}
-
-/* Stop taking Vegas samples for now. */
-#define tcp_vegas_disable(__tp) ((__tp)->vegas.doing_vegas_now = 0)
-
-static inline void tcp_vegas_enable(struct tcp_sock *tp)
-{
- /* There are several situations when we must "re-start" Vegas:
- *
- * o when a connection is established
- * o after an RTO
- * o after fast recovery
- * o when we send a packet and there is no outstanding
- * unacknowledged data (restarting an idle connection)
- *
- * In these circumstances we cannot do a Vegas calculation at the
- * end of the first RTT, because any calculation we do is using
- * stale info -- both the saved cwnd and congestion feedback are
- * stale.
- *
- * Instead we must wait until the completion of an RTT during
- * which we actually receive ACKs.
- */
-
- /* Begin taking Vegas samples next time we send something. */
- tp->vegas.doing_vegas_now = 1;
-
- /* Set the beginning of the next send window. */
- tp->vegas.beg_snd_nxt = tp->snd_nxt;
-
- tp->vegas.cntRTT = 0;
- tp->vegas.minRTT = 0x7fffffff;
-}
-
-/* Should we be taking Vegas samples right now? */
-#define tcp_vegas_enabled(__tp) ((__tp)->vegas.doing_vegas_now)
-
-extern void tcp_ca_init(struct tcp_sock *tp);
-
-static inline void tcp_set_ca_state(struct tcp_sock *tp, u8 ca_state)
-{
- if (tcp_is_vegas(tp)) {
- if (ca_state == TCP_CA_Open)
- tcp_vegas_enable(tp);
- else
- tcp_vegas_disable(tp);
- }
- tp->ca_state = ca_state;
-}
-
/* If cwnd > ssthresh, we may raise ssthresh to be half-way to cwnd.
* The exception is rate halving phase, when cwnd is decreasing towards
* ssthresh.
static inline void __tcp_enter_cwr(struct tcp_sock *tp)
{
tp->undo_marker = 0;
- tp->snd_ssthresh = tcp_recalc_ssthresh(tp);
+ tp->snd_ssthresh = tp->ca_ops->ssthresh(tp);
tp->snd_cwnd = min(tp->snd_cwnd,
tcp_packets_in_flight(tp) + 1U);
tp->snd_cwnd_cnt = 0;
return tcp_win_from_space(sk->sk_rcvbuf);
}
-static inline void tcp_acceptq_queue(struct sock *sk, struct open_request *req,
+static inline void tcp_acceptq_queue(struct sock *sk, struct request_sock *req,
struct sock *child)
{
- struct tcp_sock *tp = tcp_sk(sk);
-
- req->sk = child;
- sk_acceptq_added(sk);
-
- if (!tp->accept_queue_tail) {
- tp->accept_queue = req;
- } else {
- tp->accept_queue_tail->dl_next = req;
- }
- tp->accept_queue_tail = req;
- req->dl_next = NULL;
+ reqsk_queue_add(&tcp_sk(sk)->accept_queue, req, sk, child);
}
-struct tcp_listen_opt
-{
- u8 max_qlen_log; /* log_2 of maximal queued SYNs */
- int qlen;
- int qlen_young;
- int clock_hand;
- u32 hash_rnd;
- struct open_request *syn_table[TCP_SYNQ_HSIZE];
-};
-
static inline void
-tcp_synq_removed(struct sock *sk, struct open_request *req)
+tcp_synq_removed(struct sock *sk, struct request_sock *req)
{
- struct tcp_listen_opt *lopt = tcp_sk(sk)->listen_opt;
-
- if (--lopt->qlen == 0)
+ if (reqsk_queue_removed(&tcp_sk(sk)->accept_queue, req) == 0)
tcp_delete_keepalive_timer(sk);
- if (req->retrans == 0)
- lopt->qlen_young--;
}
static inline void tcp_synq_added(struct sock *sk)
{
- struct tcp_listen_opt *lopt = tcp_sk(sk)->listen_opt;
-
- if (lopt->qlen++ == 0)
+ if (reqsk_queue_added(&tcp_sk(sk)->accept_queue) == 0)
tcp_reset_keepalive_timer(sk, TCP_TIMEOUT_INIT);
- lopt->qlen_young++;
}
static inline int tcp_synq_len(struct sock *sk)
{
- return tcp_sk(sk)->listen_opt->qlen;
+ return reqsk_queue_len(&tcp_sk(sk)->accept_queue);
}
static inline int tcp_synq_young(struct sock *sk)
{
- return tcp_sk(sk)->listen_opt->qlen_young;
+ return reqsk_queue_len_young(&tcp_sk(sk)->accept_queue);
}
static inline int tcp_synq_is_full(struct sock *sk)
{
- return tcp_synq_len(sk) >> tcp_sk(sk)->listen_opt->max_qlen_log;
+ return reqsk_queue_is_full(&tcp_sk(sk)->accept_queue);
}
-static inline void tcp_synq_unlink(struct tcp_sock *tp, struct open_request *req,
- struct open_request **prev)
+static inline void tcp_synq_unlink(struct tcp_sock *tp, struct request_sock *req,
+ struct request_sock **prev)
{
- write_lock(&tp->syn_wait_lock);
- *prev = req->dl_next;
- write_unlock(&tp->syn_wait_lock);
+ reqsk_queue_unlink(&tp->accept_queue, req, prev);
}
-static inline void tcp_synq_drop(struct sock *sk, struct open_request *req,
- struct open_request **prev)
+static inline void tcp_synq_drop(struct sock *sk, struct request_sock *req,
+ struct request_sock **prev)
{
tcp_synq_unlink(tcp_sk(sk), req, prev);
tcp_synq_removed(sk, req);
- tcp_openreq_free(req);
+ reqsk_free(req);
}
-static __inline__ void tcp_openreq_init(struct open_request *req,
+static __inline__ void tcp_openreq_init(struct request_sock *req,
struct tcp_options_received *rx_opt,
struct sk_buff *skb)
{
+ struct inet_request_sock *ireq = inet_rsk(req);
+
req->rcv_wnd = 0; /* So that tcp_send_synack() knows! */
- req->rcv_isn = TCP_SKB_CB(skb)->seq;
+ tcp_rsk(req)->rcv_isn = TCP_SKB_CB(skb)->seq;
req->mss = rx_opt->mss_clamp;
req->ts_recent = rx_opt->saw_tstamp ? rx_opt->rcv_tsval : 0;
- req->tstamp_ok = rx_opt->tstamp_ok;
- req->sack_ok = rx_opt->sack_ok;
- req->snd_wscale = rx_opt->snd_wscale;
- req->wscale_ok = rx_opt->wscale_ok;
- req->acked = 0;
- req->ecn_ok = 0;
- req->rmt_port = skb->h.th->source;
+ ireq->tstamp_ok = rx_opt->tstamp_ok;
+ ireq->sack_ok = rx_opt->sack_ok;
+ ireq->snd_wscale = rx_opt->snd_wscale;
+ ireq->wscale_ok = rx_opt->wscale_ok;
+ ireq->acked = 0;
+ ireq->ecn_ok = 0;
+ ireq->rmt_port = skb->h.th->source;
}
extern void tcp_enter_memory_pressure(void);
extern int tcp_proc_register(struct tcp_seq_afinfo *afinfo);
extern void tcp_proc_unregister(struct tcp_seq_afinfo *afinfo);
-/* TCP Westwood functions and constants */
-
-#define TCP_WESTWOOD_INIT_RTT (20*HZ) /* maybe too conservative?! */
-#define TCP_WESTWOOD_RTT_MIN (HZ/20) /* 50ms */
-
-static inline void tcp_westwood_update_rtt(struct tcp_sock *tp, __u32 rtt_seq)
-{
- if (tcp_is_westwood(tp))
- tp->westwood.rtt = rtt_seq;
-}
-
-static inline __u32 __tcp_westwood_bw_rttmin(const struct tcp_sock *tp)
-{
- return max((tp->westwood.bw_est) * (tp->westwood.rtt_min) /
- (__u32) (tp->mss_cache_std),
- 2U);
-}
-
-static inline __u32 tcp_westwood_bw_rttmin(const struct tcp_sock *tp)
-{
- return tcp_is_westwood(tp) ? __tcp_westwood_bw_rttmin(tp) : 0;
-}
-
-static inline int tcp_westwood_ssthresh(struct tcp_sock *tp)
-{
- __u32 ssthresh = 0;
-
- if (tcp_is_westwood(tp)) {
- ssthresh = __tcp_westwood_bw_rttmin(tp);
- if (ssthresh)
- tp->snd_ssthresh = ssthresh;
- }
-
- return (ssthresh != 0);
-}
-
-static inline int tcp_westwood_cwnd(struct tcp_sock *tp)
-{
- __u32 cwnd = 0;
-
- if (tcp_is_westwood(tp)) {
- cwnd = __tcp_westwood_bw_rttmin(tp);
- if (cwnd)
- tp->snd_cwnd = cwnd;
- }
-
- return (cwnd != 0);
-}
#endif /* _TCP_H */
projects / powerpc.git / blobdiff