X-Git-Url: http://ftp.safe.ca/?a=blobdiff_plain;f=net%2Fdccp%2Foutput.c;h=c0e88c16d0888b5222ca250c01a27bf3078a7e2a;hb=adf30907d63893e4208dfe3f5c88ae12bc2f25d5;hp=3435542e96528f5cce06257980e7ee95cc28ccf8;hpb=e89862f4c5b3c4ac9afcd8cb1365d2f1e16ddc3b;p=safe%2Fjmp%2Flinux-2.6 diff --git a/net/dccp/output.c b/net/dccp/output.c index 3435542..c0e88c1 100644 --- a/net/dccp/output.c +++ b/net/dccp/output.c @@ -53,41 +53,48 @@ static int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb) dccp_packet_hdr_len(dcb->dccpd_type); int err, set_ack = 1; u64 ackno = dp->dccps_gsr; - - dccp_inc_seqno(&dp->dccps_gss); + /* + * Increment GSS here already in case the option code needs it. + * Update GSS for real only if option processing below succeeds. + */ + dcb->dccpd_seq = ADD48(dp->dccps_gss, 1); switch (dcb->dccpd_type) { case DCCP_PKT_DATA: set_ack = 0; /* fall through */ case DCCP_PKT_DATAACK: + case DCCP_PKT_RESET: break; case DCCP_PKT_REQUEST: set_ack = 0; + /* Use ISS on the first (non-retransmitted) Request. */ + if (icsk->icsk_retransmits == 0) + dcb->dccpd_seq = dp->dccps_iss; /* fall through */ case DCCP_PKT_SYNC: case DCCP_PKT_SYNCACK: - ackno = dcb->dccpd_seq; + ackno = dcb->dccpd_ack_seq; /* fall through */ default: /* - * Only data packets should come through with skb->sk - * set. + * Set owner/destructor: some skbs are allocated via + * alloc_skb (e.g. when retransmission may happen). + * Only Data, DataAck, and Reset packets should come + * through here with skb->sk set. */ WARN_ON(skb->sk); skb_set_owner_w(skb, sk); break; } - dcb->dccpd_seq = dp->dccps_gss; - if (dccp_insert_options(sk, skb)) { kfree_skb(skb); return -EPROTO; } - + /* Build DCCP header and checksum it. */ dh = dccp_zeroed_hdr(skb, dccp_header_size); @@ -100,7 +107,7 @@ static int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb) /* XXX For now we're using only 48 bits sequence numbers */ dh->dccph_x = 1; - dp->dccps_awh = dp->dccps_gss; + dccp_update_gss(sk, dcb->dccpd_seq); dccp_hdr_set_seq(dh, dp->dccps_gss); if (set_ack) dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), ackno); @@ -109,6 +116,11 @@ static int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb) case DCCP_PKT_REQUEST: dccp_hdr_request(skb)->dccph_req_service = dp->dccps_service; + /* + * Limit Ack window to ISS <= P.ackno <= GSS, so that + * only Responses to Requests we sent are considered. + */ + dp->dccps_awl = dp->dccps_iss; break; case DCCP_PKT_RESET: dccp_hdr_reset(skb)->dccph_reset_code = @@ -123,38 +135,59 @@ static int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb) DCCP_INC_STATS(DCCP_MIB_OUTSEGS); - memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); err = icsk->icsk_af_ops->queue_xmit(skb, 0); return net_xmit_eval(err); } return -ENOBUFS; } +/** + * dccp_determine_ccmps - Find out about CCID-specfic packet-size limits + * We only consider the HC-sender CCID for setting the CCMPS (RFC 4340, 14.), + * since the RX CCID is restricted to feedback packets (Acks), which are small + * in comparison with the data traffic. A value of 0 means "no current CCMPS". + */ +static u32 dccp_determine_ccmps(const struct dccp_sock *dp) +{ + const struct ccid *tx_ccid = dp->dccps_hc_tx_ccid; + + if (tx_ccid == NULL || tx_ccid->ccid_ops == NULL) + return 0; + return tx_ccid->ccid_ops->ccid_ccmps; +} + unsigned int dccp_sync_mss(struct sock *sk, u32 pmtu) { struct inet_connection_sock *icsk = inet_csk(sk); struct dccp_sock *dp = dccp_sk(sk); - int mss_now = (pmtu - icsk->icsk_af_ops->net_header_len - - sizeof(struct dccp_hdr) - sizeof(struct dccp_hdr_ext)); + u32 ccmps = dccp_determine_ccmps(dp); + u32 cur_mps = ccmps ? min(pmtu, ccmps) : pmtu; - /* Now subtract optional transport overhead */ - mss_now -= icsk->icsk_ext_hdr_len; + /* Account for header lengths and IPv4/v6 option overhead */ + cur_mps -= (icsk->icsk_af_ops->net_header_len + icsk->icsk_ext_hdr_len + + sizeof(struct dccp_hdr) + sizeof(struct dccp_hdr_ext)); /* - * FIXME: this should come from the CCID infrastructure, where, say, - * TFRC will say it wants TIMESTAMPS, ELAPSED time, etc, for now lets - * put a rough estimate for NDP + TIMESTAMP + TIMESTAMP_ECHO + ELAPSED - * TIME + TFRC_OPT_LOSS_EVENT_RATE + TFRC_OPT_RECEIVE_RATE + padding to - * make it a multiple of 4 + * Leave enough headroom for common DCCP header options. + * This only considers options which may appear on DCCP-Data packets, as + * per table 3 in RFC 4340, 5.8. When running out of space for other + * options (eg. Ack Vector which can take up to 255 bytes), it is better + * to schedule a separate Ack. Thus we leave headroom for the following: + * - 1 byte for Slow Receiver (11.6) + * - 6 bytes for Timestamp (13.1) + * - 10 bytes for Timestamp Echo (13.3) + * - 8 bytes for NDP count (7.7, when activated) + * - 6 bytes for Data Checksum (9.3) + * - %DCCPAV_MIN_OPTLEN bytes for Ack Vector size (11.4, when enabled) */ - - mss_now -= ((5 + 6 + 10 + 6 + 6 + 6 + 3) / 4) * 4; + cur_mps -= roundup(1 + 6 + 10 + dp->dccps_send_ndp_count * 8 + 6 + + (dp->dccps_hc_rx_ackvec ? DCCPAV_MIN_OPTLEN : 0), 4); /* And store cached results */ icsk->icsk_pmtu_cookie = pmtu; - dp->dccps_mss_cache = mss_now; + dp->dccps_mss_cache = cur_mps; - return mss_now; + return cur_mps; } EXPORT_SYMBOL_GPL(dccp_sync_mss); @@ -167,40 +200,45 @@ void dccp_write_space(struct sock *sk) wake_up_interruptible(sk->sk_sleep); /* Should agree with poll, otherwise some programs break */ if (sock_writeable(sk)) - sk_wake_async(sk, 2, POLL_OUT); + sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT); read_unlock(&sk->sk_callback_lock); } /** * dccp_wait_for_ccid - Wait for ccid to tell us we can send a packet - * @sk: socket to wait for + * @sk: socket to wait for + * @skb: current skb to pass on for waiting + * @delay: sleep timeout in milliseconds (> 0) + * This function is called by default when the socket is closed, and + * when a non-zero linger time is set on the socket. For consistency */ -static int dccp_wait_for_ccid(struct sock *sk, struct sk_buff *skb) +static int dccp_wait_for_ccid(struct sock *sk, struct sk_buff *skb, int delay) { struct dccp_sock *dp = dccp_sk(sk); DEFINE_WAIT(wait); - unsigned long delay; + unsigned long jiffdelay; int rc; - while (1) { + do { + dccp_pr_debug("delayed send by %d msec\n", delay); + jiffdelay = msecs_to_jiffies(delay); + prepare_to_wait(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE); + sk->sk_write_pending++; + release_sock(sk); + schedule_timeout(jiffdelay); + lock_sock(sk); + sk->sk_write_pending--; + if (sk->sk_err) goto do_error; if (signal_pending(current)) goto do_interrupted; rc = ccid_hc_tx_send_packet(dp->dccps_hc_tx_ccid, sk, skb); - if (rc <= 0) - break; - delay = msecs_to_jiffies(rc); - sk->sk_write_pending++; - release_sock(sk); - schedule_timeout(delay); - lock_sock(sk); - sk->sk_write_pending--; - } + } while ((delay = rc) > 0); out: finish_wait(sk->sk_sleep, &wait); return rc; @@ -213,19 +251,6 @@ do_interrupted: goto out; } -static void dccp_write_xmit_timer(unsigned long data) { - struct sock *sk = (struct sock *)data; - struct dccp_sock *dp = dccp_sk(sk); - - bh_lock_sock(sk); - if (sock_owned_by_user(sk)) - sk_reset_timer(sk, &dp->dccps_xmit_timer, jiffies+1); - else - dccp_write_xmit(sk, 0); - bh_unlock_sock(sk); - sock_put(sk); -} - void dccp_write_xmit(struct sock *sk, int block) { struct dccp_sock *dp = dccp_sk(sk); @@ -240,7 +265,7 @@ void dccp_write_xmit(struct sock *sk, int block) msecs_to_jiffies(err)+jiffies); break; } else - err = dccp_wait_for_ccid(sk, skb); + err = dccp_wait_for_ccid(sk, skb, err); if (err && err != -EINTR) DCCP_BUG("err=%d after dccp_wait_for_ccid", err); } @@ -251,7 +276,20 @@ void dccp_write_xmit(struct sock *sk, int block) const int len = skb->len; if (sk->sk_state == DCCP_PARTOPEN) { - /* See 8.1.5. Handshake Completion */ + const u32 cur_mps = dp->dccps_mss_cache - DCCP_FEATNEG_OVERHEAD; + /* + * See 8.1.5 - Handshake Completion. + * + * For robustness we resend Confirm options until the client has + * entered OPEN. During the initial feature negotiation, the MPS + * is smaller than usual, reduced by the Change/Confirm options. + */ + if (!list_empty(&dp->dccps_featneg) && len > cur_mps) { + DCCP_WARN("Payload too large (%d) for featneg.\n", len); + dccp_send_ack(sk); + dccp_feat_list_purge(&dp->dccps_featneg); + } + inet_csk_schedule_ack(sk); inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, inet_csk(sk)->icsk_rto, @@ -268,20 +306,32 @@ void dccp_write_xmit(struct sock *sk, int block) DCCP_BUG("err=%d after ccid_hc_tx_packet_sent", err); } else { - dccp_pr_debug("packet discarded\n"); - kfree(skb); + dccp_pr_debug("packet discarded due to err=%d\n", err); + kfree_skb(skb); } } } -int dccp_retransmit_skb(struct sock *sk, struct sk_buff *skb) +/** + * dccp_retransmit_skb - Retransmit Request, Close, or CloseReq packets + * There are only four retransmittable packet types in DCCP: + * - Request in client-REQUEST state (sec. 8.1.1), + * - CloseReq in server-CLOSEREQ state (sec. 8.3), + * - Close in node-CLOSING state (sec. 8.3), + * - Acks in client-PARTOPEN state (sec. 8.1.5, handled by dccp_delack_timer()). + * This function expects sk->sk_send_head to contain the original skb. + */ +int dccp_retransmit_skb(struct sock *sk) { + WARN_ON(sk->sk_send_head == NULL); + if (inet_csk(sk)->icsk_af_ops->rebuild_header(sk) != 0) return -EHOSTUNREACH; /* Routing failure or similar. */ - return dccp_transmit_skb(sk, (skb_cloned(skb) ? - pskb_copy(skb, GFP_ATOMIC): - skb_clone(skb, GFP_ATOMIC))); + /* this count is used to distinguish original and retransmitted skb */ + inet_csk(sk)->icsk_retransmits++; + + return dccp_transmit_skb(sk, skb_clone(sk->sk_send_head, GFP_ATOMIC)); } struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst, @@ -300,7 +350,7 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst, /* Reserve space for headers. */ skb_reserve(skb, sk->sk_prot->max_header); - skb->dst = dst_clone(dst); + skb_dst_set(skb, dst_clone(dst)); dreq = dccp_rsk(req); if (inet_rsk(req)->acked) /* increase ISS upon retransmission */ @@ -308,15 +358,17 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst, DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESPONSE; DCCP_SKB_CB(skb)->dccpd_seq = dreq->dreq_iss; - if (dccp_insert_options(sk, skb)) { - kfree_skb(skb); - return NULL; - } + /* Resolve feature dependencies resulting from choice of CCID */ + if (dccp_feat_server_ccid_dependencies(dreq)) + goto response_failed; + + if (dccp_insert_options_rsk(dreq, skb)) + goto response_failed; /* Build and checksum header */ dh = dccp_zeroed_hdr(skb, dccp_header_size); - dh->dccph_sport = inet_sk(sk)->sport; + dh->dccph_sport = inet_rsk(req)->loc_port; dh->dccph_dport = inet_rsk(req)->rmt_port; dh->dccph_doff = (dccp_header_size + DCCP_SKB_CB(skb)->dccpd_opt_len) / 4; @@ -332,76 +384,88 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst, inet_rsk(req)->acked = 1; DCCP_INC_STATS(DCCP_MIB_OUTSEGS); return skb; +response_failed: + kfree_skb(skb); + return NULL; } EXPORT_SYMBOL_GPL(dccp_make_response); -static struct sk_buff *dccp_make_reset(struct sock *sk, struct dst_entry *dst, - const enum dccp_reset_codes code) +/* answer offending packet in @rcv_skb with Reset from control socket @ctl */ +struct sk_buff *dccp_ctl_make_reset(struct sock *sk, struct sk_buff *rcv_skb) { - struct dccp_hdr *dh; - struct dccp_sock *dp = dccp_sk(sk); - const u32 dccp_header_size = sizeof(struct dccp_hdr) + - sizeof(struct dccp_hdr_ext) + - sizeof(struct dccp_hdr_reset); - struct sk_buff *skb = sock_wmalloc(sk, sk->sk_prot->max_header, 1, - GFP_ATOMIC); + struct dccp_hdr *rxdh = dccp_hdr(rcv_skb), *dh; + struct dccp_skb_cb *dcb = DCCP_SKB_CB(rcv_skb); + const u32 dccp_hdr_reset_len = sizeof(struct dccp_hdr) + + sizeof(struct dccp_hdr_ext) + + sizeof(struct dccp_hdr_reset); + struct dccp_hdr_reset *dhr; + struct sk_buff *skb; + + skb = alloc_skb(sk->sk_prot->max_header, GFP_ATOMIC); if (skb == NULL) return NULL; - /* Reserve space for headers. */ skb_reserve(skb, sk->sk_prot->max_header); - skb->dst = dst_clone(dst); - - dccp_inc_seqno(&dp->dccps_gss); - - DCCP_SKB_CB(skb)->dccpd_reset_code = code; - DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESET; - DCCP_SKB_CB(skb)->dccpd_seq = dp->dccps_gss; - - if (dccp_insert_options(sk, skb)) { - kfree_skb(skb); - return NULL; - } - - dh = dccp_zeroed_hdr(skb, dccp_header_size); - - dh->dccph_sport = inet_sk(sk)->sport; - dh->dccph_dport = inet_sk(sk)->dport; - dh->dccph_doff = (dccp_header_size + - DCCP_SKB_CB(skb)->dccpd_opt_len) / 4; + /* Swap the send and the receive. */ + dh = dccp_zeroed_hdr(skb, dccp_hdr_reset_len); dh->dccph_type = DCCP_PKT_RESET; + dh->dccph_sport = rxdh->dccph_dport; + dh->dccph_dport = rxdh->dccph_sport; + dh->dccph_doff = dccp_hdr_reset_len / 4; dh->dccph_x = 1; - dccp_hdr_set_seq(dh, dp->dccps_gss); - dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dp->dccps_gsr); - dccp_hdr_reset(skb)->dccph_reset_code = code; - inet_csk(sk)->icsk_af_ops->send_check(sk, 0, skb); + dhr = dccp_hdr_reset(skb); + dhr->dccph_reset_code = dcb->dccpd_reset_code; + + switch (dcb->dccpd_reset_code) { + case DCCP_RESET_CODE_PACKET_ERROR: + dhr->dccph_reset_data[0] = rxdh->dccph_type; + break; + case DCCP_RESET_CODE_OPTION_ERROR: /* fall through */ + case DCCP_RESET_CODE_MANDATORY_ERROR: + memcpy(dhr->dccph_reset_data, dcb->dccpd_reset_data, 3); + break; + } + /* + * From RFC 4340, 8.3.1: + * If P.ackno exists, set R.seqno := P.ackno + 1. + * Else set R.seqno := 0. + */ + if (dcb->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ) + dccp_hdr_set_seq(dh, ADD48(dcb->dccpd_ack_seq, 1)); + dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dcb->dccpd_seq); - DCCP_INC_STATS(DCCP_MIB_OUTSEGS); + dccp_csum_outgoing(skb); return skb; } +EXPORT_SYMBOL_GPL(dccp_ctl_make_reset); + +/* send Reset on established socket, to close or abort the connection */ int dccp_send_reset(struct sock *sk, enum dccp_reset_codes code) { + struct sk_buff *skb; /* * FIXME: what if rebuild_header fails? * Should we be doing a rebuild_header here? */ - int err = inet_sk_rebuild_header(sk); - - if (err == 0) { - struct sk_buff *skb = dccp_make_reset(sk, sk->sk_dst_cache, - code); - if (skb != NULL) { - memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); - err = inet_csk(sk)->icsk_af_ops->queue_xmit(skb, 0); - return net_xmit_eval(err); - } - } + int err = inet_csk(sk)->icsk_af_ops->rebuild_header(sk); + + if (err != 0) + return err; + + skb = sock_wmalloc(sk, sk->sk_prot->max_header, 1, GFP_ATOMIC); + if (skb == NULL) + return -ENOBUFS; + + /* Reserve space for headers and prepare control bits. */ + skb_reserve(skb, sk->sk_prot->max_header); + DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESET; + DCCP_SKB_CB(skb)->dccpd_reset_code = code; - return err; + return dccp_transmit_skb(sk, skb); } /* @@ -415,28 +479,13 @@ static inline void dccp_connect_init(struct sock *sk) sk->sk_err = 0; sock_reset_flag(sk, SOCK_DONE); - - dccp_sync_mss(sk, dst_mtu(dst)); - /* - * SWL and AWL are initially adjusted so that they are not less than - * the initial Sequence Numbers received and sent, respectively: - * SWL := max(GSR + 1 - floor(W/4), ISR), - * AWL := max(GSS - W' + 1, ISS). - * These adjustments MUST be applied only at the beginning of the - * connection. - */ - dccp_update_gss(sk, dp->dccps_iss); - dccp_set_seqno(&dp->dccps_awl, max48(dp->dccps_awl, dp->dccps_iss)); + dccp_sync_mss(sk, dst_mtu(dst)); - /* S.GAR - greatest valid acknowledgement number received on a non-Sync; - * initialized to S.ISS (sec. 8.5) */ + /* Initialise GAR as per 8.5; AWL/AWH are set in dccp_transmit_skb() */ dp->dccps_gar = dp->dccps_iss; icsk->icsk_retransmits = 0; - init_timer(&dp->dccps_xmit_timer); - dp->dccps_xmit_timer.data = (unsigned long)sk; - dp->dccps_xmit_timer.function = dccp_write_xmit_timer; } int dccp_connect(struct sock *sk) @@ -444,6 +493,10 @@ int dccp_connect(struct sock *sk) struct sk_buff *skb; struct inet_connection_sock *icsk = inet_csk(sk); + /* do not connect if feature negotiation setup fails */ + if (dccp_feat_finalise_settings(dccp_sk(sk))) + return -EPROTO; + dccp_connect_init(sk); skb = alloc_skb(sk->sk_prot->max_header, sk->sk_allocation); @@ -492,6 +545,8 @@ void dccp_send_ack(struct sock *sk) EXPORT_SYMBOL_GPL(dccp_send_ack); +#if 0 +/* FIXME: Is this still necessary (11.3) - currently nowhere used by DCCP. */ void dccp_send_delayed_ack(struct sock *sk) { struct inet_connection_sock *icsk = inet_csk(sk); @@ -521,8 +576,9 @@ void dccp_send_delayed_ack(struct sock *sk) icsk->icsk_ack.timeout = timeout; sk_reset_timer(sk, &icsk->icsk_delack_timer, timeout); } +#endif -void dccp_send_sync(struct sock *sk, const u64 seq, +void dccp_send_sync(struct sock *sk, const u64 ackno, const enum dccp_pkt_type pkt_type) { /* @@ -532,14 +588,16 @@ void dccp_send_sync(struct sock *sk, const u64 seq, */ struct sk_buff *skb = alloc_skb(sk->sk_prot->max_header, GFP_ATOMIC); - if (skb == NULL) + if (skb == NULL) { /* FIXME: how to make sure the sync is sent? */ + DCCP_CRIT("could not send %s", dccp_packet_name(pkt_type)); return; + } /* Reserve space for headers and prepare control bits. */ skb_reserve(skb, sk->sk_prot->max_header); DCCP_SKB_CB(skb)->dccpd_type = pkt_type; - DCCP_SKB_CB(skb)->dccpd_seq = seq; + DCCP_SKB_CB(skb)->dccpd_ack_seq = ackno; dccp_transmit_skb(sk, skb); } @@ -563,14 +621,27 @@ void dccp_send_close(struct sock *sk, const int active) /* Reserve space for headers and prepare control bits. */ skb_reserve(skb, sk->sk_prot->max_header); - DCCP_SKB_CB(skb)->dccpd_type = dp->dccps_role == DCCP_ROLE_CLIENT ? - DCCP_PKT_CLOSE : DCCP_PKT_CLOSEREQ; + if (dp->dccps_role == DCCP_ROLE_SERVER && !dp->dccps_server_timewait) + DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_CLOSEREQ; + else + DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_CLOSE; if (active) { dccp_write_xmit(sk, 1); dccp_skb_entail(sk, skb); dccp_transmit_skb(sk, skb_clone(skb, prio)); - /* FIXME do we need a retransmit timer here? */ + /* + * Retransmission timer for active-close: RFC 4340, 8.3 requires + * to retransmit the Close/CloseReq until the CLOSING/CLOSEREQ + * state can be left. The initial timeout is 2 RTTs. + * Since RTT measurement is done by the CCIDs, there is no easy + * way to get an RTT sample. The fallback RTT from RFC 4340, 3.4 + * is too low (200ms); we use a high value to avoid unnecessary + * retransmissions when the link RTT is > 0.2 seconds. + * FIXME: Let main module sample RTTs and use that instead. + */ + inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS, + DCCP_TIMEOUT_INIT, DCCP_RTO_MAX); } else dccp_transmit_skb(sk, skb); }