diff options
Diffstat (limited to 'net/dccp/ccids')
| -rw-r--r-- | net/dccp/ccids/Kconfig | 55 | ||||
| -rw-r--r-- | net/dccp/ccids/ccid2.c | 793 | ||||
| -rw-r--r-- | net/dccp/ccids/ccid2.h | 121 | ||||
| -rw-r--r-- | net/dccp/ccids/ccid3.c | 866 | ||||
| -rw-r--r-- | net/dccp/ccids/ccid3.h | 148 | ||||
| -rw-r--r-- | net/dccp/ccids/lib/loss_interval.c | 184 | ||||
| -rw-r--r-- | net/dccp/ccids/lib/loss_interval.h | 69 | ||||
| -rw-r--r-- | net/dccp/ccids/lib/packet_history.c | 439 | ||||
| -rw-r--r-- | net/dccp/ccids/lib/packet_history.h | 142 | ||||
| -rw-r--r-- | net/dccp/ccids/lib/tfrc.c | 46 | ||||
| -rw-r--r-- | net/dccp/ccids/lib/tfrc.h | 73 | ||||
| -rw-r--r-- | net/dccp/ccids/lib/tfrc_equation.c | 702 |
12 files changed, 3638 insertions, 0 deletions
diff --git a/net/dccp/ccids/Kconfig b/net/dccp/ccids/Kconfig new file mode 100644 index 000000000..a3eeb84d1 --- /dev/null +++ b/net/dccp/ccids/Kconfig @@ -0,0 +1,55 @@ +# SPDX-License-Identifier: GPL-2.0-only +menu "DCCP CCIDs Configuration" + +config IP_DCCP_CCID2_DEBUG + bool "CCID-2 debugging messages" + help + Enable CCID-2 specific debugging messages. + + The debugging output can additionally be toggled by setting the + ccid2_debug parameter to 0 or 1. + + If in doubt, say N. + +config IP_DCCP_CCID3 + bool "CCID-3 (TCP-Friendly)" + def_bool y if (IP_DCCP = y || IP_DCCP = m) + help + CCID-3 denotes TCP-Friendly Rate Control (TFRC), an equation-based + rate-controlled congestion control mechanism. TFRC is designed to + be reasonably fair when competing for bandwidth with TCP-like flows, + where a flow is "reasonably fair" if its sending rate is generally + within a factor of two of the sending rate of a TCP flow under the + same conditions. However, TFRC has a much lower variation of + throughput over time compared with TCP, which makes CCID-3 more + suitable than CCID-2 for applications such streaming media where a + relatively smooth sending rate is of importance. + + CCID-3 is further described in RFC 4342, + https://www.ietf.org/rfc/rfc4342.txt + + The TFRC congestion control algorithms were initially described in + RFC 5348. + + This text was extracted from RFC 4340 (sec. 10.2), + https://www.ietf.org/rfc/rfc4340.txt + + If in doubt, say N. + +config IP_DCCP_CCID3_DEBUG + bool "CCID-3 debugging messages" + depends on IP_DCCP_CCID3 + help + Enable CCID-3 specific debugging messages. + + The debugging output can additionally be toggled by setting the + ccid3_debug parameter to 0 or 1. + + If in doubt, say N. + +config IP_DCCP_TFRC_LIB + def_bool y if IP_DCCP_CCID3 + +config IP_DCCP_TFRC_DEBUG + def_bool y if IP_DCCP_CCID3_DEBUG +endmenu diff --git a/net/dccp/ccids/ccid2.c b/net/dccp/ccids/ccid2.c new file mode 100644 index 000000000..4d9823d6d --- /dev/null +++ b/net/dccp/ccids/ccid2.c @@ -0,0 +1,793 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * Copyright (c) 2005, 2006 Andrea Bittau <a.bittau@cs.ucl.ac.uk> + * + * Changes to meet Linux coding standards, and DCCP infrastructure fixes. + * + * Copyright (c) 2006 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + */ + +/* + * This implementation should follow RFC 4341 + */ +#include <linux/slab.h> +#include "../feat.h" +#include "ccid2.h" + + +#ifdef CONFIG_IP_DCCP_CCID2_DEBUG +static bool ccid2_debug; +#define ccid2_pr_debug(format, a...) DCCP_PR_DEBUG(ccid2_debug, format, ##a) +#else +#define ccid2_pr_debug(format, a...) +#endif + +static int ccid2_hc_tx_alloc_seq(struct ccid2_hc_tx_sock *hc) +{ + struct ccid2_seq *seqp; + int i; + + /* check if we have space to preserve the pointer to the buffer */ + if (hc->tx_seqbufc >= (sizeof(hc->tx_seqbuf) / + sizeof(struct ccid2_seq *))) + return -ENOMEM; + + /* allocate buffer and initialize linked list */ + seqp = kmalloc_array(CCID2_SEQBUF_LEN, sizeof(struct ccid2_seq), + gfp_any()); + if (seqp == NULL) + return -ENOMEM; + + for (i = 0; i < (CCID2_SEQBUF_LEN - 1); i++) { + seqp[i].ccid2s_next = &seqp[i + 1]; + seqp[i + 1].ccid2s_prev = &seqp[i]; + } + seqp[CCID2_SEQBUF_LEN - 1].ccid2s_next = seqp; + seqp->ccid2s_prev = &seqp[CCID2_SEQBUF_LEN - 1]; + + /* This is the first allocation. Initiate the head and tail. */ + if (hc->tx_seqbufc == 0) + hc->tx_seqh = hc->tx_seqt = seqp; + else { + /* link the existing list with the one we just created */ + hc->tx_seqh->ccid2s_next = seqp; + seqp->ccid2s_prev = hc->tx_seqh; + + hc->tx_seqt->ccid2s_prev = &seqp[CCID2_SEQBUF_LEN - 1]; + seqp[CCID2_SEQBUF_LEN - 1].ccid2s_next = hc->tx_seqt; + } + + /* store the original pointer to the buffer so we can free it */ + hc->tx_seqbuf[hc->tx_seqbufc] = seqp; + hc->tx_seqbufc++; + + return 0; +} + +static int ccid2_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb) +{ + if (ccid2_cwnd_network_limited(ccid2_hc_tx_sk(sk))) + return CCID_PACKET_WILL_DEQUEUE_LATER; + return CCID_PACKET_SEND_AT_ONCE; +} + +static void ccid2_change_l_ack_ratio(struct sock *sk, u32 val) +{ + u32 max_ratio = DIV_ROUND_UP(ccid2_hc_tx_sk(sk)->tx_cwnd, 2); + + /* + * Ensure that Ack Ratio does not exceed ceil(cwnd/2), which is (2) from + * RFC 4341, 6.1.2. We ignore the statement that Ack Ratio 2 is always + * acceptable since this causes starvation/deadlock whenever cwnd < 2. + * The same problem arises when Ack Ratio is 0 (ie. Ack Ratio disabled). + */ + if (val == 0 || val > max_ratio) { + DCCP_WARN("Limiting Ack Ratio (%u) to %u\n", val, max_ratio); + val = max_ratio; + } + dccp_feat_signal_nn_change(sk, DCCPF_ACK_RATIO, + min_t(u32, val, DCCPF_ACK_RATIO_MAX)); +} + +static void ccid2_check_l_ack_ratio(struct sock *sk) +{ + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + + /* + * After a loss, idle period, application limited period, or RTO we + * need to check that the ack ratio is still less than the congestion + * window. Otherwise, we will send an entire congestion window of + * packets and got no response because we haven't sent ack ratio + * packets yet. + * If the ack ratio does need to be reduced, we reduce it to half of + * the congestion window (or 1 if that's zero) instead of to the + * congestion window. This prevents problems if one ack is lost. + */ + if (dccp_feat_nn_get(sk, DCCPF_ACK_RATIO) > hc->tx_cwnd) + ccid2_change_l_ack_ratio(sk, hc->tx_cwnd/2 ? : 1U); +} + +static void ccid2_change_l_seq_window(struct sock *sk, u64 val) +{ + dccp_feat_signal_nn_change(sk, DCCPF_SEQUENCE_WINDOW, + clamp_val(val, DCCPF_SEQ_WMIN, + DCCPF_SEQ_WMAX)); +} + +static void dccp_tasklet_schedule(struct sock *sk) +{ + struct tasklet_struct *t = &dccp_sk(sk)->dccps_xmitlet; + + if (!test_and_set_bit(TASKLET_STATE_SCHED, &t->state)) { + sock_hold(sk); + __tasklet_schedule(t); + } +} + +static void ccid2_hc_tx_rto_expire(struct timer_list *t) +{ + struct ccid2_hc_tx_sock *hc = from_timer(hc, t, tx_rtotimer); + struct sock *sk = hc->sk; + const bool sender_was_blocked = ccid2_cwnd_network_limited(hc); + + bh_lock_sock(sk); + if (sock_owned_by_user(sk)) { + sk_reset_timer(sk, &hc->tx_rtotimer, jiffies + HZ / 5); + goto out; + } + + ccid2_pr_debug("RTO_EXPIRE\n"); + + if (sk->sk_state == DCCP_CLOSED) + goto out; + + /* back-off timer */ + hc->tx_rto <<= 1; + if (hc->tx_rto > DCCP_RTO_MAX) + hc->tx_rto = DCCP_RTO_MAX; + + /* adjust pipe, cwnd etc */ + hc->tx_ssthresh = hc->tx_cwnd / 2; + if (hc->tx_ssthresh < 2) + hc->tx_ssthresh = 2; + hc->tx_cwnd = 1; + hc->tx_pipe = 0; + + /* clear state about stuff we sent */ + hc->tx_seqt = hc->tx_seqh; + hc->tx_packets_acked = 0; + + /* clear ack ratio state. */ + hc->tx_rpseq = 0; + hc->tx_rpdupack = -1; + ccid2_change_l_ack_ratio(sk, 1); + + /* if we were blocked before, we may now send cwnd=1 packet */ + if (sender_was_blocked) + dccp_tasklet_schedule(sk); + /* restart backed-off timer */ + sk_reset_timer(sk, &hc->tx_rtotimer, jiffies + hc->tx_rto); +out: + bh_unlock_sock(sk); + sock_put(sk); +} + +/* + * Congestion window validation (RFC 2861). + */ +static bool ccid2_do_cwv = true; +module_param(ccid2_do_cwv, bool, 0644); +MODULE_PARM_DESC(ccid2_do_cwv, "Perform RFC2861 Congestion Window Validation"); + +/** + * ccid2_update_used_window - Track how much of cwnd is actually used + * @hc: socket to update window + * @new_wnd: new window values to add into the filter + * + * This is done in addition to CWV. The sender needs to have an idea of how many + * packets may be in flight, to set the local Sequence Window value accordingly + * (RFC 4340, 7.5.2). The CWV mechanism is exploited to keep track of the + * maximum-used window. We use an EWMA low-pass filter to filter out noise. + */ +static void ccid2_update_used_window(struct ccid2_hc_tx_sock *hc, u32 new_wnd) +{ + hc->tx_expected_wnd = (3 * hc->tx_expected_wnd + new_wnd) / 4; +} + +/* This borrows the code of tcp_cwnd_application_limited() */ +static void ccid2_cwnd_application_limited(struct sock *sk, const u32 now) +{ + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + /* don't reduce cwnd below the initial window (IW) */ + u32 init_win = rfc3390_bytes_to_packets(dccp_sk(sk)->dccps_mss_cache), + win_used = max(hc->tx_cwnd_used, init_win); + + if (win_used < hc->tx_cwnd) { + hc->tx_ssthresh = max(hc->tx_ssthresh, + (hc->tx_cwnd >> 1) + (hc->tx_cwnd >> 2)); + hc->tx_cwnd = (hc->tx_cwnd + win_used) >> 1; + } + hc->tx_cwnd_used = 0; + hc->tx_cwnd_stamp = now; + + ccid2_check_l_ack_ratio(sk); +} + +/* This borrows the code of tcp_cwnd_restart() */ +static void ccid2_cwnd_restart(struct sock *sk, const u32 now) +{ + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + u32 cwnd = hc->tx_cwnd, restart_cwnd, + iwnd = rfc3390_bytes_to_packets(dccp_sk(sk)->dccps_mss_cache); + s32 delta = now - hc->tx_lsndtime; + + hc->tx_ssthresh = max(hc->tx_ssthresh, (cwnd >> 1) + (cwnd >> 2)); + + /* don't reduce cwnd below the initial window (IW) */ + restart_cwnd = min(cwnd, iwnd); + + while ((delta -= hc->tx_rto) >= 0 && cwnd > restart_cwnd) + cwnd >>= 1; + hc->tx_cwnd = max(cwnd, restart_cwnd); + hc->tx_cwnd_stamp = now; + hc->tx_cwnd_used = 0; + + ccid2_check_l_ack_ratio(sk); +} + +static void ccid2_hc_tx_packet_sent(struct sock *sk, unsigned int len) +{ + struct dccp_sock *dp = dccp_sk(sk); + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + const u32 now = ccid2_jiffies32; + struct ccid2_seq *next; + + /* slow-start after idle periods (RFC 2581, RFC 2861) */ + if (ccid2_do_cwv && !hc->tx_pipe && + (s32)(now - hc->tx_lsndtime) >= hc->tx_rto) + ccid2_cwnd_restart(sk, now); + + hc->tx_lsndtime = now; + hc->tx_pipe += 1; + + /* see whether cwnd was fully used (RFC 2861), update expected window */ + if (ccid2_cwnd_network_limited(hc)) { + ccid2_update_used_window(hc, hc->tx_cwnd); + hc->tx_cwnd_used = 0; + hc->tx_cwnd_stamp = now; + } else { + if (hc->tx_pipe > hc->tx_cwnd_used) + hc->tx_cwnd_used = hc->tx_pipe; + + ccid2_update_used_window(hc, hc->tx_cwnd_used); + + if (ccid2_do_cwv && (s32)(now - hc->tx_cwnd_stamp) >= hc->tx_rto) + ccid2_cwnd_application_limited(sk, now); + } + + hc->tx_seqh->ccid2s_seq = dp->dccps_gss; + hc->tx_seqh->ccid2s_acked = 0; + hc->tx_seqh->ccid2s_sent = now; + + next = hc->tx_seqh->ccid2s_next; + /* check if we need to alloc more space */ + if (next == hc->tx_seqt) { + if (ccid2_hc_tx_alloc_seq(hc)) { + DCCP_CRIT("packet history - out of memory!"); + /* FIXME: find a more graceful way to bail out */ + return; + } + next = hc->tx_seqh->ccid2s_next; + BUG_ON(next == hc->tx_seqt); + } + hc->tx_seqh = next; + + ccid2_pr_debug("cwnd=%d pipe=%d\n", hc->tx_cwnd, hc->tx_pipe); + + /* + * FIXME: The code below is broken and the variables have been removed + * from the socket struct. The `ackloss' variable was always set to 0, + * and with arsent there are several problems: + * (i) it doesn't just count the number of Acks, but all sent packets; + * (ii) it is expressed in # of packets, not # of windows, so the + * comparison below uses the wrong formula: Appendix A of RFC 4341 + * comes up with the number K = cwnd / (R^2 - R) of consecutive windows + * of data with no lost or marked Ack packets. If arsent were the # of + * consecutive Acks received without loss, then Ack Ratio needs to be + * decreased by 1 when + * arsent >= K * cwnd / R = cwnd^2 / (R^3 - R^2) + * where cwnd / R is the number of Acks received per window of data + * (cf. RFC 4341, App. A). The problems are that + * - arsent counts other packets as well; + * - the comparison uses a formula different from RFC 4341; + * - computing a cubic/quadratic equation each time is too complicated. + * Hence a different algorithm is needed. + */ +#if 0 + /* Ack Ratio. Need to maintain a concept of how many windows we sent */ + hc->tx_arsent++; + /* We had an ack loss in this window... */ + if (hc->tx_ackloss) { + if (hc->tx_arsent >= hc->tx_cwnd) { + hc->tx_arsent = 0; + hc->tx_ackloss = 0; + } + } else { + /* No acks lost up to now... */ + /* decrease ack ratio if enough packets were sent */ + if (dp->dccps_l_ack_ratio > 1) { + /* XXX don't calculate denominator each time */ + int denom = dp->dccps_l_ack_ratio * dp->dccps_l_ack_ratio - + dp->dccps_l_ack_ratio; + + denom = hc->tx_cwnd * hc->tx_cwnd / denom; + + if (hc->tx_arsent >= denom) { + ccid2_change_l_ack_ratio(sk, dp->dccps_l_ack_ratio - 1); + hc->tx_arsent = 0; + } + } else { + /* we can't increase ack ratio further [1] */ + hc->tx_arsent = 0; /* or maybe set it to cwnd*/ + } + } +#endif + + sk_reset_timer(sk, &hc->tx_rtotimer, jiffies + hc->tx_rto); + +#ifdef CONFIG_IP_DCCP_CCID2_DEBUG + do { + struct ccid2_seq *seqp = hc->tx_seqt; + + while (seqp != hc->tx_seqh) { + ccid2_pr_debug("out seq=%llu acked=%d time=%u\n", + (unsigned long long)seqp->ccid2s_seq, + seqp->ccid2s_acked, seqp->ccid2s_sent); + seqp = seqp->ccid2s_next; + } + } while (0); + ccid2_pr_debug("=========\n"); +#endif +} + +/** + * ccid2_rtt_estimator - Sample RTT and compute RTO using RFC2988 algorithm + * @sk: socket to perform estimator on + * + * This code is almost identical with TCP's tcp_rtt_estimator(), since + * - it has a higher sampling frequency (recommended by RFC 1323), + * - the RTO does not collapse into RTT due to RTTVAR going towards zero, + * - it is simple (cf. more complex proposals such as Eifel timer or research + * which suggests that the gain should be set according to window size), + * - in tests it was found to work well with CCID2 [gerrit]. + */ +static void ccid2_rtt_estimator(struct sock *sk, const long mrtt) +{ + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + long m = mrtt ? : 1; + + if (hc->tx_srtt == 0) { + /* First measurement m */ + hc->tx_srtt = m << 3; + hc->tx_mdev = m << 1; + + hc->tx_mdev_max = max(hc->tx_mdev, tcp_rto_min(sk)); + hc->tx_rttvar = hc->tx_mdev_max; + + hc->tx_rtt_seq = dccp_sk(sk)->dccps_gss; + } else { + /* Update scaled SRTT as SRTT += 1/8 * (m - SRTT) */ + m -= (hc->tx_srtt >> 3); + hc->tx_srtt += m; + + /* Similarly, update scaled mdev with regard to |m| */ + if (m < 0) { + m = -m; + m -= (hc->tx_mdev >> 2); + /* + * This neutralises RTO increase when RTT < SRTT - mdev + * (see P. Sarolahti, A. Kuznetsov,"Congestion Control + * in Linux TCP", USENIX 2002, pp. 49-62). + */ + if (m > 0) + m >>= 3; + } else { + m -= (hc->tx_mdev >> 2); + } + hc->tx_mdev += m; + + if (hc->tx_mdev > hc->tx_mdev_max) { + hc->tx_mdev_max = hc->tx_mdev; + if (hc->tx_mdev_max > hc->tx_rttvar) + hc->tx_rttvar = hc->tx_mdev_max; + } + + /* + * Decay RTTVAR at most once per flight, exploiting that + * 1) pipe <= cwnd <= Sequence_Window = W (RFC 4340, 7.5.2) + * 2) AWL = GSS-W+1 <= GAR <= GSS (RFC 4340, 7.5.1) + * GAR is a useful bound for FlightSize = pipe. + * AWL is probably too low here, as it over-estimates pipe. + */ + if (after48(dccp_sk(sk)->dccps_gar, hc->tx_rtt_seq)) { + if (hc->tx_mdev_max < hc->tx_rttvar) + hc->tx_rttvar -= (hc->tx_rttvar - + hc->tx_mdev_max) >> 2; + hc->tx_rtt_seq = dccp_sk(sk)->dccps_gss; + hc->tx_mdev_max = tcp_rto_min(sk); + } + } + + /* + * Set RTO from SRTT and RTTVAR + * As in TCP, 4 * RTTVAR >= TCP_RTO_MIN, giving a minimum RTO of 200 ms. + * This agrees with RFC 4341, 5: + * "Because DCCP does not retransmit data, DCCP does not require + * TCP's recommended minimum timeout of one second". + */ + hc->tx_rto = (hc->tx_srtt >> 3) + hc->tx_rttvar; + + if (hc->tx_rto > DCCP_RTO_MAX) + hc->tx_rto = DCCP_RTO_MAX; +} + +static void ccid2_new_ack(struct sock *sk, struct ccid2_seq *seqp, + unsigned int *maxincr) +{ + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + struct dccp_sock *dp = dccp_sk(sk); + int r_seq_used = hc->tx_cwnd / dp->dccps_l_ack_ratio; + + if (hc->tx_cwnd < dp->dccps_l_seq_win && + r_seq_used < dp->dccps_r_seq_win) { + if (hc->tx_cwnd < hc->tx_ssthresh) { + if (*maxincr > 0 && ++hc->tx_packets_acked >= 2) { + hc->tx_cwnd += 1; + *maxincr -= 1; + hc->tx_packets_acked = 0; + } + } else if (++hc->tx_packets_acked >= hc->tx_cwnd) { + hc->tx_cwnd += 1; + hc->tx_packets_acked = 0; + } + } + + /* + * Adjust the local sequence window and the ack ratio to allow about + * 5 times the number of packets in the network (RFC 4340 7.5.2) + */ + if (r_seq_used * CCID2_WIN_CHANGE_FACTOR >= dp->dccps_r_seq_win) + ccid2_change_l_ack_ratio(sk, dp->dccps_l_ack_ratio * 2); + else if (r_seq_used * CCID2_WIN_CHANGE_FACTOR < dp->dccps_r_seq_win/2) + ccid2_change_l_ack_ratio(sk, dp->dccps_l_ack_ratio / 2 ? : 1U); + + if (hc->tx_cwnd * CCID2_WIN_CHANGE_FACTOR >= dp->dccps_l_seq_win) + ccid2_change_l_seq_window(sk, dp->dccps_l_seq_win * 2); + else if (hc->tx_cwnd * CCID2_WIN_CHANGE_FACTOR < dp->dccps_l_seq_win/2) + ccid2_change_l_seq_window(sk, dp->dccps_l_seq_win / 2); + + /* + * FIXME: RTT is sampled several times per acknowledgment (for each + * entry in the Ack Vector), instead of once per Ack (as in TCP SACK). + * This causes the RTT to be over-estimated, since the older entries + * in the Ack Vector have earlier sending times. + * The cleanest solution is to not use the ccid2s_sent field at all + * and instead use DCCP timestamps: requires changes in other places. + */ + ccid2_rtt_estimator(sk, ccid2_jiffies32 - seqp->ccid2s_sent); +} + +static void ccid2_congestion_event(struct sock *sk, struct ccid2_seq *seqp) +{ + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + + if ((s32)(seqp->ccid2s_sent - hc->tx_last_cong) < 0) { + ccid2_pr_debug("Multiple losses in an RTT---treating as one\n"); + return; + } + + hc->tx_last_cong = ccid2_jiffies32; + + hc->tx_cwnd = hc->tx_cwnd / 2 ? : 1U; + hc->tx_ssthresh = max(hc->tx_cwnd, 2U); + + ccid2_check_l_ack_ratio(sk); +} + +static int ccid2_hc_tx_parse_options(struct sock *sk, u8 packet_type, + u8 option, u8 *optval, u8 optlen) +{ + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + + switch (option) { + case DCCPO_ACK_VECTOR_0: + case DCCPO_ACK_VECTOR_1: + return dccp_ackvec_parsed_add(&hc->tx_av_chunks, optval, optlen, + option - DCCPO_ACK_VECTOR_0); + } + return 0; +} + +static void ccid2_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb) +{ + struct dccp_sock *dp = dccp_sk(sk); + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + const bool sender_was_blocked = ccid2_cwnd_network_limited(hc); + struct dccp_ackvec_parsed *avp; + u64 ackno, seqno; + struct ccid2_seq *seqp; + int done = 0; + unsigned int maxincr = 0; + + /* check reverse path congestion */ + seqno = DCCP_SKB_CB(skb)->dccpd_seq; + + /* XXX this whole "algorithm" is broken. Need to fix it to keep track + * of the seqnos of the dupacks so that rpseq and rpdupack are correct + * -sorbo. + */ + /* need to bootstrap */ + if (hc->tx_rpdupack == -1) { + hc->tx_rpdupack = 0; + hc->tx_rpseq = seqno; + } else { + /* check if packet is consecutive */ + if (dccp_delta_seqno(hc->tx_rpseq, seqno) == 1) + hc->tx_rpseq = seqno; + /* it's a later packet */ + else if (after48(seqno, hc->tx_rpseq)) { + hc->tx_rpdupack++; + + /* check if we got enough dupacks */ + if (hc->tx_rpdupack >= NUMDUPACK) { + hc->tx_rpdupack = -1; /* XXX lame */ + hc->tx_rpseq = 0; +#ifdef __CCID2_COPES_GRACEFULLY_WITH_ACK_CONGESTION_CONTROL__ + /* + * FIXME: Ack Congestion Control is broken; in + * the current state instabilities occurred with + * Ack Ratios greater than 1; causing hang-ups + * and long RTO timeouts. This needs to be fixed + * before opening up dynamic changes. -- gerrit + */ + ccid2_change_l_ack_ratio(sk, 2 * dp->dccps_l_ack_ratio); +#endif + } + } + } + + /* check forward path congestion */ + if (dccp_packet_without_ack(skb)) + return; + + /* still didn't send out new data packets */ + if (hc->tx_seqh == hc->tx_seqt) + goto done; + + ackno = DCCP_SKB_CB(skb)->dccpd_ack_seq; + if (after48(ackno, hc->tx_high_ack)) + hc->tx_high_ack = ackno; + + seqp = hc->tx_seqt; + while (before48(seqp->ccid2s_seq, ackno)) { + seqp = seqp->ccid2s_next; + if (seqp == hc->tx_seqh) { + seqp = hc->tx_seqh->ccid2s_prev; + break; + } + } + + /* + * In slow-start, cwnd can increase up to a maximum of Ack Ratio/2 + * packets per acknowledgement. Rounding up avoids that cwnd is not + * advanced when Ack Ratio is 1 and gives a slight edge otherwise. + */ + if (hc->tx_cwnd < hc->tx_ssthresh) + maxincr = DIV_ROUND_UP(dp->dccps_l_ack_ratio, 2); + + /* go through all ack vectors */ + list_for_each_entry(avp, &hc->tx_av_chunks, node) { + /* go through this ack vector */ + for (; avp->len--; avp->vec++) { + u64 ackno_end_rl = SUB48(ackno, + dccp_ackvec_runlen(avp->vec)); + + ccid2_pr_debug("ackvec %llu |%u,%u|\n", + (unsigned long long)ackno, + dccp_ackvec_state(avp->vec) >> 6, + dccp_ackvec_runlen(avp->vec)); + /* if the seqno we are analyzing is larger than the + * current ackno, then move towards the tail of our + * seqnos. + */ + while (after48(seqp->ccid2s_seq, ackno)) { + if (seqp == hc->tx_seqt) { + done = 1; + break; + } + seqp = seqp->ccid2s_prev; + } + if (done) + break; + + /* check all seqnos in the range of the vector + * run length + */ + while (between48(seqp->ccid2s_seq,ackno_end_rl,ackno)) { + const u8 state = dccp_ackvec_state(avp->vec); + + /* new packet received or marked */ + if (state != DCCPAV_NOT_RECEIVED && + !seqp->ccid2s_acked) { + if (state == DCCPAV_ECN_MARKED) + ccid2_congestion_event(sk, + seqp); + else + ccid2_new_ack(sk, seqp, + &maxincr); + + seqp->ccid2s_acked = 1; + ccid2_pr_debug("Got ack for %llu\n", + (unsigned long long)seqp->ccid2s_seq); + hc->tx_pipe--; + } + if (seqp == hc->tx_seqt) { + done = 1; + break; + } + seqp = seqp->ccid2s_prev; + } + if (done) + break; + + ackno = SUB48(ackno_end_rl, 1); + } + if (done) + break; + } + + /* The state about what is acked should be correct now + * Check for NUMDUPACK + */ + seqp = hc->tx_seqt; + while (before48(seqp->ccid2s_seq, hc->tx_high_ack)) { + seqp = seqp->ccid2s_next; + if (seqp == hc->tx_seqh) { + seqp = hc->tx_seqh->ccid2s_prev; + break; + } + } + done = 0; + while (1) { + if (seqp->ccid2s_acked) { + done++; + if (done == NUMDUPACK) + break; + } + if (seqp == hc->tx_seqt) + break; + seqp = seqp->ccid2s_prev; + } + + /* If there are at least 3 acknowledgements, anything unacknowledged + * below the last sequence number is considered lost + */ + if (done == NUMDUPACK) { + struct ccid2_seq *last_acked = seqp; + + /* check for lost packets */ + while (1) { + if (!seqp->ccid2s_acked) { + ccid2_pr_debug("Packet lost: %llu\n", + (unsigned long long)seqp->ccid2s_seq); + /* XXX need to traverse from tail -> head in + * order to detect multiple congestion events in + * one ack vector. + */ + ccid2_congestion_event(sk, seqp); + hc->tx_pipe--; + } + if (seqp == hc->tx_seqt) + break; + seqp = seqp->ccid2s_prev; + } + + hc->tx_seqt = last_acked; + } + + /* trim acked packets in tail */ + while (hc->tx_seqt != hc->tx_seqh) { + if (!hc->tx_seqt->ccid2s_acked) + break; + + hc->tx_seqt = hc->tx_seqt->ccid2s_next; + } + + /* restart RTO timer if not all outstanding data has been acked */ + if (hc->tx_pipe == 0) + sk_stop_timer(sk, &hc->tx_rtotimer); + else + sk_reset_timer(sk, &hc->tx_rtotimer, jiffies + hc->tx_rto); +done: + /* check if incoming Acks allow pending packets to be sent */ + if (sender_was_blocked && !ccid2_cwnd_network_limited(hc)) + dccp_tasklet_schedule(sk); + dccp_ackvec_parsed_cleanup(&hc->tx_av_chunks); +} + +static int ccid2_hc_tx_init(struct ccid *ccid, struct sock *sk) +{ + struct ccid2_hc_tx_sock *hc = ccid_priv(ccid); + struct dccp_sock *dp = dccp_sk(sk); + u32 max_ratio; + + /* RFC 4341, 5: initialise ssthresh to arbitrarily high (max) value */ + hc->tx_ssthresh = ~0U; + + /* Use larger initial windows (RFC 4341, section 5). */ + hc->tx_cwnd = rfc3390_bytes_to_packets(dp->dccps_mss_cache); + hc->tx_expected_wnd = hc->tx_cwnd; + + /* Make sure that Ack Ratio is enabled and within bounds. */ + max_ratio = DIV_ROUND_UP(hc->tx_cwnd, 2); + if (dp->dccps_l_ack_ratio == 0 || dp->dccps_l_ack_ratio > max_ratio) + dp->dccps_l_ack_ratio = max_ratio; + + /* XXX init ~ to window size... */ + if (ccid2_hc_tx_alloc_seq(hc)) + return -ENOMEM; + + hc->tx_rto = DCCP_TIMEOUT_INIT; + hc->tx_rpdupack = -1; + hc->tx_last_cong = hc->tx_lsndtime = hc->tx_cwnd_stamp = ccid2_jiffies32; + hc->tx_cwnd_used = 0; + hc->sk = sk; + timer_setup(&hc->tx_rtotimer, ccid2_hc_tx_rto_expire, 0); + INIT_LIST_HEAD(&hc->tx_av_chunks); + return 0; +} + +static void ccid2_hc_tx_exit(struct sock *sk) +{ + struct ccid2_hc_tx_sock *hc = ccid2_hc_tx_sk(sk); + int i; + + sk_stop_timer(sk, &hc->tx_rtotimer); + + for (i = 0; i < hc->tx_seqbufc; i++) + kfree(hc->tx_seqbuf[i]); + hc->tx_seqbufc = 0; + dccp_ackvec_parsed_cleanup(&hc->tx_av_chunks); +} + +static void ccid2_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb) +{ + struct ccid2_hc_rx_sock *hc = ccid2_hc_rx_sk(sk); + + if (!dccp_data_packet(skb)) + return; + + if (++hc->rx_num_data_pkts >= dccp_sk(sk)->dccps_r_ack_ratio) { + dccp_send_ack(sk); + hc->rx_num_data_pkts = 0; + } +} + +struct ccid_operations ccid2_ops = { + .ccid_id = DCCPC_CCID2, + .ccid_name = "TCP-like", + .ccid_hc_tx_obj_size = sizeof(struct ccid2_hc_tx_sock), + .ccid_hc_tx_init = ccid2_hc_tx_init, + .ccid_hc_tx_exit = ccid2_hc_tx_exit, + .ccid_hc_tx_send_packet = ccid2_hc_tx_send_packet, + .ccid_hc_tx_packet_sent = ccid2_hc_tx_packet_sent, + .ccid_hc_tx_parse_options = ccid2_hc_tx_parse_options, + .ccid_hc_tx_packet_recv = ccid2_hc_tx_packet_recv, + .ccid_hc_rx_obj_size = sizeof(struct ccid2_hc_rx_sock), + .ccid_hc_rx_packet_recv = ccid2_hc_rx_packet_recv, +}; + +#ifdef CONFIG_IP_DCCP_CCID2_DEBUG +module_param(ccid2_debug, bool, 0644); +MODULE_PARM_DESC(ccid2_debug, "Enable CCID-2 debug messages"); +#endif diff --git a/net/dccp/ccids/ccid2.h b/net/dccp/ccids/ccid2.h new file mode 100644 index 000000000..330c7b4ec --- /dev/null +++ b/net/dccp/ccids/ccid2.h @@ -0,0 +1,121 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * Copyright (c) 2005 Andrea Bittau <a.bittau@cs.ucl.ac.uk> + */ +#ifndef _DCCP_CCID2_H_ +#define _DCCP_CCID2_H_ + +#include <linux/timer.h> +#include <linux/types.h> +#include "../ccid.h" +#include "../dccp.h" + +/* + * CCID-2 timestamping faces the same issues as TCP timestamping. + * Hence we reuse/share as much of the code as possible. + */ +#define ccid2_jiffies32 ((u32)jiffies) + +/* NUMDUPACK parameter from RFC 4341, p. 6 */ +#define NUMDUPACK 3 + +struct ccid2_seq { + u64 ccid2s_seq; + u32 ccid2s_sent; + int ccid2s_acked; + struct ccid2_seq *ccid2s_prev; + struct ccid2_seq *ccid2s_next; +}; + +#define CCID2_SEQBUF_LEN 1024 +#define CCID2_SEQBUF_MAX 128 + +/* + * Multiple of congestion window to keep the sequence window at + * (RFC 4340 7.5.2) + */ +#define CCID2_WIN_CHANGE_FACTOR 5 + +/** + * struct ccid2_hc_tx_sock - CCID2 TX half connection + * @tx_{cwnd,ssthresh,pipe}: as per RFC 4341, section 5 + * @tx_packets_acked: Ack counter for deriving cwnd growth (RFC 3465) + * @tx_srtt: smoothed RTT estimate, scaled by 2^3 + * @tx_mdev: smoothed RTT variation, scaled by 2^2 + * @tx_mdev_max: maximum of @mdev during one flight + * @tx_rttvar: moving average/maximum of @mdev_max + * @tx_rto: RTO value deriving from SRTT and RTTVAR (RFC 2988) + * @tx_rtt_seq: to decay RTTVAR at most once per flight + * @tx_cwnd_used: actually used cwnd, W_used of RFC 2861 + * @tx_expected_wnd: moving average of @tx_cwnd_used + * @tx_cwnd_stamp: to track idle periods in CWV + * @tx_lsndtime: last time (in jiffies) a data packet was sent + * @tx_rpseq: last consecutive seqno + * @tx_rpdupack: dupacks since rpseq + * @tx_av_chunks: list of Ack Vectors received on current skb + */ +struct ccid2_hc_tx_sock { + u32 tx_cwnd; + u32 tx_ssthresh; + u32 tx_pipe; + u32 tx_packets_acked; + struct ccid2_seq *tx_seqbuf[CCID2_SEQBUF_MAX]; + int tx_seqbufc; + struct ccid2_seq *tx_seqh; + struct ccid2_seq *tx_seqt; + + /* RTT measurement: variables/principles are the same as in TCP */ + u32 tx_srtt, + tx_mdev, + tx_mdev_max, + tx_rttvar, + tx_rto; + u64 tx_rtt_seq:48; + struct timer_list tx_rtotimer; + struct sock *sk; + + /* Congestion Window validation (optional, RFC 2861) */ + u32 tx_cwnd_used, + tx_expected_wnd, + tx_cwnd_stamp, + tx_lsndtime; + + u64 tx_rpseq; + int tx_rpdupack; + u32 tx_last_cong; + u64 tx_high_ack; + struct list_head tx_av_chunks; +}; + +static inline bool ccid2_cwnd_network_limited(struct ccid2_hc_tx_sock *hc) +{ + return hc->tx_pipe >= hc->tx_cwnd; +} + +/* + * Convert RFC 3390 larger initial window into an equivalent number of packets. + * This is based on the numbers specified in RFC 5681, 3.1. + */ +static inline u32 rfc3390_bytes_to_packets(const u32 smss) +{ + return smss <= 1095 ? 4 : (smss > 2190 ? 2 : 3); +} + +/** + * struct ccid2_hc_rx_sock - Receiving end of CCID-2 half-connection + * @rx_num_data_pkts: number of data packets received since last feedback + */ +struct ccid2_hc_rx_sock { + u32 rx_num_data_pkts; +}; + +static inline struct ccid2_hc_tx_sock *ccid2_hc_tx_sk(const struct sock *sk) +{ + return ccid_priv(dccp_sk(sk)->dccps_hc_tx_ccid); +} + +static inline struct ccid2_hc_rx_sock *ccid2_hc_rx_sk(const struct sock *sk) +{ + return ccid_priv(dccp_sk(sk)->dccps_hc_rx_ccid); +} +#endif /* _DCCP_CCID2_H_ */ diff --git a/net/dccp/ccids/ccid3.c b/net/dccp/ccids/ccid3.c new file mode 100644 index 000000000..ca8670f78 --- /dev/null +++ b/net/dccp/ccids/ccid3.c @@ -0,0 +1,866 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * Copyright (c) 2007 The University of Aberdeen, Scotland, UK + * Copyright (c) 2005-7 The University of Waikato, Hamilton, New Zealand. + * Copyright (c) 2005-7 Ian McDonald <ian.mcdonald@jandi.co.nz> + * + * An implementation of the DCCP protocol + * + * This code has been developed by the University of Waikato WAND + * research group. For further information please see https://www.wand.net.nz/ + * + * This code also uses code from Lulea University, rereleased as GPL by its + * authors: + * Copyright (c) 2003 Nils-Erik Mattsson, Joacim Haggmark, Magnus Erixzon + * + * Changes to meet Linux coding standards, to make it meet latest ccid3 draft + * and to make it work as a loadable module in the DCCP stack written by + * Arnaldo Carvalho de Melo <acme@conectiva.com.br>. + * + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + */ +#include "../dccp.h" +#include "ccid3.h" + +#include <asm/unaligned.h> + +#ifdef CONFIG_IP_DCCP_CCID3_DEBUG +static bool ccid3_debug; +#define ccid3_pr_debug(format, a...) DCCP_PR_DEBUG(ccid3_debug, format, ##a) +#else +#define ccid3_pr_debug(format, a...) +#endif + +/* + * Transmitter Half-Connection Routines + */ +#ifdef CONFIG_IP_DCCP_CCID3_DEBUG +static const char *ccid3_tx_state_name(enum ccid3_hc_tx_states state) +{ + static const char *const ccid3_state_names[] = { + [TFRC_SSTATE_NO_SENT] = "NO_SENT", + [TFRC_SSTATE_NO_FBACK] = "NO_FBACK", + [TFRC_SSTATE_FBACK] = "FBACK", + }; + + return ccid3_state_names[state]; +} +#endif + +static void ccid3_hc_tx_set_state(struct sock *sk, + enum ccid3_hc_tx_states state) +{ + struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + enum ccid3_hc_tx_states oldstate = hc->tx_state; + + ccid3_pr_debug("%s(%p) %-8.8s -> %s\n", + dccp_role(sk), sk, ccid3_tx_state_name(oldstate), + ccid3_tx_state_name(state)); + WARN_ON(state == oldstate); + hc->tx_state = state; +} + +/* + * Compute the initial sending rate X_init in the manner of RFC 3390: + * + * X_init = min(4 * s, max(2 * s, 4380 bytes)) / RTT + * + * Note that RFC 3390 uses MSS, RFC 4342 refers to RFC 3390, and rfc3448bis + * (rev-02) clarifies the use of RFC 3390 with regard to the above formula. + * For consistency with other parts of the code, X_init is scaled by 2^6. + */ +static inline u64 rfc3390_initial_rate(struct sock *sk) +{ + const struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + const __u32 w_init = clamp_t(__u32, 4380U, 2 * hc->tx_s, 4 * hc->tx_s); + + return scaled_div(w_init << 6, hc->tx_rtt); +} + +/** + * ccid3_update_send_interval - Calculate new t_ipi = s / X_inst + * @hc: socket to have the send interval updated + * + * This respects the granularity of X_inst (64 * bytes/second). + */ +static void ccid3_update_send_interval(struct ccid3_hc_tx_sock *hc) +{ + hc->tx_t_ipi = scaled_div32(((u64)hc->tx_s) << 6, hc->tx_x); + + DCCP_BUG_ON(hc->tx_t_ipi == 0); + ccid3_pr_debug("t_ipi=%u, s=%u, X=%u\n", hc->tx_t_ipi, + hc->tx_s, (unsigned int)(hc->tx_x >> 6)); +} + +static u32 ccid3_hc_tx_idle_rtt(struct ccid3_hc_tx_sock *hc, ktime_t now) +{ + u32 delta = ktime_us_delta(now, hc->tx_t_last_win_count); + + return delta / hc->tx_rtt; +} + +/** + * ccid3_hc_tx_update_x - Update allowed sending rate X + * @sk: socket to be updated + * @stamp: most recent time if available - can be left NULL. + * + * This function tracks draft rfc3448bis, check there for latest details. + * + * Note: X and X_recv are both stored in units of 64 * bytes/second, to support + * fine-grained resolution of sending rates. This requires scaling by 2^6 + * throughout the code. Only X_calc is unscaled (in bytes/second). + * + */ +static void ccid3_hc_tx_update_x(struct sock *sk, ktime_t *stamp) +{ + struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + __u64 min_rate = 2 * hc->tx_x_recv; + const __u64 old_x = hc->tx_x; + ktime_t now = stamp ? *stamp : ktime_get_real(); + + /* + * Handle IDLE periods: do not reduce below RFC3390 initial sending rate + * when idling [RFC 4342, 5.1]. Definition of idling is from rfc3448bis: + * a sender is idle if it has not sent anything over a 2-RTT-period. + * For consistency with X and X_recv, min_rate is also scaled by 2^6. + */ + if (ccid3_hc_tx_idle_rtt(hc, now) >= 2) { + min_rate = rfc3390_initial_rate(sk); + min_rate = max(min_rate, 2 * hc->tx_x_recv); + } + + if (hc->tx_p > 0) { + + hc->tx_x = min(((__u64)hc->tx_x_calc) << 6, min_rate); + hc->tx_x = max(hc->tx_x, (((__u64)hc->tx_s) << 6) / TFRC_T_MBI); + + } else if (ktime_us_delta(now, hc->tx_t_ld) - (s64)hc->tx_rtt >= 0) { + + hc->tx_x = min(2 * hc->tx_x, min_rate); + hc->tx_x = max(hc->tx_x, + scaled_div(((__u64)hc->tx_s) << 6, hc->tx_rtt)); + hc->tx_t_ld = now; + } + + if (hc->tx_x != old_x) { + ccid3_pr_debug("X_prev=%u, X_now=%u, X_calc=%u, " + "X_recv=%u\n", (unsigned int)(old_x >> 6), + (unsigned int)(hc->tx_x >> 6), hc->tx_x_calc, + (unsigned int)(hc->tx_x_recv >> 6)); + + ccid3_update_send_interval(hc); + } +} + +/** + * ccid3_hc_tx_update_s - Track the mean packet size `s' + * @hc: socket to be updated + * @len: DCCP packet payload size in bytes + * + * cf. RFC 4342, 5.3 and RFC 3448, 4.1 + */ +static inline void ccid3_hc_tx_update_s(struct ccid3_hc_tx_sock *hc, int len) +{ + const u16 old_s = hc->tx_s; + + hc->tx_s = tfrc_ewma(hc->tx_s, len, 9); + + if (hc->tx_s != old_s) + ccid3_update_send_interval(hc); +} + +/* + * Update Window Counter using the algorithm from [RFC 4342, 8.1]. + * As elsewhere, RTT > 0 is assumed by using dccp_sample_rtt(). + */ +static inline void ccid3_hc_tx_update_win_count(struct ccid3_hc_tx_sock *hc, + ktime_t now) +{ + u32 delta = ktime_us_delta(now, hc->tx_t_last_win_count), + quarter_rtts = (4 * delta) / hc->tx_rtt; + + if (quarter_rtts > 0) { + hc->tx_t_last_win_count = now; + hc->tx_last_win_count += min(quarter_rtts, 5U); + hc->tx_last_win_count &= 0xF; /* mod 16 */ + } +} + +static void ccid3_hc_tx_no_feedback_timer(struct timer_list *t) +{ + struct ccid3_hc_tx_sock *hc = from_timer(hc, t, tx_no_feedback_timer); + struct sock *sk = hc->sk; + unsigned long t_nfb = USEC_PER_SEC / 5; + + bh_lock_sock(sk); + if (sock_owned_by_user(sk)) { + /* Try again later. */ + /* XXX: set some sensible MIB */ + goto restart_timer; + } + + ccid3_pr_debug("%s(%p, state=%s) - entry\n", dccp_role(sk), sk, + ccid3_tx_state_name(hc->tx_state)); + + /* Ignore and do not restart after leaving the established state */ + if ((1 << sk->sk_state) & ~(DCCPF_OPEN | DCCPF_PARTOPEN)) + goto out; + + /* Reset feedback state to "no feedback received" */ + if (hc->tx_state == TFRC_SSTATE_FBACK) + ccid3_hc_tx_set_state(sk, TFRC_SSTATE_NO_FBACK); + + /* + * Determine new allowed sending rate X as per draft rfc3448bis-00, 4.4 + * RTO is 0 if and only if no feedback has been received yet. + */ + if (hc->tx_t_rto == 0 || hc->tx_p == 0) { + + /* halve send rate directly */ + hc->tx_x = max(hc->tx_x / 2, + (((__u64)hc->tx_s) << 6) / TFRC_T_MBI); + ccid3_update_send_interval(hc); + } else { + /* + * Modify the cached value of X_recv + * + * If (X_calc > 2 * X_recv) + * X_recv = max(X_recv / 2, s / (2 * t_mbi)); + * Else + * X_recv = X_calc / 4; + * + * Note that X_recv is scaled by 2^6 while X_calc is not + */ + if (hc->tx_x_calc > (hc->tx_x_recv >> 5)) + hc->tx_x_recv = + max(hc->tx_x_recv / 2, + (((__u64)hc->tx_s) << 6) / (2*TFRC_T_MBI)); + else { + hc->tx_x_recv = hc->tx_x_calc; + hc->tx_x_recv <<= 4; + } + ccid3_hc_tx_update_x(sk, NULL); + } + ccid3_pr_debug("Reduced X to %llu/64 bytes/sec\n", + (unsigned long long)hc->tx_x); + + /* + * Set new timeout for the nofeedback timer. + * See comments in packet_recv() regarding the value of t_RTO. + */ + if (unlikely(hc->tx_t_rto == 0)) /* no feedback received yet */ + t_nfb = TFRC_INITIAL_TIMEOUT; + else + t_nfb = max(hc->tx_t_rto, 2 * hc->tx_t_ipi); + +restart_timer: + sk_reset_timer(sk, &hc->tx_no_feedback_timer, + jiffies + usecs_to_jiffies(t_nfb)); +out: + bh_unlock_sock(sk); + sock_put(sk); +} + +/** + * ccid3_hc_tx_send_packet - Delay-based dequeueing of TX packets + * @sk: socket to send packet from + * @skb: next packet candidate to send on @sk + * + * This function uses the convention of ccid_packet_dequeue_eval() and + * returns a millisecond-delay value between 0 and t_mbi = 64000 msec. + */ +static int ccid3_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb) +{ + struct dccp_sock *dp = dccp_sk(sk); + struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + ktime_t now = ktime_get_real(); + s64 delay; + + /* + * This function is called only for Data and DataAck packets. Sending + * zero-sized Data(Ack)s is theoretically possible, but for congestion + * control this case is pathological - ignore it. + */ + if (unlikely(skb->len == 0)) + return -EBADMSG; + + if (hc->tx_state == TFRC_SSTATE_NO_SENT) { + sk_reset_timer(sk, &hc->tx_no_feedback_timer, (jiffies + + usecs_to_jiffies(TFRC_INITIAL_TIMEOUT))); + hc->tx_last_win_count = 0; + hc->tx_t_last_win_count = now; + + /* Set t_0 for initial packet */ + hc->tx_t_nom = now; + + hc->tx_s = skb->len; + + /* + * Use initial RTT sample when available: recommended by erratum + * to RFC 4342. This implements the initialisation procedure of + * draft rfc3448bis, section 4.2. Remember, X is scaled by 2^6. + */ + if (dp->dccps_syn_rtt) { + ccid3_pr_debug("SYN RTT = %uus\n", dp->dccps_syn_rtt); + hc->tx_rtt = dp->dccps_syn_rtt; + hc->tx_x = rfc3390_initial_rate(sk); + hc->tx_t_ld = now; + } else { + /* + * Sender does not have RTT sample: + * - set fallback RTT (RFC 4340, 3.4) since a RTT value + * is needed in several parts (e.g. window counter); + * - set sending rate X_pps = 1pps as per RFC 3448, 4.2. + */ + hc->tx_rtt = DCCP_FALLBACK_RTT; + hc->tx_x = hc->tx_s; + hc->tx_x <<= 6; + } + ccid3_update_send_interval(hc); + + ccid3_hc_tx_set_state(sk, TFRC_SSTATE_NO_FBACK); + + } else { + delay = ktime_us_delta(hc->tx_t_nom, now); + ccid3_pr_debug("delay=%ld\n", (long)delay); + /* + * Scheduling of packet transmissions (RFC 5348, 8.3) + * + * if (t_now > t_nom - delta) + * // send the packet now + * else + * // send the packet in (t_nom - t_now) milliseconds. + */ + if (delay >= TFRC_T_DELTA) + return (u32)delay / USEC_PER_MSEC; + + ccid3_hc_tx_update_win_count(hc, now); + } + + /* prepare to send now (add options etc.) */ + dp->dccps_hc_tx_insert_options = 1; + DCCP_SKB_CB(skb)->dccpd_ccval = hc->tx_last_win_count; + + /* set the nominal send time for the next following packet */ + hc->tx_t_nom = ktime_add_us(hc->tx_t_nom, hc->tx_t_ipi); + return CCID_PACKET_SEND_AT_ONCE; +} + +static void ccid3_hc_tx_packet_sent(struct sock *sk, unsigned int len) +{ + struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + + ccid3_hc_tx_update_s(hc, len); + + if (tfrc_tx_hist_add(&hc->tx_hist, dccp_sk(sk)->dccps_gss)) + DCCP_CRIT("packet history - out of memory!"); +} + +static void ccid3_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb) +{ + struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + struct tfrc_tx_hist_entry *acked; + ktime_t now; + unsigned long t_nfb; + u32 r_sample; + + /* we are only interested in ACKs */ + if (!(DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_ACK || + DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_DATAACK)) + return; + /* + * Locate the acknowledged packet in the TX history. + * + * Returning "entry not found" here can for instance happen when + * - the host has not sent out anything (e.g. a passive server), + * - the Ack is outdated (packet with higher Ack number was received), + * - it is a bogus Ack (for a packet not sent on this connection). + */ + acked = tfrc_tx_hist_find_entry(hc->tx_hist, dccp_hdr_ack_seq(skb)); + if (acked == NULL) + return; + /* For the sake of RTT sampling, ignore/remove all older entries */ + tfrc_tx_hist_purge(&acked->next); + + /* Update the moving average for the RTT estimate (RFC 3448, 4.3) */ + now = ktime_get_real(); + r_sample = dccp_sample_rtt(sk, ktime_us_delta(now, acked->stamp)); + hc->tx_rtt = tfrc_ewma(hc->tx_rtt, r_sample, 9); + + /* + * Update allowed sending rate X as per draft rfc3448bis-00, 4.2/3 + */ + if (hc->tx_state == TFRC_SSTATE_NO_FBACK) { + ccid3_hc_tx_set_state(sk, TFRC_SSTATE_FBACK); + + if (hc->tx_t_rto == 0) { + /* + * Initial feedback packet: Larger Initial Windows (4.2) + */ + hc->tx_x = rfc3390_initial_rate(sk); + hc->tx_t_ld = now; + + ccid3_update_send_interval(hc); + + goto done_computing_x; + } else if (hc->tx_p == 0) { + /* + * First feedback after nofeedback timer expiry (4.3) + */ + goto done_computing_x; + } + } + + /* Update sending rate (step 4 of [RFC 3448, 4.3]) */ + if (hc->tx_p > 0) + hc->tx_x_calc = tfrc_calc_x(hc->tx_s, hc->tx_rtt, hc->tx_p); + ccid3_hc_tx_update_x(sk, &now); + +done_computing_x: + ccid3_pr_debug("%s(%p), RTT=%uus (sample=%uus), s=%u, " + "p=%u, X_calc=%u, X_recv=%u, X=%u\n", + dccp_role(sk), sk, hc->tx_rtt, r_sample, + hc->tx_s, hc->tx_p, hc->tx_x_calc, + (unsigned int)(hc->tx_x_recv >> 6), + (unsigned int)(hc->tx_x >> 6)); + + /* unschedule no feedback timer */ + sk_stop_timer(sk, &hc->tx_no_feedback_timer); + + /* + * As we have calculated new ipi, delta, t_nom it is possible + * that we now can send a packet, so wake up dccp_wait_for_ccid + */ + sk->sk_write_space(sk); + + /* + * Update timeout interval for the nofeedback timer. In order to control + * rate halving on networks with very low RTTs (<= 1 ms), use per-route + * tunable RTAX_RTO_MIN value as the lower bound. + */ + hc->tx_t_rto = max_t(u32, 4 * hc->tx_rtt, + USEC_PER_SEC/HZ * tcp_rto_min(sk)); + /* + * Schedule no feedback timer to expire in + * max(t_RTO, 2 * s/X) = max(t_RTO, 2 * t_ipi) + */ + t_nfb = max(hc->tx_t_rto, 2 * hc->tx_t_ipi); + + ccid3_pr_debug("%s(%p), Scheduled no feedback timer to " + "expire in %lu jiffies (%luus)\n", + dccp_role(sk), sk, usecs_to_jiffies(t_nfb), t_nfb); + + sk_reset_timer(sk, &hc->tx_no_feedback_timer, + jiffies + usecs_to_jiffies(t_nfb)); +} + +static int ccid3_hc_tx_parse_options(struct sock *sk, u8 packet_type, + u8 option, u8 *optval, u8 optlen) +{ + struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + __be32 opt_val; + + switch (option) { + case TFRC_OPT_RECEIVE_RATE: + case TFRC_OPT_LOSS_EVENT_RATE: + /* Must be ignored on Data packets, cf. RFC 4342 8.3 and 8.5 */ + if (packet_type == DCCP_PKT_DATA) + break; + if (unlikely(optlen != 4)) { + DCCP_WARN("%s(%p), invalid len %d for %u\n", + dccp_role(sk), sk, optlen, option); + return -EINVAL; + } + opt_val = ntohl(get_unaligned((__be32 *)optval)); + + if (option == TFRC_OPT_RECEIVE_RATE) { + /* Receive Rate is kept in units of 64 bytes/second */ + hc->tx_x_recv = opt_val; + hc->tx_x_recv <<= 6; + + ccid3_pr_debug("%s(%p), RECEIVE_RATE=%u\n", + dccp_role(sk), sk, opt_val); + } else { + /* Update the fixpoint Loss Event Rate fraction */ + hc->tx_p = tfrc_invert_loss_event_rate(opt_val); + + ccid3_pr_debug("%s(%p), LOSS_EVENT_RATE=%u\n", + dccp_role(sk), sk, opt_val); + } + } + return 0; +} + +static int ccid3_hc_tx_init(struct ccid *ccid, struct sock *sk) +{ + struct ccid3_hc_tx_sock *hc = ccid_priv(ccid); + + hc->tx_state = TFRC_SSTATE_NO_SENT; + hc->tx_hist = NULL; + hc->sk = sk; + timer_setup(&hc->tx_no_feedback_timer, + ccid3_hc_tx_no_feedback_timer, 0); + return 0; +} + +static void ccid3_hc_tx_exit(struct sock *sk) +{ + struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + + sk_stop_timer(sk, &hc->tx_no_feedback_timer); + tfrc_tx_hist_purge(&hc->tx_hist); +} + +static void ccid3_hc_tx_get_info(struct sock *sk, struct tcp_info *info) +{ + info->tcpi_rto = ccid3_hc_tx_sk(sk)->tx_t_rto; + info->tcpi_rtt = ccid3_hc_tx_sk(sk)->tx_rtt; +} + +static int ccid3_hc_tx_getsockopt(struct sock *sk, const int optname, int len, + u32 __user *optval, int __user *optlen) +{ + const struct ccid3_hc_tx_sock *hc = ccid3_hc_tx_sk(sk); + struct tfrc_tx_info tfrc; + const void *val; + + switch (optname) { + case DCCP_SOCKOPT_CCID_TX_INFO: + if (len < sizeof(tfrc)) + return -EINVAL; + memset(&tfrc, 0, sizeof(tfrc)); + tfrc.tfrctx_x = hc->tx_x; + tfrc.tfrctx_x_recv = hc->tx_x_recv; + tfrc.tfrctx_x_calc = hc->tx_x_calc; + tfrc.tfrctx_rtt = hc->tx_rtt; + tfrc.tfrctx_p = hc->tx_p; + tfrc.tfrctx_rto = hc->tx_t_rto; + tfrc.tfrctx_ipi = hc->tx_t_ipi; + len = sizeof(tfrc); + val = &tfrc; + break; + default: + return -ENOPROTOOPT; + } + + if (put_user(len, optlen) || copy_to_user(optval, val, len)) + return -EFAULT; + + return 0; +} + +/* + * Receiver Half-Connection Routines + */ + +/* CCID3 feedback types */ +enum ccid3_fback_type { + CCID3_FBACK_NONE = 0, + CCID3_FBACK_INITIAL, + CCID3_FBACK_PERIODIC, + CCID3_FBACK_PARAM_CHANGE +}; + +#ifdef CONFIG_IP_DCCP_CCID3_DEBUG +static const char *ccid3_rx_state_name(enum ccid3_hc_rx_states state) +{ + static const char *const ccid3_rx_state_names[] = { + [TFRC_RSTATE_NO_DATA] = "NO_DATA", + [TFRC_RSTATE_DATA] = "DATA", + }; + + return ccid3_rx_state_names[state]; +} +#endif + +static void ccid3_hc_rx_set_state(struct sock *sk, + enum ccid3_hc_rx_states state) +{ + struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk); + enum ccid3_hc_rx_states oldstate = hc->rx_state; + + ccid3_pr_debug("%s(%p) %-8.8s -> %s\n", + dccp_role(sk), sk, ccid3_rx_state_name(oldstate), + ccid3_rx_state_name(state)); + WARN_ON(state == oldstate); + hc->rx_state = state; +} + +static void ccid3_hc_rx_send_feedback(struct sock *sk, + const struct sk_buff *skb, + enum ccid3_fback_type fbtype) +{ + struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk); + struct dccp_sock *dp = dccp_sk(sk); + ktime_t now = ktime_get(); + s64 delta = 0; + + switch (fbtype) { + case CCID3_FBACK_INITIAL: + hc->rx_x_recv = 0; + hc->rx_pinv = ~0U; /* see RFC 4342, 8.5 */ + break; + case CCID3_FBACK_PARAM_CHANGE: + /* + * When parameters change (new loss or p > p_prev), we do not + * have a reliable estimate for R_m of [RFC 3448, 6.2] and so + * need to reuse the previous value of X_recv. However, when + * X_recv was 0 (due to early loss), this would kill X down to + * s/t_mbi (i.e. one packet in 64 seconds). + * To avoid such drastic reduction, we approximate X_recv as + * the number of bytes since last feedback. + * This is a safe fallback, since X is bounded above by X_calc. + */ + if (hc->rx_x_recv > 0) + break; + fallthrough; + case CCID3_FBACK_PERIODIC: + delta = ktime_us_delta(now, hc->rx_tstamp_last_feedback); + if (delta <= 0) + delta = 1; + hc->rx_x_recv = scaled_div32(hc->rx_bytes_recv, delta); + break; + default: + return; + } + + ccid3_pr_debug("Interval %lldusec, X_recv=%u, 1/p=%u\n", delta, + hc->rx_x_recv, hc->rx_pinv); + + hc->rx_tstamp_last_feedback = now; + hc->rx_last_counter = dccp_hdr(skb)->dccph_ccval; + hc->rx_bytes_recv = 0; + + dp->dccps_hc_rx_insert_options = 1; + dccp_send_ack(sk); +} + +static int ccid3_hc_rx_insert_options(struct sock *sk, struct sk_buff *skb) +{ + const struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk); + __be32 x_recv, pinv; + + if (!(sk->sk_state == DCCP_OPEN || sk->sk_state == DCCP_PARTOPEN)) + return 0; + + if (dccp_packet_without_ack(skb)) + return 0; + + x_recv = htonl(hc->rx_x_recv); + pinv = htonl(hc->rx_pinv); + + if (dccp_insert_option(skb, TFRC_OPT_LOSS_EVENT_RATE, + &pinv, sizeof(pinv)) || + dccp_insert_option(skb, TFRC_OPT_RECEIVE_RATE, + &x_recv, sizeof(x_recv))) + return -1; + + return 0; +} + +/** + * ccid3_first_li - Implements [RFC 5348, 6.3.1] + * @sk: socket to calculate loss interval for + * + * Determine the length of the first loss interval via inverse lookup. + * Assume that X_recv can be computed by the throughput equation + * s + * X_recv = -------- + * R * fval + * Find some p such that f(p) = fval; return 1/p (scaled). + */ +static u32 ccid3_first_li(struct sock *sk) +{ + struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk); + u32 x_recv, p; + s64 delta; + u64 fval; + + if (hc->rx_rtt == 0) { + DCCP_WARN("No RTT estimate available, using fallback RTT\n"); + hc->rx_rtt = DCCP_FALLBACK_RTT; + } + + delta = ktime_us_delta(ktime_get(), hc->rx_tstamp_last_feedback); + if (delta <= 0) + delta = 1; + x_recv = scaled_div32(hc->rx_bytes_recv, delta); + if (x_recv == 0) { /* would also trigger divide-by-zero */ + DCCP_WARN("X_recv==0\n"); + if (hc->rx_x_recv == 0) { + DCCP_BUG("stored value of X_recv is zero"); + return ~0U; + } + x_recv = hc->rx_x_recv; + } + + fval = scaled_div(hc->rx_s, hc->rx_rtt); + fval = scaled_div32(fval, x_recv); + p = tfrc_calc_x_reverse_lookup(fval); + + ccid3_pr_debug("%s(%p), receive rate=%u bytes/s, implied " + "loss rate=%u\n", dccp_role(sk), sk, x_recv, p); + + return p == 0 ? ~0U : scaled_div(1, p); +} + +static void ccid3_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb) +{ + struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk); + enum ccid3_fback_type do_feedback = CCID3_FBACK_NONE; + const u64 ndp = dccp_sk(sk)->dccps_options_received.dccpor_ndp; + const bool is_data_packet = dccp_data_packet(skb); + + if (unlikely(hc->rx_state == TFRC_RSTATE_NO_DATA)) { + if (is_data_packet) { + const u32 payload = skb->len - dccp_hdr(skb)->dccph_doff * 4; + do_feedback = CCID3_FBACK_INITIAL; + ccid3_hc_rx_set_state(sk, TFRC_RSTATE_DATA); + hc->rx_s = payload; + /* + * Not necessary to update rx_bytes_recv here, + * since X_recv = 0 for the first feedback packet (cf. + * RFC 3448, 6.3) -- gerrit + */ + } + goto update_records; + } + + if (tfrc_rx_hist_duplicate(&hc->rx_hist, skb)) + return; /* done receiving */ + + if (is_data_packet) { + const u32 payload = skb->len - dccp_hdr(skb)->dccph_doff * 4; + /* + * Update moving-average of s and the sum of received payload bytes + */ + hc->rx_s = tfrc_ewma(hc->rx_s, payload, 9); + hc->rx_bytes_recv += payload; + } + + /* + * Perform loss detection and handle pending losses + */ + if (tfrc_rx_handle_loss(&hc->rx_hist, &hc->rx_li_hist, + skb, ndp, ccid3_first_li, sk)) { + do_feedback = CCID3_FBACK_PARAM_CHANGE; + goto done_receiving; + } + + if (tfrc_rx_hist_loss_pending(&hc->rx_hist)) + return; /* done receiving */ + + /* + * Handle data packets: RTT sampling and monitoring p + */ + if (unlikely(!is_data_packet)) + goto update_records; + + if (!tfrc_lh_is_initialised(&hc->rx_li_hist)) { + const u32 sample = tfrc_rx_hist_sample_rtt(&hc->rx_hist, skb); + /* + * Empty loss history: no loss so far, hence p stays 0. + * Sample RTT values, since an RTT estimate is required for the + * computation of p when the first loss occurs; RFC 3448, 6.3.1. + */ + if (sample != 0) + hc->rx_rtt = tfrc_ewma(hc->rx_rtt, sample, 9); + + } else if (tfrc_lh_update_i_mean(&hc->rx_li_hist, skb)) { + /* + * Step (3) of [RFC 3448, 6.1]: Recompute I_mean and, if I_mean + * has decreased (resp. p has increased), send feedback now. + */ + do_feedback = CCID3_FBACK_PARAM_CHANGE; + } + + /* + * Check if the periodic once-per-RTT feedback is due; RFC 4342, 10.3 + */ + if (SUB16(dccp_hdr(skb)->dccph_ccval, hc->rx_last_counter) > 3) + do_feedback = CCID3_FBACK_PERIODIC; + +update_records: + tfrc_rx_hist_add_packet(&hc->rx_hist, skb, ndp); + +done_receiving: + if (do_feedback) + ccid3_hc_rx_send_feedback(sk, skb, do_feedback); +} + +static int ccid3_hc_rx_init(struct ccid *ccid, struct sock *sk) +{ + struct ccid3_hc_rx_sock *hc = ccid_priv(ccid); + + hc->rx_state = TFRC_RSTATE_NO_DATA; + tfrc_lh_init(&hc->rx_li_hist); + return tfrc_rx_hist_alloc(&hc->rx_hist); +} + +static void ccid3_hc_rx_exit(struct sock *sk) +{ + struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk); + + tfrc_rx_hist_purge(&hc->rx_hist); + tfrc_lh_cleanup(&hc->rx_li_hist); +} + +static void ccid3_hc_rx_get_info(struct sock *sk, struct tcp_info *info) +{ + info->tcpi_ca_state = ccid3_hc_rx_sk(sk)->rx_state; + info->tcpi_options |= TCPI_OPT_TIMESTAMPS; + info->tcpi_rcv_rtt = ccid3_hc_rx_sk(sk)->rx_rtt; +} + +static int ccid3_hc_rx_getsockopt(struct sock *sk, const int optname, int len, + u32 __user *optval, int __user *optlen) +{ + const struct ccid3_hc_rx_sock *hc = ccid3_hc_rx_sk(sk); + struct tfrc_rx_info rx_info; + const void *val; + + switch (optname) { + case DCCP_SOCKOPT_CCID_RX_INFO: + if (len < sizeof(rx_info)) + return -EINVAL; + rx_info.tfrcrx_x_recv = hc->rx_x_recv; + rx_info.tfrcrx_rtt = hc->rx_rtt; + rx_info.tfrcrx_p = tfrc_invert_loss_event_rate(hc->rx_pinv); + len = sizeof(rx_info); + val = &rx_info; + break; + default: + return -ENOPROTOOPT; + } + + if (put_user(len, optlen) || copy_to_user(optval, val, len)) + return -EFAULT; + + return 0; +} + +struct ccid_operations ccid3_ops = { + .ccid_id = DCCPC_CCID3, + .ccid_name = "TCP-Friendly Rate Control", + .ccid_hc_tx_obj_size = sizeof(struct ccid3_hc_tx_sock), + .ccid_hc_tx_init = ccid3_hc_tx_init, + .ccid_hc_tx_exit = ccid3_hc_tx_exit, + .ccid_hc_tx_send_packet = ccid3_hc_tx_send_packet, + .ccid_hc_tx_packet_sent = ccid3_hc_tx_packet_sent, + .ccid_hc_tx_packet_recv = ccid3_hc_tx_packet_recv, + .ccid_hc_tx_parse_options = ccid3_hc_tx_parse_options, + .ccid_hc_rx_obj_size = sizeof(struct ccid3_hc_rx_sock), + .ccid_hc_rx_init = ccid3_hc_rx_init, + .ccid_hc_rx_exit = ccid3_hc_rx_exit, + .ccid_hc_rx_insert_options = ccid3_hc_rx_insert_options, + .ccid_hc_rx_packet_recv = ccid3_hc_rx_packet_recv, + .ccid_hc_rx_get_info = ccid3_hc_rx_get_info, + .ccid_hc_tx_get_info = ccid3_hc_tx_get_info, + .ccid_hc_rx_getsockopt = ccid3_hc_rx_getsockopt, + .ccid_hc_tx_getsockopt = ccid3_hc_tx_getsockopt, +}; + +#ifdef CONFIG_IP_DCCP_CCID3_DEBUG +module_param(ccid3_debug, bool, 0644); +MODULE_PARM_DESC(ccid3_debug, "Enable CCID-3 debug messages"); +#endif diff --git a/net/dccp/ccids/ccid3.h b/net/dccp/ccids/ccid3.h new file mode 100644 index 000000000..02e0fc9f6 --- /dev/null +++ b/net/dccp/ccids/ccid3.h @@ -0,0 +1,148 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * Copyright (c) 2005-7 The University of Waikato, Hamilton, New Zealand. + * Copyright (c) 2007 The University of Aberdeen, Scotland, UK + * + * An implementation of the DCCP protocol + * + * This code has been developed by the University of Waikato WAND + * research group. For further information please see https://www.wand.net.nz/ + * or e-mail Ian McDonald - ian.mcdonald@jandi.co.nz + * + * This code also uses code from Lulea University, rereleased as GPL by its + * authors: + * Copyright (c) 2003 Nils-Erik Mattsson, Joacim Haggmark, Magnus Erixzon + * + * Changes to meet Linux coding standards, to make it meet latest ccid3 draft + * and to make it work as a loadable module in the DCCP stack written by + * Arnaldo Carvalho de Melo <acme@conectiva.com.br>. + * + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + */ +#ifndef _DCCP_CCID3_H_ +#define _DCCP_CCID3_H_ + +#include <linux/ktime.h> +#include <linux/list.h> +#include <linux/types.h> +#include <linux/tfrc.h> +#include "lib/tfrc.h" +#include "../ccid.h" + +/* Two seconds as per RFC 5348, 4.2 */ +#define TFRC_INITIAL_TIMEOUT (2 * USEC_PER_SEC) + +/* Parameter t_mbi from [RFC 3448, 4.3]: backoff interval in seconds */ +#define TFRC_T_MBI 64 + +/* + * The t_delta parameter (RFC 5348, 8.3): delays of less than %USEC_PER_MSEC are + * rounded down to 0, since sk_reset_timer() here uses millisecond granularity. + * Hence we can use a constant t_delta = %USEC_PER_MSEC when HZ >= 500. A coarse + * resolution of HZ < 500 means that the error is below one timer tick (t_gran) + * when using the constant t_delta = t_gran / 2 = %USEC_PER_SEC / (2 * HZ). + */ +#if (HZ >= 500) +# define TFRC_T_DELTA USEC_PER_MSEC +#else +# define TFRC_T_DELTA (USEC_PER_SEC / (2 * HZ)) +#endif + +enum ccid3_options { + TFRC_OPT_LOSS_EVENT_RATE = 192, + TFRC_OPT_LOSS_INTERVALS = 193, + TFRC_OPT_RECEIVE_RATE = 194, +}; + +/* TFRC sender states */ +enum ccid3_hc_tx_states { + TFRC_SSTATE_NO_SENT = 1, + TFRC_SSTATE_NO_FBACK, + TFRC_SSTATE_FBACK, +}; + +/** + * struct ccid3_hc_tx_sock - CCID3 sender half-connection socket + * @tx_x: Current sending rate in 64 * bytes per second + * @tx_x_recv: Receive rate in 64 * bytes per second + * @tx_x_calc: Calculated rate in bytes per second + * @tx_rtt: Estimate of current round trip time in usecs + * @tx_p: Current loss event rate (0-1) scaled by 1000000 + * @tx_s: Packet size in bytes + * @tx_t_rto: Nofeedback Timer setting in usecs + * @tx_t_ipi: Interpacket (send) interval (RFC 3448, 4.6) in usecs + * @tx_state: Sender state, one of %ccid3_hc_tx_states + * @tx_last_win_count: Last window counter sent + * @tx_t_last_win_count: Timestamp of earliest packet + * with last_win_count value sent + * @tx_no_feedback_timer: Handle to no feedback timer + * @tx_t_ld: Time last doubled during slow start + * @tx_t_nom: Nominal send time of next packet + * @tx_hist: Packet history + */ +struct ccid3_hc_tx_sock { + u64 tx_x; + u64 tx_x_recv; + u32 tx_x_calc; + u32 tx_rtt; + u32 tx_p; + u32 tx_t_rto; + u32 tx_t_ipi; + u16 tx_s; + enum ccid3_hc_tx_states tx_state:8; + u8 tx_last_win_count; + ktime_t tx_t_last_win_count; + struct timer_list tx_no_feedback_timer; + struct sock *sk; + ktime_t tx_t_ld; + ktime_t tx_t_nom; + struct tfrc_tx_hist_entry *tx_hist; +}; + +static inline struct ccid3_hc_tx_sock *ccid3_hc_tx_sk(const struct sock *sk) +{ + struct ccid3_hc_tx_sock *hctx = ccid_priv(dccp_sk(sk)->dccps_hc_tx_ccid); + BUG_ON(hctx == NULL); + return hctx; +} + +/* TFRC receiver states */ +enum ccid3_hc_rx_states { + TFRC_RSTATE_NO_DATA = 1, + TFRC_RSTATE_DATA, +}; + +/** + * struct ccid3_hc_rx_sock - CCID3 receiver half-connection socket + * @rx_last_counter: Tracks window counter (RFC 4342, 8.1) + * @rx_state: Receiver state, one of %ccid3_hc_rx_states + * @rx_bytes_recv: Total sum of DCCP payload bytes + * @rx_x_recv: Receiver estimate of send rate (RFC 3448, sec. 4.3) + * @rx_rtt: Receiver estimate of RTT + * @rx_tstamp_last_feedback: Time at which last feedback was sent + * @rx_hist: Packet history (loss detection + RTT sampling) + * @rx_li_hist: Loss Interval database + * @rx_s: Received packet size in bytes + * @rx_pinv: Inverse of Loss Event Rate (RFC 4342, sec. 8.5) + */ +struct ccid3_hc_rx_sock { + u8 rx_last_counter:4; + enum ccid3_hc_rx_states rx_state:8; + u32 rx_bytes_recv; + u32 rx_x_recv; + u32 rx_rtt; + ktime_t rx_tstamp_last_feedback; + struct tfrc_rx_hist rx_hist; + struct tfrc_loss_hist rx_li_hist; + u16 rx_s; +#define rx_pinv rx_li_hist.i_mean +}; + +static inline struct ccid3_hc_rx_sock *ccid3_hc_rx_sk(const struct sock *sk) +{ + struct ccid3_hc_rx_sock *hcrx = ccid_priv(dccp_sk(sk)->dccps_hc_rx_ccid); + BUG_ON(hcrx == NULL); + return hcrx; +} + +#endif /* _DCCP_CCID3_H_ */ diff --git a/net/dccp/ccids/lib/loss_interval.c b/net/dccp/ccids/lib/loss_interval.c new file mode 100644 index 000000000..da9531984 --- /dev/null +++ b/net/dccp/ccids/lib/loss_interval.c @@ -0,0 +1,184 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * Copyright (c) 2007 The University of Aberdeen, Scotland, UK + * Copyright (c) 2005-7 The University of Waikato, Hamilton, New Zealand. + * Copyright (c) 2005-7 Ian McDonald <ian.mcdonald@jandi.co.nz> + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + */ +#include <net/sock.h> +#include "tfrc.h" + +static struct kmem_cache *tfrc_lh_slab __read_mostly; +/* Loss Interval weights from [RFC 3448, 5.4], scaled by 10 */ +static const int tfrc_lh_weights[NINTERVAL] = { 10, 10, 10, 10, 8, 6, 4, 2 }; + +/* implements LIFO semantics on the array */ +static inline u8 LIH_INDEX(const u8 ctr) +{ + return LIH_SIZE - 1 - (ctr % LIH_SIZE); +} + +/* the `counter' index always points at the next entry to be populated */ +static inline struct tfrc_loss_interval *tfrc_lh_peek(struct tfrc_loss_hist *lh) +{ + return lh->counter ? lh->ring[LIH_INDEX(lh->counter - 1)] : NULL; +} + +/* given i with 0 <= i <= k, return I_i as per the rfc3448bis notation */ +static inline u32 tfrc_lh_get_interval(struct tfrc_loss_hist *lh, const u8 i) +{ + BUG_ON(i >= lh->counter); + return lh->ring[LIH_INDEX(lh->counter - i - 1)]->li_length; +} + +/* + * On-demand allocation and de-allocation of entries + */ +static struct tfrc_loss_interval *tfrc_lh_demand_next(struct tfrc_loss_hist *lh) +{ + if (lh->ring[LIH_INDEX(lh->counter)] == NULL) + lh->ring[LIH_INDEX(lh->counter)] = kmem_cache_alloc(tfrc_lh_slab, + GFP_ATOMIC); + return lh->ring[LIH_INDEX(lh->counter)]; +} + +void tfrc_lh_cleanup(struct tfrc_loss_hist *lh) +{ + if (!tfrc_lh_is_initialised(lh)) + return; + + for (lh->counter = 0; lh->counter < LIH_SIZE; lh->counter++) + if (lh->ring[LIH_INDEX(lh->counter)] != NULL) { + kmem_cache_free(tfrc_lh_slab, + lh->ring[LIH_INDEX(lh->counter)]); + lh->ring[LIH_INDEX(lh->counter)] = NULL; + } +} + +static void tfrc_lh_calc_i_mean(struct tfrc_loss_hist *lh) +{ + u32 i_i, i_tot0 = 0, i_tot1 = 0, w_tot = 0; + int i, k = tfrc_lh_length(lh) - 1; /* k is as in rfc3448bis, 5.4 */ + + if (k <= 0) + return; + + for (i = 0; i <= k; i++) { + i_i = tfrc_lh_get_interval(lh, i); + + if (i < k) { + i_tot0 += i_i * tfrc_lh_weights[i]; + w_tot += tfrc_lh_weights[i]; + } + if (i > 0) + i_tot1 += i_i * tfrc_lh_weights[i-1]; + } + + lh->i_mean = max(i_tot0, i_tot1) / w_tot; +} + +/** + * tfrc_lh_update_i_mean - Update the `open' loss interval I_0 + * @lh: histogram to update + * @skb: received socket triggering loss interval update + * + * For recomputing p: returns `true' if p > p_prev <=> 1/p < 1/p_prev + */ +u8 tfrc_lh_update_i_mean(struct tfrc_loss_hist *lh, struct sk_buff *skb) +{ + struct tfrc_loss_interval *cur = tfrc_lh_peek(lh); + u32 old_i_mean = lh->i_mean; + s64 len; + + if (cur == NULL) /* not initialised */ + return 0; + + len = dccp_delta_seqno(cur->li_seqno, DCCP_SKB_CB(skb)->dccpd_seq) + 1; + + if (len - (s64)cur->li_length <= 0) /* duplicate or reordered */ + return 0; + + if (SUB16(dccp_hdr(skb)->dccph_ccval, cur->li_ccval) > 4) + /* + * Implements RFC 4342, 10.2: + * If a packet S (skb) exists whose seqno comes `after' the one + * starting the current loss interval (cur) and if the modulo-16 + * distance from C(cur) to C(S) is greater than 4, consider all + * subsequent packets as belonging to a new loss interval. This + * test is necessary since CCVal may wrap between intervals. + */ + cur->li_is_closed = 1; + + if (tfrc_lh_length(lh) == 1) /* due to RFC 3448, 6.3.1 */ + return 0; + + cur->li_length = len; + tfrc_lh_calc_i_mean(lh); + + return lh->i_mean < old_i_mean; +} + +/* Determine if `new_loss' does begin a new loss interval [RFC 4342, 10.2] */ +static inline u8 tfrc_lh_is_new_loss(struct tfrc_loss_interval *cur, + struct tfrc_rx_hist_entry *new_loss) +{ + return dccp_delta_seqno(cur->li_seqno, new_loss->tfrchrx_seqno) > 0 && + (cur->li_is_closed || SUB16(new_loss->tfrchrx_ccval, cur->li_ccval) > 4); +} + +/** + * tfrc_lh_interval_add - Insert new record into the Loss Interval database + * @lh: Loss Interval database + * @rh: Receive history containing a fresh loss event + * @calc_first_li: Caller-dependent routine to compute length of first interval + * @sk: Used by @calc_first_li in caller-specific way (subtyping) + * + * Updates I_mean and returns 1 if a new interval has in fact been added to @lh. + */ +int tfrc_lh_interval_add(struct tfrc_loss_hist *lh, struct tfrc_rx_hist *rh, + u32 (*calc_first_li)(struct sock *), struct sock *sk) +{ + struct tfrc_loss_interval *cur = tfrc_lh_peek(lh), *new; + + if (cur != NULL && !tfrc_lh_is_new_loss(cur, tfrc_rx_hist_loss_prev(rh))) + return 0; + + new = tfrc_lh_demand_next(lh); + if (unlikely(new == NULL)) { + DCCP_CRIT("Cannot allocate/add loss record."); + return 0; + } + + new->li_seqno = tfrc_rx_hist_loss_prev(rh)->tfrchrx_seqno; + new->li_ccval = tfrc_rx_hist_loss_prev(rh)->tfrchrx_ccval; + new->li_is_closed = 0; + + if (++lh->counter == 1) + lh->i_mean = new->li_length = (*calc_first_li)(sk); + else { + cur->li_length = dccp_delta_seqno(cur->li_seqno, new->li_seqno); + new->li_length = dccp_delta_seqno(new->li_seqno, + tfrc_rx_hist_last_rcv(rh)->tfrchrx_seqno) + 1; + if (lh->counter > (2*LIH_SIZE)) + lh->counter -= LIH_SIZE; + + tfrc_lh_calc_i_mean(lh); + } + return 1; +} + +int __init tfrc_li_init(void) +{ + tfrc_lh_slab = kmem_cache_create("tfrc_li_hist", + sizeof(struct tfrc_loss_interval), 0, + SLAB_HWCACHE_ALIGN, NULL); + return tfrc_lh_slab == NULL ? -ENOBUFS : 0; +} + +void tfrc_li_exit(void) +{ + if (tfrc_lh_slab != NULL) { + kmem_cache_destroy(tfrc_lh_slab); + tfrc_lh_slab = NULL; + } +} diff --git a/net/dccp/ccids/lib/loss_interval.h b/net/dccp/ccids/lib/loss_interval.h new file mode 100644 index 000000000..c3d95f85e --- /dev/null +++ b/net/dccp/ccids/lib/loss_interval.h @@ -0,0 +1,69 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +#ifndef _DCCP_LI_HIST_ +#define _DCCP_LI_HIST_ +/* + * Copyright (c) 2007 The University of Aberdeen, Scotland, UK + * Copyright (c) 2005-7 The University of Waikato, Hamilton, New Zealand. + * Copyright (c) 2005-7 Ian McDonald <ian.mcdonald@jandi.co.nz> + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + */ +#include <linux/ktime.h> +#include <linux/list.h> +#include <linux/slab.h> + +/* + * Number of loss intervals (RFC 4342, 8.6.1). The history size is one more than + * NINTERVAL, since the `open' interval I_0 is always stored as the first entry. + */ +#define NINTERVAL 8 +#define LIH_SIZE (NINTERVAL + 1) + +/** + * tfrc_loss_interval - Loss history record for TFRC-based protocols + * @li_seqno: Highest received seqno before the start of loss + * @li_ccval: The CCVal belonging to @li_seqno + * @li_is_closed: Whether @li_seqno is older than 1 RTT + * @li_length: Loss interval sequence length + */ +struct tfrc_loss_interval { + u64 li_seqno:48, + li_ccval:4, + li_is_closed:1; + u32 li_length; +}; + +/** + * tfrc_loss_hist - Loss record database + * @ring: Circular queue managed in LIFO manner + * @counter: Current count of entries (can be more than %LIH_SIZE) + * @i_mean: Current Average Loss Interval [RFC 3448, 5.4] + */ +struct tfrc_loss_hist { + struct tfrc_loss_interval *ring[LIH_SIZE]; + u8 counter; + u32 i_mean; +}; + +static inline void tfrc_lh_init(struct tfrc_loss_hist *lh) +{ + memset(lh, 0, sizeof(struct tfrc_loss_hist)); +} + +static inline u8 tfrc_lh_is_initialised(struct tfrc_loss_hist *lh) +{ + return lh->counter > 0; +} + +static inline u8 tfrc_lh_length(struct tfrc_loss_hist *lh) +{ + return min(lh->counter, (u8)LIH_SIZE); +} + +struct tfrc_rx_hist; + +int tfrc_lh_interval_add(struct tfrc_loss_hist *, struct tfrc_rx_hist *, + u32 (*first_li)(struct sock *), struct sock *); +u8 tfrc_lh_update_i_mean(struct tfrc_loss_hist *lh, struct sk_buff *); +void tfrc_lh_cleanup(struct tfrc_loss_hist *lh); + +#endif /* _DCCP_LI_HIST_ */ diff --git a/net/dccp/ccids/lib/packet_history.c b/net/dccp/ccids/lib/packet_history.c new file mode 100644 index 000000000..0cdda3c66 --- /dev/null +++ b/net/dccp/ccids/lib/packet_history.c @@ -0,0 +1,439 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * Copyright (c) 2007 The University of Aberdeen, Scotland, UK + * Copyright (c) 2005-7 The University of Waikato, Hamilton, New Zealand. + * + * An implementation of the DCCP protocol + * + * This code has been developed by the University of Waikato WAND + * research group. For further information please see https://www.wand.net.nz/ + * or e-mail Ian McDonald - ian.mcdonald@jandi.co.nz + * + * This code also uses code from Lulea University, rereleased as GPL by its + * authors: + * Copyright (c) 2003 Nils-Erik Mattsson, Joacim Haggmark, Magnus Erixzon + * + * Changes to meet Linux coding standards, to make it meet latest ccid3 draft + * and to make it work as a loadable module in the DCCP stack written by + * Arnaldo Carvalho de Melo <acme@conectiva.com.br>. + * + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + */ + +#include <linux/string.h> +#include <linux/slab.h> +#include "packet_history.h" +#include "../../dccp.h" + +/* + * Transmitter History Routines + */ +static struct kmem_cache *tfrc_tx_hist_slab; + +int __init tfrc_tx_packet_history_init(void) +{ + tfrc_tx_hist_slab = kmem_cache_create("tfrc_tx_hist", + sizeof(struct tfrc_tx_hist_entry), + 0, SLAB_HWCACHE_ALIGN, NULL); + return tfrc_tx_hist_slab == NULL ? -ENOBUFS : 0; +} + +void tfrc_tx_packet_history_exit(void) +{ + if (tfrc_tx_hist_slab != NULL) { + kmem_cache_destroy(tfrc_tx_hist_slab); + tfrc_tx_hist_slab = NULL; + } +} + +int tfrc_tx_hist_add(struct tfrc_tx_hist_entry **headp, u64 seqno) +{ + struct tfrc_tx_hist_entry *entry = kmem_cache_alloc(tfrc_tx_hist_slab, gfp_any()); + + if (entry == NULL) + return -ENOBUFS; + entry->seqno = seqno; + entry->stamp = ktime_get_real(); + entry->next = *headp; + *headp = entry; + return 0; +} + +void tfrc_tx_hist_purge(struct tfrc_tx_hist_entry **headp) +{ + struct tfrc_tx_hist_entry *head = *headp; + + while (head != NULL) { + struct tfrc_tx_hist_entry *next = head->next; + + kmem_cache_free(tfrc_tx_hist_slab, head); + head = next; + } + + *headp = NULL; +} + +/* + * Receiver History Routines + */ +static struct kmem_cache *tfrc_rx_hist_slab; + +int __init tfrc_rx_packet_history_init(void) +{ + tfrc_rx_hist_slab = kmem_cache_create("tfrc_rxh_cache", + sizeof(struct tfrc_rx_hist_entry), + 0, SLAB_HWCACHE_ALIGN, NULL); + return tfrc_rx_hist_slab == NULL ? -ENOBUFS : 0; +} + +void tfrc_rx_packet_history_exit(void) +{ + if (tfrc_rx_hist_slab != NULL) { + kmem_cache_destroy(tfrc_rx_hist_slab); + tfrc_rx_hist_slab = NULL; + } +} + +static inline void tfrc_rx_hist_entry_from_skb(struct tfrc_rx_hist_entry *entry, + const struct sk_buff *skb, + const u64 ndp) +{ + const struct dccp_hdr *dh = dccp_hdr(skb); + + entry->tfrchrx_seqno = DCCP_SKB_CB(skb)->dccpd_seq; + entry->tfrchrx_ccval = dh->dccph_ccval; + entry->tfrchrx_type = dh->dccph_type; + entry->tfrchrx_ndp = ndp; + entry->tfrchrx_tstamp = ktime_get_real(); +} + +void tfrc_rx_hist_add_packet(struct tfrc_rx_hist *h, + const struct sk_buff *skb, + const u64 ndp) +{ + struct tfrc_rx_hist_entry *entry = tfrc_rx_hist_last_rcv(h); + + tfrc_rx_hist_entry_from_skb(entry, skb, ndp); +} + +/* has the packet contained in skb been seen before? */ +int tfrc_rx_hist_duplicate(struct tfrc_rx_hist *h, struct sk_buff *skb) +{ + const u64 seq = DCCP_SKB_CB(skb)->dccpd_seq; + int i; + + if (dccp_delta_seqno(tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno, seq) <= 0) + return 1; + + for (i = 1; i <= h->loss_count; i++) + if (tfrc_rx_hist_entry(h, i)->tfrchrx_seqno == seq) + return 1; + + return 0; +} + +static void tfrc_rx_hist_swap(struct tfrc_rx_hist *h, const u8 a, const u8 b) +{ + const u8 idx_a = tfrc_rx_hist_index(h, a), + idx_b = tfrc_rx_hist_index(h, b); + + swap(h->ring[idx_a], h->ring[idx_b]); +} + +/* + * Private helper functions for loss detection. + * + * In the descriptions, `Si' refers to the sequence number of entry number i, + * whose NDP count is `Ni' (lower case is used for variables). + * Note: All __xxx_loss functions expect that a test against duplicates has been + * performed already: the seqno of the skb must not be less than the seqno + * of loss_prev; and it must not equal that of any valid history entry. + */ +static void __do_track_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u64 n1) +{ + u64 s0 = tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno, + s1 = DCCP_SKB_CB(skb)->dccpd_seq; + + if (!dccp_loss_free(s0, s1, n1)) { /* gap between S0 and S1 */ + h->loss_count = 1; + tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 1), skb, n1); + } +} + +static void __one_after_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u32 n2) +{ + u64 s0 = tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno, + s1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_seqno, + s2 = DCCP_SKB_CB(skb)->dccpd_seq; + + if (likely(dccp_delta_seqno(s1, s2) > 0)) { /* S1 < S2 */ + h->loss_count = 2; + tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 2), skb, n2); + return; + } + + /* S0 < S2 < S1 */ + + if (dccp_loss_free(s0, s2, n2)) { + u64 n1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_ndp; + + if (dccp_loss_free(s2, s1, n1)) { + /* hole is filled: S0, S2, and S1 are consecutive */ + h->loss_count = 0; + h->loss_start = tfrc_rx_hist_index(h, 1); + } else + /* gap between S2 and S1: just update loss_prev */ + tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_loss_prev(h), skb, n2); + + } else { /* gap between S0 and S2 */ + /* + * Reorder history to insert S2 between S0 and S1 + */ + tfrc_rx_hist_swap(h, 0, 3); + h->loss_start = tfrc_rx_hist_index(h, 3); + tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 1), skb, n2); + h->loss_count = 2; + } +} + +/* return 1 if a new loss event has been identified */ +static int __two_after_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u32 n3) +{ + u64 s0 = tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno, + s1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_seqno, + s2 = tfrc_rx_hist_entry(h, 2)->tfrchrx_seqno, + s3 = DCCP_SKB_CB(skb)->dccpd_seq; + + if (likely(dccp_delta_seqno(s2, s3) > 0)) { /* S2 < S3 */ + h->loss_count = 3; + tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 3), skb, n3); + return 1; + } + + /* S3 < S2 */ + + if (dccp_delta_seqno(s1, s3) > 0) { /* S1 < S3 < S2 */ + /* + * Reorder history to insert S3 between S1 and S2 + */ + tfrc_rx_hist_swap(h, 2, 3); + tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 2), skb, n3); + h->loss_count = 3; + return 1; + } + + /* S0 < S3 < S1 */ + + if (dccp_loss_free(s0, s3, n3)) { + u64 n1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_ndp; + + if (dccp_loss_free(s3, s1, n1)) { + /* hole between S0 and S1 filled by S3 */ + u64 n2 = tfrc_rx_hist_entry(h, 2)->tfrchrx_ndp; + + if (dccp_loss_free(s1, s2, n2)) { + /* entire hole filled by S0, S3, S1, S2 */ + h->loss_start = tfrc_rx_hist_index(h, 2); + h->loss_count = 0; + } else { + /* gap remains between S1 and S2 */ + h->loss_start = tfrc_rx_hist_index(h, 1); + h->loss_count = 1; + } + + } else /* gap exists between S3 and S1, loss_count stays at 2 */ + tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_loss_prev(h), skb, n3); + + return 0; + } + + /* + * The remaining case: S0 < S3 < S1 < S2; gap between S0 and S3 + * Reorder history to insert S3 between S0 and S1. + */ + tfrc_rx_hist_swap(h, 0, 3); + h->loss_start = tfrc_rx_hist_index(h, 3); + tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 1), skb, n3); + h->loss_count = 3; + + return 1; +} + +/* recycle RX history records to continue loss detection if necessary */ +static void __three_after_loss(struct tfrc_rx_hist *h) +{ + /* + * At this stage we know already that there is a gap between S0 and S1 + * (since S0 was the highest sequence number received before detecting + * the loss). To recycle the loss record, it is thus only necessary to + * check for other possible gaps between S1/S2 and between S2/S3. + */ + u64 s1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_seqno, + s2 = tfrc_rx_hist_entry(h, 2)->tfrchrx_seqno, + s3 = tfrc_rx_hist_entry(h, 3)->tfrchrx_seqno; + u64 n2 = tfrc_rx_hist_entry(h, 2)->tfrchrx_ndp, + n3 = tfrc_rx_hist_entry(h, 3)->tfrchrx_ndp; + + if (dccp_loss_free(s1, s2, n2)) { + + if (dccp_loss_free(s2, s3, n3)) { + /* no gap between S2 and S3: entire hole is filled */ + h->loss_start = tfrc_rx_hist_index(h, 3); + h->loss_count = 0; + } else { + /* gap between S2 and S3 */ + h->loss_start = tfrc_rx_hist_index(h, 2); + h->loss_count = 1; + } + + } else { /* gap between S1 and S2 */ + h->loss_start = tfrc_rx_hist_index(h, 1); + h->loss_count = 2; + } +} + +/** + * tfrc_rx_handle_loss - Loss detection and further processing + * @h: The non-empty RX history object + * @lh: Loss Intervals database to update + * @skb: Currently received packet + * @ndp: The NDP count belonging to @skb + * @calc_first_li: Caller-dependent computation of first loss interval in @lh + * @sk: Used by @calc_first_li (see tfrc_lh_interval_add) + * + * Chooses action according to pending loss, updates LI database when a new + * loss was detected, and does required post-processing. Returns 1 when caller + * should send feedback, 0 otherwise. + * Since it also takes care of reordering during loss detection and updates the + * records accordingly, the caller should not perform any more RX history + * operations when loss_count is greater than 0 after calling this function. + */ +int tfrc_rx_handle_loss(struct tfrc_rx_hist *h, + struct tfrc_loss_hist *lh, + struct sk_buff *skb, const u64 ndp, + u32 (*calc_first_li)(struct sock *), struct sock *sk) +{ + int is_new_loss = 0; + + if (h->loss_count == 0) { + __do_track_loss(h, skb, ndp); + } else if (h->loss_count == 1) { + __one_after_loss(h, skb, ndp); + } else if (h->loss_count != 2) { + DCCP_BUG("invalid loss_count %d", h->loss_count); + } else if (__two_after_loss(h, skb, ndp)) { + /* + * Update Loss Interval database and recycle RX records + */ + is_new_loss = tfrc_lh_interval_add(lh, h, calc_first_li, sk); + __three_after_loss(h); + } + return is_new_loss; +} + +int tfrc_rx_hist_alloc(struct tfrc_rx_hist *h) +{ + int i; + + for (i = 0; i <= TFRC_NDUPACK; i++) { + h->ring[i] = kmem_cache_alloc(tfrc_rx_hist_slab, GFP_ATOMIC); + if (h->ring[i] == NULL) + goto out_free; + } + + h->loss_count = h->loss_start = 0; + return 0; + +out_free: + while (i-- != 0) { + kmem_cache_free(tfrc_rx_hist_slab, h->ring[i]); + h->ring[i] = NULL; + } + return -ENOBUFS; +} + +void tfrc_rx_hist_purge(struct tfrc_rx_hist *h) +{ + int i; + + for (i = 0; i <= TFRC_NDUPACK; ++i) + if (h->ring[i] != NULL) { + kmem_cache_free(tfrc_rx_hist_slab, h->ring[i]); + h->ring[i] = NULL; + } +} + +/** + * tfrc_rx_hist_rtt_last_s - reference entry to compute RTT samples against + * @h: The non-empty RX history object + */ +static inline struct tfrc_rx_hist_entry * + tfrc_rx_hist_rtt_last_s(const struct tfrc_rx_hist *h) +{ + return h->ring[0]; +} + +/** + * tfrc_rx_hist_rtt_prev_s - previously suitable (wrt rtt_last_s) RTT-sampling entry + * @h: The non-empty RX history object + */ +static inline struct tfrc_rx_hist_entry * + tfrc_rx_hist_rtt_prev_s(const struct tfrc_rx_hist *h) +{ + return h->ring[h->rtt_sample_prev]; +} + +/** + * tfrc_rx_hist_sample_rtt - Sample RTT from timestamp / CCVal + * @h: receive histogram + * @skb: packet containing timestamp. + * + * Based on ideas presented in RFC 4342, 8.1. Returns 0 if it was not able + * to compute a sample with given data - calling function should check this. + */ +u32 tfrc_rx_hist_sample_rtt(struct tfrc_rx_hist *h, const struct sk_buff *skb) +{ + u32 sample = 0, + delta_v = SUB16(dccp_hdr(skb)->dccph_ccval, + tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval); + + if (delta_v < 1 || delta_v > 4) { /* unsuitable CCVal delta */ + if (h->rtt_sample_prev == 2) { /* previous candidate stored */ + sample = SUB16(tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_ccval, + tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval); + if (sample) + sample = 4 / sample * + ktime_us_delta(tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_tstamp, + tfrc_rx_hist_rtt_last_s(h)->tfrchrx_tstamp); + else /* + * FIXME: This condition is in principle not + * possible but occurs when CCID is used for + * two-way data traffic. I have tried to trace + * it, but the cause does not seem to be here. + */ + DCCP_BUG("please report to dccp@vger.kernel.org" + " => prev = %u, last = %u", + tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_ccval, + tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval); + } else if (delta_v < 1) { + h->rtt_sample_prev = 1; + goto keep_ref_for_next_time; + } + + } else if (delta_v == 4) /* optimal match */ + sample = ktime_to_us(net_timedelta(tfrc_rx_hist_rtt_last_s(h)->tfrchrx_tstamp)); + else { /* suboptimal match */ + h->rtt_sample_prev = 2; + goto keep_ref_for_next_time; + } + + if (unlikely(sample > DCCP_SANE_RTT_MAX)) { + DCCP_WARN("RTT sample %u too large, using max\n", sample); + sample = DCCP_SANE_RTT_MAX; + } + + h->rtt_sample_prev = 0; /* use current entry as next reference */ +keep_ref_for_next_time: + + return sample; +} diff --git a/net/dccp/ccids/lib/packet_history.h b/net/dccp/ccids/lib/packet_history.h new file mode 100644 index 000000000..159cc9326 --- /dev/null +++ b/net/dccp/ccids/lib/packet_history.h @@ -0,0 +1,142 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * Packet RX/TX history data structures and routines for TFRC-based protocols. + * + * Copyright (c) 2007 The University of Aberdeen, Scotland, UK + * Copyright (c) 2005-6 The University of Waikato, Hamilton, New Zealand. + * + * This code has been developed by the University of Waikato WAND + * research group. For further information please see https://www.wand.net.nz/ + * or e-mail Ian McDonald - ian.mcdonald@jandi.co.nz + * + * This code also uses code from Lulea University, rereleased as GPL by its + * authors: + * Copyright (c) 2003 Nils-Erik Mattsson, Joacim Haggmark, Magnus Erixzon + * + * Changes to meet Linux coding standards, to make it meet latest ccid3 draft + * and to make it work as a loadable module in the DCCP stack written by + * Arnaldo Carvalho de Melo <acme@conectiva.com.br>. + * + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + */ + +#ifndef _DCCP_PKT_HIST_ +#define _DCCP_PKT_HIST_ + +#include <linux/list.h> +#include <linux/slab.h> +#include "tfrc.h" + +/** + * tfrc_tx_hist_entry - Simple singly-linked TX history list + * @next: next oldest entry (LIFO order) + * @seqno: sequence number of this entry + * @stamp: send time of packet with sequence number @seqno + */ +struct tfrc_tx_hist_entry { + struct tfrc_tx_hist_entry *next; + u64 seqno; + ktime_t stamp; +}; + +static inline struct tfrc_tx_hist_entry * + tfrc_tx_hist_find_entry(struct tfrc_tx_hist_entry *head, u64 seqno) +{ + while (head != NULL && head->seqno != seqno) + head = head->next; + return head; +} + +int tfrc_tx_hist_add(struct tfrc_tx_hist_entry **headp, u64 seqno); +void tfrc_tx_hist_purge(struct tfrc_tx_hist_entry **headp); + +/* Subtraction a-b modulo-16, respects circular wrap-around */ +#define SUB16(a, b) (((a) + 16 - (b)) & 0xF) + +/* Number of packets to wait after a missing packet (RFC 4342, 6.1) */ +#define TFRC_NDUPACK 3 + +/** + * tfrc_rx_hist_entry - Store information about a single received packet + * @tfrchrx_seqno: DCCP packet sequence number + * @tfrchrx_ccval: window counter value of packet (RFC 4342, 8.1) + * @tfrchrx_ndp: the NDP count (if any) of the packet + * @tfrchrx_tstamp: actual receive time of packet + */ +struct tfrc_rx_hist_entry { + u64 tfrchrx_seqno:48, + tfrchrx_ccval:4, + tfrchrx_type:4; + u64 tfrchrx_ndp:48; + ktime_t tfrchrx_tstamp; +}; + +/** + * tfrc_rx_hist - RX history structure for TFRC-based protocols + * @ring: Packet history for RTT sampling and loss detection + * @loss_count: Number of entries in circular history + * @loss_start: Movable index (for loss detection) + * @rtt_sample_prev: Used during RTT sampling, points to candidate entry + */ +struct tfrc_rx_hist { + struct tfrc_rx_hist_entry *ring[TFRC_NDUPACK + 1]; + u8 loss_count:2, + loss_start:2; +#define rtt_sample_prev loss_start +}; + +/** + * tfrc_rx_hist_index - index to reach n-th entry after loss_start + */ +static inline u8 tfrc_rx_hist_index(const struct tfrc_rx_hist *h, const u8 n) +{ + return (h->loss_start + n) & TFRC_NDUPACK; +} + +/** + * tfrc_rx_hist_last_rcv - entry with highest-received-seqno so far + */ +static inline struct tfrc_rx_hist_entry * + tfrc_rx_hist_last_rcv(const struct tfrc_rx_hist *h) +{ + return h->ring[tfrc_rx_hist_index(h, h->loss_count)]; +} + +/** + * tfrc_rx_hist_entry - return the n-th history entry after loss_start + */ +static inline struct tfrc_rx_hist_entry * + tfrc_rx_hist_entry(const struct tfrc_rx_hist *h, const u8 n) +{ + return h->ring[tfrc_rx_hist_index(h, n)]; +} + +/** + * tfrc_rx_hist_loss_prev - entry with highest-received-seqno before loss was detected + */ +static inline struct tfrc_rx_hist_entry * + tfrc_rx_hist_loss_prev(const struct tfrc_rx_hist *h) +{ + return h->ring[h->loss_start]; +} + +/* indicate whether previously a packet was detected missing */ +static inline bool tfrc_rx_hist_loss_pending(const struct tfrc_rx_hist *h) +{ + return h->loss_count > 0; +} + +void tfrc_rx_hist_add_packet(struct tfrc_rx_hist *h, const struct sk_buff *skb, + const u64 ndp); + +int tfrc_rx_hist_duplicate(struct tfrc_rx_hist *h, struct sk_buff *skb); + +struct tfrc_loss_hist; +int tfrc_rx_handle_loss(struct tfrc_rx_hist *h, struct tfrc_loss_hist *lh, + struct sk_buff *skb, const u64 ndp, + u32 (*first_li)(struct sock *sk), struct sock *sk); +u32 tfrc_rx_hist_sample_rtt(struct tfrc_rx_hist *h, const struct sk_buff *skb); +int tfrc_rx_hist_alloc(struct tfrc_rx_hist *h); +void tfrc_rx_hist_purge(struct tfrc_rx_hist *h); + +#endif /* _DCCP_PKT_HIST_ */ diff --git a/net/dccp/ccids/lib/tfrc.c b/net/dccp/ccids/lib/tfrc.c new file mode 100644 index 000000000..d7f265e1f --- /dev/null +++ b/net/dccp/ccids/lib/tfrc.c @@ -0,0 +1,46 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * TFRC library initialisation + * + * Copyright (c) 2007 The University of Aberdeen, Scotland, UK + * Copyright (c) 2007 Arnaldo Carvalho de Melo <acme@redhat.com> + */ +#include <linux/moduleparam.h> +#include "tfrc.h" + +#ifdef CONFIG_IP_DCCP_TFRC_DEBUG +bool tfrc_debug; +module_param(tfrc_debug, bool, 0644); +MODULE_PARM_DESC(tfrc_debug, "Enable TFRC debug messages"); +#endif + +int __init tfrc_lib_init(void) +{ + int rc = tfrc_li_init(); + + if (rc) + goto out; + + rc = tfrc_tx_packet_history_init(); + if (rc) + goto out_free_loss_intervals; + + rc = tfrc_rx_packet_history_init(); + if (rc) + goto out_free_tx_history; + return 0; + +out_free_tx_history: + tfrc_tx_packet_history_exit(); +out_free_loss_intervals: + tfrc_li_exit(); +out: + return rc; +} + +void tfrc_lib_exit(void) +{ + tfrc_rx_packet_history_exit(); + tfrc_tx_packet_history_exit(); + tfrc_li_exit(); +} diff --git a/net/dccp/ccids/lib/tfrc.h b/net/dccp/ccids/lib/tfrc.h new file mode 100644 index 000000000..0a63e8750 --- /dev/null +++ b/net/dccp/ccids/lib/tfrc.h @@ -0,0 +1,73 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +#ifndef _TFRC_H_ +#define _TFRC_H_ +/* + * Copyright (c) 2007 The University of Aberdeen, Scotland, UK + * Copyright (c) 2005-6 The University of Waikato, Hamilton, New Zealand. + * Copyright (c) 2005-6 Ian McDonald <ian.mcdonald@jandi.co.nz> + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + * Copyright (c) 2003 Nils-Erik Mattsson, Joacim Haggmark, Magnus Erixzon + */ +#include <linux/types.h> +#include <linux/math64.h> +#include "../../dccp.h" + +/* internal includes that this library exports: */ +#include "loss_interval.h" +#include "packet_history.h" + +#ifdef CONFIG_IP_DCCP_TFRC_DEBUG +extern bool tfrc_debug; +#define tfrc_pr_debug(format, a...) DCCP_PR_DEBUG(tfrc_debug, format, ##a) +#else +#define tfrc_pr_debug(format, a...) +#endif + +/* integer-arithmetic divisions of type (a * 1000000)/b */ +static inline u64 scaled_div(u64 a, u64 b) +{ + BUG_ON(b == 0); + return div64_u64(a * 1000000, b); +} + +static inline u32 scaled_div32(u64 a, u64 b) +{ + u64 result = scaled_div(a, b); + + if (result > UINT_MAX) { + DCCP_CRIT("Overflow: %llu/%llu > UINT_MAX", + (unsigned long long)a, (unsigned long long)b); + return UINT_MAX; + } + return result; +} + +/** + * tfrc_ewma - Exponentially weighted moving average + * @weight: Weight to be used as damping factor, in units of 1/10 + */ +static inline u32 tfrc_ewma(const u32 avg, const u32 newval, const u8 weight) +{ + return avg ? (weight * avg + (10 - weight) * newval) / 10 : newval; +} + +u32 tfrc_calc_x(u16 s, u32 R, u32 p); +u32 tfrc_calc_x_reverse_lookup(u32 fvalue); +u32 tfrc_invert_loss_event_rate(u32 loss_event_rate); + +int tfrc_tx_packet_history_init(void); +void tfrc_tx_packet_history_exit(void); +int tfrc_rx_packet_history_init(void); +void tfrc_rx_packet_history_exit(void); + +int tfrc_li_init(void); +void tfrc_li_exit(void); + +#ifdef CONFIG_IP_DCCP_TFRC_LIB +int tfrc_lib_init(void); +void tfrc_lib_exit(void); +#else +#define tfrc_lib_init() (0) +#define tfrc_lib_exit() +#endif +#endif /* _TFRC_H_ */ diff --git a/net/dccp/ccids/lib/tfrc_equation.c b/net/dccp/ccids/lib/tfrc_equation.c new file mode 100644 index 000000000..92a8c6bea --- /dev/null +++ b/net/dccp/ccids/lib/tfrc_equation.c @@ -0,0 +1,702 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * Copyright (c) 2005 The University of Waikato, Hamilton, New Zealand. + * Copyright (c) 2005 Ian McDonald <ian.mcdonald@jandi.co.nz> + * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> + * Copyright (c) 2003 Nils-Erik Mattsson, Joacim Haggmark, Magnus Erixzon + */ + +#include <linux/module.h> +#include "../../dccp.h" +#include "tfrc.h" + +#define TFRC_CALC_X_ARRSIZE 500 +#define TFRC_CALC_X_SPLIT 50000 /* 0.05 * 1000000, details below */ +#define TFRC_SMALLEST_P (TFRC_CALC_X_SPLIT/TFRC_CALC_X_ARRSIZE) + +/* + TFRC TCP Reno Throughput Equation Lookup Table for f(p) + + The following two-column lookup table implements a part of the TCP throughput + equation from [RFC 3448, sec. 3.1]: + + s + X_calc = -------------------------------------------------------------- + R * sqrt(2*b*p/3) + (3 * t_RTO * sqrt(3*b*p/8) * (p + 32*p^3)) + + Where: + X is the transmit rate in bytes/second + s is the packet size in bytes + R is the round trip time in seconds + p is the loss event rate, between 0 and 1.0, of the number of loss + events as a fraction of the number of packets transmitted + t_RTO is the TCP retransmission timeout value in seconds + b is the number of packets acknowledged by a single TCP ACK + + We can assume that b = 1 and t_RTO is 4 * R. The equation now becomes: + + s + X_calc = ------------------------------------------------------- + R * sqrt(p*2/3) + (12 * R * sqrt(p*3/8) * (p + 32*p^3)) + + which we can break down into: + + s + X_calc = --------- + R * f(p) + + where f(p) is given for 0 < p <= 1 by: + + f(p) = sqrt(2*p/3) + 12 * sqrt(3*p/8) * (p + 32*p^3) + + Since this is kernel code, floating-point arithmetic is avoided in favour of + integer arithmetic. This means that nearly all fractional parameters are + scaled by 1000000: + * the parameters p and R + * the return result f(p) + The lookup table therefore actually tabulates the following function g(q): + + g(q) = 1000000 * f(q/1000000) + + Hence, when p <= 1, q must be less than or equal to 1000000. To achieve finer + granularity for the practically more relevant case of small values of p (up to + 5%), the second column is used; the first one ranges up to 100%. This split + corresponds to the value of q = TFRC_CALC_X_SPLIT. At the same time this also + determines the smallest resolution possible with this lookup table: + + TFRC_SMALLEST_P = TFRC_CALC_X_SPLIT / TFRC_CALC_X_ARRSIZE + + The entire table is generated by: + for(i=0; i < TFRC_CALC_X_ARRSIZE; i++) { + lookup[i][0] = g((i+1) * 1000000/TFRC_CALC_X_ARRSIZE); + lookup[i][1] = g((i+1) * TFRC_CALC_X_SPLIT/TFRC_CALC_X_ARRSIZE); + } + + With the given configuration, we have, with M = TFRC_CALC_X_ARRSIZE-1, + lookup[0][0] = g(1000000/(M+1)) = 1000000 * f(0.2%) + lookup[M][0] = g(1000000) = 1000000 * f(100%) + lookup[0][1] = g(TFRC_SMALLEST_P) = 1000000 * f(0.01%) + lookup[M][1] = g(TFRC_CALC_X_SPLIT) = 1000000 * f(5%) + + In summary, the two columns represent f(p) for the following ranges: + * The first column is for 0.002 <= p <= 1.0 + * The second column is for 0.0001 <= p <= 0.05 + Where the columns overlap, the second (finer-grained) is given preference, + i.e. the first column is used only for p >= 0.05. + */ +static const u32 tfrc_calc_x_lookup[TFRC_CALC_X_ARRSIZE][2] = { + { 37172, 8172 }, + { 53499, 11567 }, + { 66664, 14180 }, + { 78298, 16388 }, + { 89021, 18339 }, + { 99147, 20108 }, + { 108858, 21738 }, + { 118273, 23260 }, + { 127474, 24693 }, + { 136520, 26052 }, + { 145456, 27348 }, + { 154316, 28589 }, + { 163130, 29783 }, + { 171919, 30935 }, + { 180704, 32049 }, + { 189502, 33130 }, + { 198328, 34180 }, + { 207194, 35202 }, + { 216114, 36198 }, + { 225097, 37172 }, + { 234153, 38123 }, + { 243294, 39055 }, + { 252527, 39968 }, + { 261861, 40864 }, + { 271305, 41743 }, + { 280866, 42607 }, + { 290553, 43457 }, + { 300372, 44293 }, + { 310333, 45117 }, + { 320441, 45929 }, + { 330705, 46729 }, + { 341131, 47518 }, + { 351728, 48297 }, + { 362501, 49066 }, + { 373460, 49826 }, + { 384609, 50577 }, + { 395958, 51320 }, + { 407513, 52054 }, + { 419281, 52780 }, + { 431270, 53499 }, + { 443487, 54211 }, + { 455940, 54916 }, + { 468635, 55614 }, + { 481581, 56306 }, + { 494785, 56991 }, + { 508254, 57671 }, + { 521996, 58345 }, + { 536019, 59014 }, + { 550331, 59677 }, + { 564939, 60335 }, + { 579851, 60988 }, + { 595075, 61636 }, + { 610619, 62279 }, + { 626491, 62918 }, + { 642700, 63553 }, + { 659253, 64183 }, + { 676158, 64809 }, + { 693424, 65431 }, + { 711060, 66050 }, + { 729073, 66664 }, + { 747472, 67275 }, + { 766266, 67882 }, + { 785464, 68486 }, + { 805073, 69087 }, + { 825103, 69684 }, + { 845562, 70278 }, + { 866460, 70868 }, + { 887805, 71456 }, + { 909606, 72041 }, + { 931873, 72623 }, + { 954614, 73202 }, + { 977839, 73778 }, + { 1001557, 74352 }, + { 1025777, 74923 }, + { 1050508, 75492 }, + { 1075761, 76058 }, + { 1101544, 76621 }, + { 1127867, 77183 }, + { 1154739, 77741 }, + { 1182172, 78298 }, + { 1210173, 78852 }, + { 1238753, 79405 }, + { 1267922, 79955 }, + { 1297689, 80503 }, + { 1328066, 81049 }, + { 1359060, 81593 }, + { 1390684, 82135 }, + { 1422947, 82675 }, + { 1455859, 83213 }, + { 1489430, 83750 }, + { 1523671, 84284 }, + { 1558593, 84817 }, + { 1594205, 85348 }, + { 1630518, 85878 }, + { 1667543, 86406 }, + { 1705290, 86932 }, + { 1743770, 87457 }, + { 1782994, 87980 }, + { 1822973, 88501 }, + { 1863717, 89021 }, + { 1905237, 89540 }, + { 1947545, 90057 }, + { 1990650, 90573 }, + { 2034566, 91087 }, + { 2079301, 91600 }, + { 2124869, 92111 }, + { 2171279, 92622 }, + { 2218543, 93131 }, + { 2266673, 93639 }, + { 2315680, 94145 }, + { 2365575, 94650 }, + { 2416371, 95154 }, + { 2468077, 95657 }, + { 2520707, 96159 }, + { 2574271, 96660 }, + { 2628782, 97159 }, + { 2684250, 97658 }, + { 2740689, 98155 }, + { 2798110, 98651 }, + { 2856524, 99147 }, + { 2915944, 99641 }, + { 2976382, 100134 }, + { 3037850, 100626 }, + { 3100360, 101117 }, + { 3163924, 101608 }, + { 3228554, 102097 }, + { 3294263, 102586 }, + { 3361063, 103073 }, + { 3428966, 103560 }, + { 3497984, 104045 }, + { 3568131, 104530 }, + { 3639419, 105014 }, + { 3711860, 105498 }, + { 3785467, 105980 }, + { 3860253, 106462 }, + { 3936229, 106942 }, + { 4013410, 107422 }, + { 4091808, 107902 }, + { 4171435, 108380 }, + { 4252306, 108858 }, + { 4334431, 109335 }, + { 4417825, 109811 }, + { 4502501, 110287 }, + { 4588472, 110762 }, + { 4675750, 111236 }, + { 4764349, 111709 }, + { 4854283, 112182 }, + { 4945564, 112654 }, + { 5038206, 113126 }, + { 5132223, 113597 }, + { 5227627, 114067 }, + { 5324432, 114537 }, + { 5422652, 115006 }, + { 5522299, 115474 }, + { 5623389, 115942 }, + { 5725934, 116409 }, + { 5829948, 116876 }, + { 5935446, 117342 }, + { 6042439, 117808 }, + { 6150943, 118273 }, + { 6260972, 118738 }, + { 6372538, 119202 }, + { 6485657, 119665 }, + { 6600342, 120128 }, + { 6716607, 120591 }, + { 6834467, 121053 }, + { 6953935, 121514 }, + { 7075025, 121976 }, + { 7197752, 122436 }, + { 7322131, 122896 }, + { 7448175, 123356 }, + { 7575898, 123815 }, + { 7705316, 124274 }, + { 7836442, 124733 }, + { 7969291, 125191 }, + { 8103877, 125648 }, + { 8240216, 126105 }, + { 8378321, 126562 }, + { 8518208, 127018 }, + { 8659890, 127474 }, + { 8803384, 127930 }, + { 8948702, 128385 }, + { 9095861, 128840 }, + { 9244875, 129294 }, + { 9395760, 129748 }, + { 9548529, 130202 }, + { 9703198, 130655 }, + { 9859782, 131108 }, + { 10018296, 131561 }, + { 10178755, 132014 }, + { 10341174, 132466 }, + { 10505569, 132917 }, + { 10671954, 133369 }, + { 10840345, 133820 }, + { 11010757, 134271 }, + { 11183206, 134721 }, + { 11357706, 135171 }, + { 11534274, 135621 }, + { 11712924, 136071 }, + { 11893673, 136520 }, + { 12076536, 136969 }, + { 12261527, 137418 }, + { 12448664, 137867 }, + { 12637961, 138315 }, + { 12829435, 138763 }, + { 13023101, 139211 }, + { 13218974, 139658 }, + { 13417071, 140106 }, + { 13617407, 140553 }, + { 13819999, 140999 }, + { 14024862, 141446 }, + { 14232012, 141892 }, + { 14441465, 142339 }, + { 14653238, 142785 }, + { 14867346, 143230 }, + { 15083805, 143676 }, + { 15302632, 144121 }, + { 15523842, 144566 }, + { 15747453, 145011 }, + { 15973479, 145456 }, + { 16201939, 145900 }, + { 16432847, 146345 }, + { 16666221, 146789 }, + { 16902076, 147233 }, + { 17140429, 147677 }, + { 17381297, 148121 }, + { 17624696, 148564 }, + { 17870643, 149007 }, + { 18119154, 149451 }, + { 18370247, 149894 }, + { 18623936, 150336 }, + { 18880241, 150779 }, + { 19139176, 151222 }, + { 19400759, 151664 }, + { 19665007, 152107 }, + { 19931936, 152549 }, + { 20201564, 152991 }, + { 20473907, 153433 }, + { 20748982, 153875 }, + { 21026807, 154316 }, + { 21307399, 154758 }, + { 21590773, 155199 }, + { 21876949, 155641 }, + { 22165941, 156082 }, + { 22457769, 156523 }, + { 22752449, 156964 }, + { 23049999, 157405 }, + { 23350435, 157846 }, + { 23653774, 158287 }, + { 23960036, 158727 }, + { 24269236, 159168 }, + { 24581392, 159608 }, + { 24896521, 160049 }, + { 25214642, 160489 }, + { 25535772, 160929 }, + { 25859927, 161370 }, + { 26187127, 161810 }, + { 26517388, 162250 }, + { 26850728, 162690 }, + { 27187165, 163130 }, + { 27526716, 163569 }, + { 27869400, 164009 }, + { 28215234, 164449 }, + { 28564236, 164889 }, + { 28916423, 165328 }, + { 29271815, 165768 }, + { 29630428, 166208 }, + { 29992281, 166647 }, + { 30357392, 167087 }, + { 30725779, 167526 }, + { 31097459, 167965 }, + { 31472452, 168405 }, + { 31850774, 168844 }, + { 32232445, 169283 }, + { 32617482, 169723 }, + { 33005904, 170162 }, + { 33397730, 170601 }, + { 33792976, 171041 }, + { 34191663, 171480 }, + { 34593807, 171919 }, + { 34999428, 172358 }, + { 35408544, 172797 }, + { 35821174, 173237 }, + { 36237335, 173676 }, + { 36657047, 174115 }, + { 37080329, 174554 }, + { 37507197, 174993 }, + { 37937673, 175433 }, + { 38371773, 175872 }, + { 38809517, 176311 }, + { 39250924, 176750 }, + { 39696012, 177190 }, + { 40144800, 177629 }, + { 40597308, 178068 }, + { 41053553, 178507 }, + { 41513554, 178947 }, + { 41977332, 179386 }, + { 42444904, 179825 }, + { 42916290, 180265 }, + { 43391509, 180704 }, + { 43870579, 181144 }, + { 44353520, 181583 }, + { 44840352, 182023 }, + { 45331092, 182462 }, + { 45825761, 182902 }, + { 46324378, 183342 }, + { 46826961, 183781 }, + { 47333531, 184221 }, + { 47844106, 184661 }, + { 48358706, 185101 }, + { 48877350, 185541 }, + { 49400058, 185981 }, + { 49926849, 186421 }, + { 50457743, 186861 }, + { 50992759, 187301 }, + { 51531916, 187741 }, + { 52075235, 188181 }, + { 52622735, 188622 }, + { 53174435, 189062 }, + { 53730355, 189502 }, + { 54290515, 189943 }, + { 54854935, 190383 }, + { 55423634, 190824 }, + { 55996633, 191265 }, + { 56573950, 191706 }, + { 57155606, 192146 }, + { 57741621, 192587 }, + { 58332014, 193028 }, + { 58926806, 193470 }, + { 59526017, 193911 }, + { 60129666, 194352 }, + { 60737774, 194793 }, + { 61350361, 195235 }, + { 61967446, 195677 }, + { 62589050, 196118 }, + { 63215194, 196560 }, + { 63845897, 197002 }, + { 64481179, 197444 }, + { 65121061, 197886 }, + { 65765563, 198328 }, + { 66414705, 198770 }, + { 67068508, 199213 }, + { 67726992, 199655 }, + { 68390177, 200098 }, + { 69058085, 200540 }, + { 69730735, 200983 }, + { 70408147, 201426 }, + { 71090343, 201869 }, + { 71777343, 202312 }, + { 72469168, 202755 }, + { 73165837, 203199 }, + { 73867373, 203642 }, + { 74573795, 204086 }, + { 75285124, 204529 }, + { 76001380, 204973 }, + { 76722586, 205417 }, + { 77448761, 205861 }, + { 78179926, 206306 }, + { 78916102, 206750 }, + { 79657310, 207194 }, + { 80403571, 207639 }, + { 81154906, 208084 }, + { 81911335, 208529 }, + { 82672880, 208974 }, + { 83439562, 209419 }, + { 84211402, 209864 }, + { 84988421, 210309 }, + { 85770640, 210755 }, + { 86558080, 211201 }, + { 87350762, 211647 }, + { 88148708, 212093 }, + { 88951938, 212539 }, + { 89760475, 212985 }, + { 90574339, 213432 }, + { 91393551, 213878 }, + { 92218133, 214325 }, + { 93048107, 214772 }, + { 93883493, 215219 }, + { 94724314, 215666 }, + { 95570590, 216114 }, + { 96422343, 216561 }, + { 97279594, 217009 }, + { 98142366, 217457 }, + { 99010679, 217905 }, + { 99884556, 218353 }, + { 100764018, 218801 }, + { 101649086, 219250 }, + { 102539782, 219698 }, + { 103436128, 220147 }, + { 104338146, 220596 }, + { 105245857, 221046 }, + { 106159284, 221495 }, + { 107078448, 221945 }, + { 108003370, 222394 }, + { 108934074, 222844 }, + { 109870580, 223294 }, + { 110812910, 223745 }, + { 111761087, 224195 }, + { 112715133, 224646 }, + { 113675069, 225097 }, + { 114640918, 225548 }, + { 115612702, 225999 }, + { 116590442, 226450 }, + { 117574162, 226902 }, + { 118563882, 227353 }, + { 119559626, 227805 }, + { 120561415, 228258 }, + { 121569272, 228710 }, + { 122583219, 229162 }, + { 123603278, 229615 }, + { 124629471, 230068 }, + { 125661822, 230521 }, + { 126700352, 230974 }, + { 127745083, 231428 }, + { 128796039, 231882 }, + { 129853241, 232336 }, + { 130916713, 232790 }, + { 131986475, 233244 }, + { 133062553, 233699 }, + { 134144966, 234153 }, + { 135233739, 234608 }, + { 136328894, 235064 }, + { 137430453, 235519 }, + { 138538440, 235975 }, + { 139652876, 236430 }, + { 140773786, 236886 }, + { 141901190, 237343 }, + { 143035113, 237799 }, + { 144175576, 238256 }, + { 145322604, 238713 }, + { 146476218, 239170 }, + { 147636442, 239627 }, + { 148803298, 240085 }, + { 149976809, 240542 }, + { 151156999, 241000 }, + { 152343890, 241459 }, + { 153537506, 241917 }, + { 154737869, 242376 }, + { 155945002, 242835 }, + { 157158929, 243294 }, + { 158379673, 243753 }, + { 159607257, 244213 }, + { 160841704, 244673 }, + { 162083037, 245133 }, + { 163331279, 245593 }, + { 164586455, 246054 }, + { 165848586, 246514 }, + { 167117696, 246975 }, + { 168393810, 247437 }, + { 169676949, 247898 }, + { 170967138, 248360 }, + { 172264399, 248822 }, + { 173568757, 249284 }, + { 174880235, 249747 }, + { 176198856, 250209 }, + { 177524643, 250672 }, + { 178857621, 251136 }, + { 180197813, 251599 }, + { 181545242, 252063 }, + { 182899933, 252527 }, + { 184261908, 252991 }, + { 185631191, 253456 }, + { 187007807, 253920 }, + { 188391778, 254385 }, + { 189783129, 254851 }, + { 191181884, 255316 }, + { 192588065, 255782 }, + { 194001698, 256248 }, + { 195422805, 256714 }, + { 196851411, 257181 }, + { 198287540, 257648 }, + { 199731215, 258115 }, + { 201182461, 258582 }, + { 202641302, 259050 }, + { 204107760, 259518 }, + { 205581862, 259986 }, + { 207063630, 260454 }, + { 208553088, 260923 }, + { 210050262, 261392 }, + { 211555174, 261861 }, + { 213067849, 262331 }, + { 214588312, 262800 }, + { 216116586, 263270 }, + { 217652696, 263741 }, + { 219196666, 264211 }, + { 220748520, 264682 }, + { 222308282, 265153 }, + { 223875978, 265625 }, + { 225451630, 266097 }, + { 227035265, 266569 }, + { 228626905, 267041 }, + { 230226576, 267514 }, + { 231834302, 267986 }, + { 233450107, 268460 }, + { 235074016, 268933 }, + { 236706054, 269407 }, + { 238346244, 269881 }, + { 239994613, 270355 }, + { 241651183, 270830 }, + { 243315981, 271305 } +}; + +/* return largest index i such that fval <= lookup[i][small] */ +static inline u32 tfrc_binsearch(u32 fval, u8 small) +{ + u32 try, low = 0, high = TFRC_CALC_X_ARRSIZE - 1; + + while (low < high) { + try = (low + high) / 2; + if (fval <= tfrc_calc_x_lookup[try][small]) + high = try; + else + low = try + 1; + } + return high; +} + +/** + * tfrc_calc_x - Calculate the send rate as per section 3.1 of RFC3448 + * @s: packet size in bytes + * @R: RTT scaled by 1000000 (i.e., microseconds) + * @p: loss ratio estimate scaled by 1000000 + * + * Returns X_calc in bytes per second (not scaled). + */ +u32 tfrc_calc_x(u16 s, u32 R, u32 p) +{ + u16 index; + u32 f; + u64 result; + + /* check against invalid parameters and divide-by-zero */ + BUG_ON(p > 1000000); /* p must not exceed 100% */ + BUG_ON(p == 0); /* f(0) = 0, divide by zero */ + if (R == 0) { /* possible divide by zero */ + DCCP_CRIT("WARNING: RTT is 0, returning maximum X_calc."); + return ~0U; + } + + if (p <= TFRC_CALC_X_SPLIT) { /* 0.0000 < p <= 0.05 */ + if (p < TFRC_SMALLEST_P) { /* 0.0000 < p < 0.0001 */ + DCCP_WARN("Value of p (%d) below resolution. " + "Substituting %d\n", p, TFRC_SMALLEST_P); + index = 0; + } else /* 0.0001 <= p <= 0.05 */ + index = p/TFRC_SMALLEST_P - 1; + + f = tfrc_calc_x_lookup[index][1]; + + } else { /* 0.05 < p <= 1.00 */ + index = p/(1000000/TFRC_CALC_X_ARRSIZE) - 1; + + f = tfrc_calc_x_lookup[index][0]; + } + + /* + * Compute X = s/(R*f(p)) in bytes per second. + * Since f(p) and R are both scaled by 1000000, we need to multiply by + * 1000000^2. To avoid overflow, the result is computed in two stages. + * This works under almost all reasonable operational conditions, for a + * wide range of parameters. Yet, should some strange combination of + * parameters result in overflow, the use of scaled_div32 will catch + * this and return UINT_MAX - which is a logically adequate consequence. + */ + result = scaled_div(s, R); + return scaled_div32(result, f); +} + +/** + * tfrc_calc_x_reverse_lookup - try to find p given f(p) + * @fvalue: function value to match, scaled by 1000000 + * + * Returns closest match for p, also scaled by 1000000 + */ +u32 tfrc_calc_x_reverse_lookup(u32 fvalue) +{ + int index; + + if (fvalue == 0) /* f(p) = 0 whenever p = 0 */ + return 0; + + /* Error cases. */ + if (fvalue < tfrc_calc_x_lookup[0][1]) { + DCCP_WARN("fvalue %u smaller than resolution\n", fvalue); + return TFRC_SMALLEST_P; + } + if (fvalue > tfrc_calc_x_lookup[TFRC_CALC_X_ARRSIZE - 1][0]) { + DCCP_WARN("fvalue %u exceeds bounds!\n", fvalue); + return 1000000; + } + + if (fvalue <= tfrc_calc_x_lookup[TFRC_CALC_X_ARRSIZE - 1][1]) { + index = tfrc_binsearch(fvalue, 1); + return (index + 1) * TFRC_CALC_X_SPLIT / TFRC_CALC_X_ARRSIZE; + } + + /* else ... it must be in the coarse-grained column */ + index = tfrc_binsearch(fvalue, 0); + return (index + 1) * 1000000 / TFRC_CALC_X_ARRSIZE; +} + +/** + * tfrc_invert_loss_event_rate - Compute p so that 10^6 corresponds to 100% + * @loss_event_rate: loss event rate to invert + * When @loss_event_rate is large, there is a chance that p is truncated to 0. + * To avoid re-entering slow-start in that case, we set p = TFRC_SMALLEST_P > 0. + */ +u32 tfrc_invert_loss_event_rate(u32 loss_event_rate) +{ + if (loss_event_rate == UINT_MAX) /* see RFC 4342, 8.5 */ + return 0; + if (unlikely(loss_event_rate == 0)) /* map 1/0 into 100% */ + return 1000000; + return max_t(u32, scaled_div(1, loss_event_rate), TFRC_SMALLEST_P); +} |