kernel-fxtec-pro1x/net/dccp/ccids/ccid3.c

786 lines
23 KiB
C
Raw Normal View History

/*
* net/dccp/ccids/ccid3.c
*
* 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 http://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>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
#include "../dccp.h"
#include "ccid3.h"
#include <asm/unaligned.h>
#ifdef CONFIG_IP_DCCP_CCID3_DEBUG
static int 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
*/
/*
* Compute the initial sending rate X_init in the manner of RFC 3390:
*
* X_init = min(4 * MPS, max(2 * MPS, 4380 bytes)) / RTT
*
* 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 u32 mps = dccp_sk(sk)->dccps_mss_cache,
w_init = clamp(4380U, 2 * mps, 4 * mps);
return scaled_div(w_init << 6, ccid3_hc_tx_sk(sk)->rtt);
}
dccp ccid-3: Bug fix for the inter-packet scheduling algorithm This fixes a subtle bug in the calculation of the inter-packet gap and shows that t_delta, as it is currently used, is not needed. And hence replaced. The algorithm from RFC 3448, 4.6 below continually computes a send time t_nom, which is initialised with the current time t_now; t_gran = 1E6 / HZ specifies the scheduling granularity, s the packet size, and X the sending rate: t_distance = t_nom - t_now; // in microseconds t_delta = min(t_ipi, t_gran) / 2; // `delta' parameter in microseconds if (t_distance >= t_delta) { reschedule after (t_distance / 1000) milliseconds; } else { t_ipi = s / X; // inter-packet interval in usec t_nom += t_ipi; // compute the next send time send packet now; } 1) Description of the bug ------------------------- Rescheduling requires a conversion into milliseconds, due to this call chain: * ccid3_hc_tx_send_packet() returns a timeout in milliseconds, * this value is converted by msecs_to_jiffies() in dccp_write_xmit(), * and finally used as jiffy-expires-value for sk_reset_timer(). The highest jiffy resolution with HZ=1000 is 1 millisecond, so using a higher granularity does not make much sense here. As a consequence, values of t_distance < 1000 are truncated to 0. This issue has so far been resolved by using instead if (t_distance >= t_delta + 1000) reschedule after (t_distance / 1000) milliseconds; The bug is in artificially inflating t_delta to t_delta' = t_delta + 1000. This is unnecessarily large, a more adequate value is t_delta' = max(t_delta, 1000). 2) Consequences of using the corrected t_delta' ----------------------------------------------- Since t_delta <= t_gran/2 = 10^6/(2*HZ), we have t_delta <= 1000 as long as HZ >= 500. This means that t_delta' = max(1000, t_delta) is constant at 1000. On the other hand, when using a coarse HZ value of HZ < 500, we have three sub-cases that can all be reduced to using another constant of t_gran/2. (a) The first case arises when t_ipi > t_gran. Here t_delta' is the constant t_delta' = max(1000, t_gran/2) = t_gran/2. (b) If t_ipi <= 2000 < t_gran = 10^6/HZ usec, then t_delta = t_ipi/2 <= 1000, so that t_delta' = max(1000, t_delta) = 1000 < t_gran/2. (c) If 2000 < t_ipi <= t_gran, we have t_delta' = max(t_delta, 1000) = t_ipi/2. In the second and third cases we have delay values less than t_gran/2, which is in the order of less than or equal to half a jiffy. How these are treated depends on how fractions of a jiffy are handled: they are either always rounded down to 0, or always rounded up to 1 jiffy (assuming non-zero values). In both cases the error is on average in the order of 50%. Thus we are not increasing the error when in the second/third case we replace a value less than t_gran/2 with 0, by setting t_delta' to the constant t_gran/2. 3) Summary ---------- Fixing (1) and considering (2), the patch replaces t_delta with a constant, whose value depends on CONFIG_HZ, changing the above algorithm to: if (t_distance >= t_delta') reschedule after (t_distance / 1000) milliseconds; where t_delta' = 10^6/(2*HZ) if HZ < 500, and t_delta' = 1000 otherwise. Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
2008-09-03 23:30:19 -06:00
/**
* ccid3_update_send_interval - Calculate new t_ipi = s / X_inst
* This respects the granularity of X_inst (64 * bytes/second).
*/
static void ccid3_update_send_interval(struct ccid3_hc_tx_sock *hctx)
{
hctx->t_ipi = scaled_div32(((u64)hctx->s) << 6, hctx->x);
dccp ccid-3: Bug fix for the inter-packet scheduling algorithm This fixes a subtle bug in the calculation of the inter-packet gap and shows that t_delta, as it is currently used, is not needed. And hence replaced. The algorithm from RFC 3448, 4.6 below continually computes a send time t_nom, which is initialised with the current time t_now; t_gran = 1E6 / HZ specifies the scheduling granularity, s the packet size, and X the sending rate: t_distance = t_nom - t_now; // in microseconds t_delta = min(t_ipi, t_gran) / 2; // `delta' parameter in microseconds if (t_distance >= t_delta) { reschedule after (t_distance / 1000) milliseconds; } else { t_ipi = s / X; // inter-packet interval in usec t_nom += t_ipi; // compute the next send time send packet now; } 1) Description of the bug ------------------------- Rescheduling requires a conversion into milliseconds, due to this call chain: * ccid3_hc_tx_send_packet() returns a timeout in milliseconds, * this value is converted by msecs_to_jiffies() in dccp_write_xmit(), * and finally used as jiffy-expires-value for sk_reset_timer(). The highest jiffy resolution with HZ=1000 is 1 millisecond, so using a higher granularity does not make much sense here. As a consequence, values of t_distance < 1000 are truncated to 0. This issue has so far been resolved by using instead if (t_distance >= t_delta + 1000) reschedule after (t_distance / 1000) milliseconds; The bug is in artificially inflating t_delta to t_delta' = t_delta + 1000. This is unnecessarily large, a more adequate value is t_delta' = max(t_delta, 1000). 2) Consequences of using the corrected t_delta' ----------------------------------------------- Since t_delta <= t_gran/2 = 10^6/(2*HZ), we have t_delta <= 1000 as long as HZ >= 500. This means that t_delta' = max(1000, t_delta) is constant at 1000. On the other hand, when using a coarse HZ value of HZ < 500, we have three sub-cases that can all be reduced to using another constant of t_gran/2. (a) The first case arises when t_ipi > t_gran. Here t_delta' is the constant t_delta' = max(1000, t_gran/2) = t_gran/2. (b) If t_ipi <= 2000 < t_gran = 10^6/HZ usec, then t_delta = t_ipi/2 <= 1000, so that t_delta' = max(1000, t_delta) = 1000 < t_gran/2. (c) If 2000 < t_ipi <= t_gran, we have t_delta' = max(t_delta, 1000) = t_ipi/2. In the second and third cases we have delay values less than t_gran/2, which is in the order of less than or equal to half a jiffy. How these are treated depends on how fractions of a jiffy are handled: they are either always rounded down to 0, or always rounded up to 1 jiffy (assuming non-zero values). In both cases the error is on average in the order of 50%. Thus we are not increasing the error when in the second/third case we replace a value less than t_gran/2 with 0, by setting t_delta' to the constant t_gran/2. 3) Summary ---------- Fixing (1) and considering (2), the patch replaces t_delta with a constant, whose value depends on CONFIG_HZ, changing the above algorithm to: if (t_distance >= t_delta') reschedule after (t_distance / 1000) milliseconds; where t_delta' = 10^6/(2*HZ) if HZ < 500, and t_delta' = 1000 otherwise. Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
2008-09-03 23:30:19 -06:00
ccid3_pr_debug("t_ipi=%u, s=%u, X=%u\n", hctx->t_ipi,
hctx->s, (unsigned)(hctx->x >> 6));
}
static u32 ccid3_hc_tx_idle_rtt(struct ccid3_hc_tx_sock *hctx, ktime_t now)
{
u32 delta = ktime_us_delta(now, hctx->t_last_win_count);
return delta / hctx->rtt;
}
/**
* ccid3_hc_tx_update_x - Update allowed sending rate X
* @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 *hctx = ccid3_hc_tx_sk(sk);
u64 min_rate = 2 * hctx->x_recv;
const u64 old_x = hctx->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(hctx, now) >= 2) {
min_rate = rfc3390_initial_rate(sk);
min_rate = max(min_rate, 2 * hctx->x_recv);
}
if (hctx->p > 0) {
hctx->x = min(((u64)hctx->x_calc) << 6, min_rate);
hctx->x = max(hctx->x, (((u64)hctx->s) << 6) / TFRC_T_MBI);
} else if (ktime_us_delta(now, hctx->t_ld) - (s64)hctx->rtt >= 0) {
hctx->x = min(2 * hctx->x, min_rate);
hctx->x = max(hctx->x,
scaled_div(((u64)hctx->s) << 6, hctx->rtt));
hctx->t_ld = now;
}
if (hctx->x != old_x) {
ccid3_pr_debug("X_prev=%u, X_now=%u, X_calc=%u, "
"X_recv=%u\n", (unsigned)(old_x >> 6),
(unsigned)(hctx->x >> 6), hctx->x_calc,
(unsigned)(hctx->x_recv >> 6));
ccid3_update_send_interval(hctx);
}
}
/*
* Track the mean packet size `s' (cf. RFC 4342, 5.3 and RFC 3448, 4.1)
* @len: DCCP packet payload size in bytes
*/
static inline void ccid3_hc_tx_update_s(struct ccid3_hc_tx_sock *hctx, int len)
{
const u16 old_s = hctx->s;
hctx->s = tfrc_ewma(hctx->s, len, 9);
if (hctx->s != old_s)
ccid3_update_send_interval(hctx);
}
/*
* 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 *hctx,
ktime_t now)
{
u32 delta = ktime_us_delta(now, hctx->t_last_win_count),
quarter_rtts = (4 * delta) / hctx->rtt;
if (quarter_rtts > 0) {
hctx->t_last_win_count = now;
hctx->last_win_count += min(quarter_rtts, 5U);
hctx->last_win_count &= 0xF; /* mod 16 */
}
}
static void ccid3_hc_tx_no_feedback_timer(unsigned long data)
{
struct sock *sk = (struct sock *)data;
struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(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) entry with%s feedback\n", dccp_role(sk), sk,
hctx->feedback ? "" : "out");
/* 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" */
hctx->feedback = false;
/*
* 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 (hctx->t_rto == 0 || hctx->p == 0) {
/* halve send rate directly */
hctx->x = max(hctx->x / 2, (((u64)hctx->s) << 6) / TFRC_T_MBI);
ccid3_update_send_interval(hctx);
} 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
*/
BUG_ON(hctx->p && !hctx->x_calc);
if (hctx->x_calc > (hctx->x_recv >> 5))
hctx->x_recv =
max(hctx->x_recv / 2,
(((__u64)hctx->s) << 6) / (2 * TFRC_T_MBI));
else {
hctx->x_recv = hctx->x_calc;
hctx->x_recv <<= 4;
}
ccid3_hc_tx_update_x(sk, NULL);
}
ccid3_pr_debug("Reduced X to %llu/64 bytes/sec\n",
(unsigned long long)hctx->x);
/*
* Set new timeout for the nofeedback timer.
* See comments in packet_recv() regarding the value of t_RTO.
*/
if (unlikely(hctx->t_rto == 0)) /* no feedback received yet */
t_nfb = TFRC_INITIAL_TIMEOUT;
else
t_nfb = max(hctx->t_rto, 2 * hctx->t_ipi);
restart_timer:
sk_reset_timer(sk, &hctx->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
* @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 *hctx = 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 (hctx->s == 0) {
sk_reset_timer(sk, &hctx->no_feedback_timer, (jiffies +
usecs_to_jiffies(TFRC_INITIAL_TIMEOUT)));
hctx->last_win_count = 0;
hctx->t_last_win_count = now;
/* Set t_0 for initial packet */
hctx->t_nom = now;
hctx->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);
hctx->rtt = dp->dccps_syn_rtt;
hctx->x = rfc3390_initial_rate(sk);
hctx->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.
*/
hctx->rtt = DCCP_FALLBACK_RTT;
hctx->x = dp->dccps_mss_cache;
hctx->x <<= 6;
}
ccid3_update_send_interval(hctx);
} else {
delay = ktime_us_delta(hctx->t_nom, now);
ccid3_pr_debug("delay=%ld\n", (long)delay);
/*
* Scheduling of packet transmissions [RFC 3448, 4.6]
*
* if (t_now > t_nom - delta)
* // send the packet now
* else
* // send the packet in (t_nom - t_now) milliseconds.
*/
dccp ccid-3: Bug fix for the inter-packet scheduling algorithm This fixes a subtle bug in the calculation of the inter-packet gap and shows that t_delta, as it is currently used, is not needed. And hence replaced. The algorithm from RFC 3448, 4.6 below continually computes a send time t_nom, which is initialised with the current time t_now; t_gran = 1E6 / HZ specifies the scheduling granularity, s the packet size, and X the sending rate: t_distance = t_nom - t_now; // in microseconds t_delta = min(t_ipi, t_gran) / 2; // `delta' parameter in microseconds if (t_distance >= t_delta) { reschedule after (t_distance / 1000) milliseconds; } else { t_ipi = s / X; // inter-packet interval in usec t_nom += t_ipi; // compute the next send time send packet now; } 1) Description of the bug ------------------------- Rescheduling requires a conversion into milliseconds, due to this call chain: * ccid3_hc_tx_send_packet() returns a timeout in milliseconds, * this value is converted by msecs_to_jiffies() in dccp_write_xmit(), * and finally used as jiffy-expires-value for sk_reset_timer(). The highest jiffy resolution with HZ=1000 is 1 millisecond, so using a higher granularity does not make much sense here. As a consequence, values of t_distance < 1000 are truncated to 0. This issue has so far been resolved by using instead if (t_distance >= t_delta + 1000) reschedule after (t_distance / 1000) milliseconds; The bug is in artificially inflating t_delta to t_delta' = t_delta + 1000. This is unnecessarily large, a more adequate value is t_delta' = max(t_delta, 1000). 2) Consequences of using the corrected t_delta' ----------------------------------------------- Since t_delta <= t_gran/2 = 10^6/(2*HZ), we have t_delta <= 1000 as long as HZ >= 500. This means that t_delta' = max(1000, t_delta) is constant at 1000. On the other hand, when using a coarse HZ value of HZ < 500, we have three sub-cases that can all be reduced to using another constant of t_gran/2. (a) The first case arises when t_ipi > t_gran. Here t_delta' is the constant t_delta' = max(1000, t_gran/2) = t_gran/2. (b) If t_ipi <= 2000 < t_gran = 10^6/HZ usec, then t_delta = t_ipi/2 <= 1000, so that t_delta' = max(1000, t_delta) = 1000 < t_gran/2. (c) If 2000 < t_ipi <= t_gran, we have t_delta' = max(t_delta, 1000) = t_ipi/2. In the second and third cases we have delay values less than t_gran/2, which is in the order of less than or equal to half a jiffy. How these are treated depends on how fractions of a jiffy are handled: they are either always rounded down to 0, or always rounded up to 1 jiffy (assuming non-zero values). In both cases the error is on average in the order of 50%. Thus we are not increasing the error when in the second/third case we replace a value less than t_gran/2 with 0, by setting t_delta' to the constant t_gran/2. 3) Summary ---------- Fixing (1) and considering (2), the patch replaces t_delta with a constant, whose value depends on CONFIG_HZ, changing the above algorithm to: if (t_distance >= t_delta') reschedule after (t_distance / 1000) milliseconds; where t_delta' = 10^6/(2*HZ) if HZ < 500, and t_delta' = 1000 otherwise. Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
2008-09-03 23:30:19 -06:00
if (delay >= TFRC_T_DELTA)
return (u32)delay / USEC_PER_MSEC;
ccid3_hc_tx_update_win_count(hctx, now);
}
/* prepare to send now (add options etc.) */
dp->dccps_hc_tx_insert_options = 1;
DCCP_SKB_CB(skb)->dccpd_ccval = hctx->last_win_count;
/* set the nominal send time for the next following packet */
hctx->t_nom = ktime_add_us(hctx->t_nom, hctx->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 *hctx = ccid3_hc_tx_sk(sk);
ccid3_hc_tx_update_s(hctx, len);
if (tfrc_tx_hist_add(&hctx->hist, dccp_sk(sk)->dccps_gss))
[DCCP] ccid3: Perform history operations only after packet has been sent This migrates all packet history operations into the routine ccid3_hc_tx_packet_sent, thereby removing synchronization problems that occur when, as before, the operations are spread over multiple routines. The following minor simplifications are also applied: * several simplifications now follow from this change - several tests are now no longer required * removal of one unnecessary variable (dp) Justification: Currently packet history operations span two different routines, one of which is likely to pass through several iterations of sleeping and awakening. The first routine, ccid3_hc_tx_send_packet, allocates an entry and sets a few fields. The remaining fields are filled in when the second routine (which is not within a sleeping context), ccid3_hc_tx_packet_sent, is called. This has several strong drawbacks: * it is not necessary to split history operations - all fields can be filled in by the second routine * the first routine is called multiple times, until a packet can be sent, and sleeps meanwhile - this causes a lot of difficulties with regard to keeping the list consistent * since both routines do not have a producer-consumer like synchronization, it is very difficult to maintain data across calls to these routines * the fact that the routines are called in different contexts (sleeping, not sleeping) adds further problems Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
2006-12-09 19:09:21 -07:00
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 *hctx = 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(hctx->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));
hctx->rtt = tfrc_ewma(hctx->rtt, r_sample, 9);
/*
* Update allowed sending rate X as per draft rfc3448bis-00, 4.2/3
*/
if (!hctx->feedback) {
hctx->feedback = true;
if (hctx->t_rto == 0) {
/*
* Initial feedback packet: Larger Initial Windows (4.2)
*/
hctx->x = rfc3390_initial_rate(sk);
hctx->t_ld = now;
ccid3_update_send_interval(hctx);
goto done_computing_x;
} else if (hctx->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 (hctx->p > 0)
hctx->x_calc = tfrc_calc_x(hctx->s, hctx->rtt, hctx->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, hctx->rtt, r_sample,
hctx->s, hctx->p, hctx->x_calc,
(unsigned)(hctx->x_recv >> 6),
(unsigned)(hctx->x >> 6));
/* unschedule no feedback timer */
sk_stop_timer(sk, &hctx->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.
* We use a configuration option to increase the lower bound.
* This can help avoid triggering the nofeedback timer too
* often ('spinning') on LANs with small RTTs.
*/
hctx->t_rto = max_t(u32, 4 * hctx->rtt, (CONFIG_IP_DCCP_CCID3_RTO *
(USEC_PER_SEC / 1000)));
/*
* Schedule no feedback timer to expire in
* max(t_RTO, 2 * s/X) = max(t_RTO, 2 * t_ipi)
*/
t_nfb = max(hctx->t_rto, 2 * hctx->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, &hctx->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 *hctx = 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 */
hctx->x_recv = opt_val;
hctx->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 */
hctx->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 *hctx = ccid_priv(ccid);
hctx->hist = NULL;
setup_timer(&hctx->no_feedback_timer,
ccid3_hc_tx_no_feedback_timer, (unsigned long)sk);
return 0;
}
static void ccid3_hc_tx_exit(struct sock *sk)
{
struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
sk_stop_timer(sk, &hctx->no_feedback_timer);
tfrc_tx_hist_purge(&hctx->hist);
}
static void ccid3_hc_tx_get_info(struct sock *sk, struct tcp_info *info)
{
info->tcpi_rto = ccid3_hc_tx_sk(sk)->t_rto;
info->tcpi_rtt = ccid3_hc_tx_sk(sk)->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 *hctx = 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;
tfrc.tfrctx_x = hctx->x;
tfrc.tfrctx_x_recv = hctx->x_recv;
tfrc.tfrctx_x_calc = hctx->x_calc;
tfrc.tfrctx_rtt = hctx->rtt;
tfrc.tfrctx_p = hctx->p;
tfrc.tfrctx_rto = hctx->t_rto;
tfrc.tfrctx_ipi = hctx->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
*/
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 *hcrx = ccid3_hc_rx_sk(sk);
switch (fbtype) {
case CCID3_FBACK_INITIAL:
hcrx->x_recv = 0;
hcrx->p_inverse = ~0U; /* see RFC 4342, 8.5 */
break;
case CCID3_FBACK_PARAM_CHANGE:
if (unlikely(hcrx->feedback == CCID3_FBACK_NONE)) {
/*
* rfc3448bis-06, 6.3.1: First packet(s) lost or marked
* FIXME: in rfc3448bis the receiver returns X_recv=0
* here as it normally would in the first feedback packet.
* However this is not possible yet, since the code still
* uses RFC 3448, i.e.
* If (p > 0)
* Calculate X_calc using the TCP throughput equation.
* X = max(min(X_calc, 2*X_recv), s/t_mbi);
* would bring X down to s/t_mbi. That is why we return
* X_recv according to rfc3448bis-06 for the moment.
*/
u32 s = tfrc_rx_hist_packet_size(&hcrx->hist),
rtt = tfrc_rx_hist_rtt(&hcrx->hist);
hcrx->x_recv = scaled_div32(s, 2 * rtt);
break;
}
/*
* 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
* always check whether at least RTT time units were covered.
*/
hcrx->x_recv = tfrc_rx_hist_x_recv(&hcrx->hist, hcrx->x_recv);
break;
case CCID3_FBACK_PERIODIC:
/*
* Step (2) of rfc3448bis-06, 6.2:
* - if no data packets have been received, just restart timer
* - if data packets have been received, re-compute X_recv
*/
if (hcrx->hist.bytes_recvd == 0)
goto prepare_for_next_time;
hcrx->x_recv = tfrc_rx_hist_x_recv(&hcrx->hist, hcrx->x_recv);
break;
default:
return;
}
ccid3_pr_debug("X_recv=%u, 1/p=%u\n", hcrx->x_recv, hcrx->p_inverse);
dccp_sk(sk)->dccps_hc_rx_insert_options = 1;
dccp_send_ack(sk);
prepare_for_next_time:
tfrc_rx_hist_restart_byte_counter(&hcrx->hist);
hcrx->last_counter = dccp_hdr(skb)->dccph_ccval;
hcrx->feedback = fbtype;
}
static int ccid3_hc_rx_insert_options(struct sock *sk, struct sk_buff *skb)
{
const struct ccid3_hc_rx_sock *hcrx = 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(hcrx->x_recv);
pinv = htonl(hcrx->p_inverse);
if (dccp_insert_option(sk, skb, TFRC_OPT_LOSS_EVENT_RATE,
&pinv, sizeof(pinv)) ||
dccp_insert_option(sk, skb, TFRC_OPT_RECEIVE_RATE,
&x_recv, sizeof(x_recv)))
return -1;
return 0;
}
/** ccid3_first_li - Implements [RFC 3448, 6.3.1]
*
* 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 *hcrx = ccid3_hc_rx_sk(sk);
u32 s = tfrc_rx_hist_packet_size(&hcrx->hist),
rtt = tfrc_rx_hist_rtt(&hcrx->hist), x_recv, p;
u64 fval;
/*
* rfc3448bis-06, 6.3.1: First data packet(s) are marked or lost. Set p
* to give the equivalent of X_target = s/(2*R). Thus fval = 2 and so p
* is about 20.64%. This yields an interval length of 4.84 (rounded up).
*/
if (unlikely(hcrx->feedback == CCID3_FBACK_NONE))
return 5;
x_recv = tfrc_rx_hist_x_recv(&hcrx->hist, hcrx->x_recv);
if (x_recv == 0)
goto failed;
fval = scaled_div32(scaled_div(s, rtt), 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);
if (p > 0)
return scaled_div(1, p);
failed:
return UINT_MAX;
}
static void ccid3_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
{
struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
const u64 ndp = dccp_sk(sk)->dccps_options_received.dccpor_ndp;
const bool is_data_packet = dccp_data_packet(skb);
/*
* Perform loss detection and handle pending losses
*/
if (tfrc_rx_congestion_event(&hcrx->hist, &hcrx->li_hist,
skb, ndp, ccid3_first_li, sk))
ccid3_hc_rx_send_feedback(sk, skb, CCID3_FBACK_PARAM_CHANGE);
/*
* Feedback for first non-empty data packet (RFC 3448, 6.3)
*/
else if (unlikely(hcrx->feedback == CCID3_FBACK_NONE && is_data_packet))
ccid3_hc_rx_send_feedback(sk, skb, CCID3_FBACK_INITIAL);
/*
* Check if the periodic once-per-RTT feedback is due; RFC 4342, 10.3
*/
else if (!tfrc_rx_hist_loss_pending(&hcrx->hist) && is_data_packet &&
SUB16(dccp_hdr(skb)->dccph_ccval, hcrx->last_counter) > 3)
ccid3_hc_rx_send_feedback(sk, skb, CCID3_FBACK_PERIODIC);
}
static int ccid3_hc_rx_init(struct ccid *ccid, struct sock *sk)
{
struct ccid3_hc_rx_sock *hcrx = ccid_priv(ccid);
tfrc_lh_init(&hcrx->li_hist);
return tfrc_rx_hist_init(&hcrx->hist, sk);
}
static void ccid3_hc_rx_exit(struct sock *sk)
{
struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
tfrc_rx_hist_purge(&hcrx->hist);
tfrc_lh_cleanup(&hcrx->li_hist);
}
static void ccid3_hc_rx_get_info(struct sock *sk, struct tcp_info *info)
{
info->tcpi_options |= TCPI_OPT_TIMESTAMPS;
info->tcpi_rcv_rtt = tfrc_rx_hist_rtt(&ccid3_hc_rx_sk(sk)->hist);
}
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 *hcrx = 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 = hcrx->x_recv;
rx_info.tfrcrx_rtt = tfrc_rx_hist_rtt(&hcrx->hist);
rx_info.tfrcrx_p = tfrc_invert_loss_event_rate(hcrx->p_inverse);
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;
}
static struct ccid_operations ccid3 = {
.ccid_id = DCCPC_CCID3,
.ccid_name = "TCP-Friendly Rate Control",
.ccid_owner = THIS_MODULE,
.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 debug messages");
#endif
static __init int ccid3_module_init(void)
{
struct timespec tp;
/*
* Without a fine-grained clock resolution, RTTs/X_recv are not sampled
* correctly and feedback is sent either too early or too late.
*/
hrtimer_get_res(CLOCK_MONOTONIC, &tp);
if (tp.tv_sec || tp.tv_nsec > DCCP_TIME_RESOLUTION * NSEC_PER_USEC) {
printk(KERN_ERR "%s: Timer too coarse (%ld usec), need %u-usec"
" resolution - check your clocksource.\n", __func__,
tp.tv_nsec/NSEC_PER_USEC, DCCP_TIME_RESOLUTION);
return -ESOCKTNOSUPPORT;
}
return ccid_register(&ccid3);
}
module_init(ccid3_module_init);
static __exit void ccid3_module_exit(void)
{
ccid_unregister(&ccid3);
}
module_exit(ccid3_module_exit);
MODULE_AUTHOR("Ian McDonald <ian.mcdonald@jandi.co.nz>, "
"Arnaldo Carvalho de Melo <acme@ghostprotocols.net>");
MODULE_DESCRIPTION("DCCP TFRC CCID3 CCID");
MODULE_LICENSE("GPL");
MODULE_ALIAS("net-dccp-ccid-3");