1
linux/net/ipv4/tcp_cdg.c

429 lines
11 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
/*
* CAIA Delay-Gradient (CDG) congestion control
*
* This implementation is based on the paper:
* D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using
* delay gradients." In IFIP Networking, pages 328-341. Springer, 2011.
*
* Scavenger traffic (Less-than-Best-Effort) should disable coexistence
* heuristics using parameters use_shadow=0 and use_ineff=0.
*
* Parameters window, backoff_beta, and backoff_factor are crucial for
* throughput and delay. Future work is needed to determine better defaults,
* and to provide guidelines for use in different environments/contexts.
*
* Except for window, knobs are configured via /sys/module/tcp_cdg/parameters/.
* Parameter window is only configurable when loading tcp_cdg as a module.
*
* Notable differences from paper/FreeBSD:
* o Using Hybrid Slow start and Proportional Rate Reduction.
* o Add toggle for shadow window mechanism. Suggested by David Hayes.
* o Add toggle for non-congestion loss tolerance.
* o Scaling parameter G is changed to a backoff factor;
* conversion is given by: backoff_factor = 1000/(G * window).
* o Limit shadow window to 2 * cwnd, or to cwnd when application limited.
* o More accurate e^-x.
*/
#include <linux/kernel.h>
#include <linux/random.h>
#include <linux/module.h>
#include <linux/sched/clock.h>
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
#include <net/tcp.h>
#define HYSTART_ACK_TRAIN 1
#define HYSTART_DELAY 2
static int window __read_mostly = 8;
static unsigned int backoff_beta __read_mostly = 0.7071 * 1024; /* sqrt 0.5 */
static unsigned int backoff_factor __read_mostly = 42;
static unsigned int hystart_detect __read_mostly = 3;
static unsigned int use_ineff __read_mostly = 5;
static bool use_shadow __read_mostly = true;
static bool use_tolerance __read_mostly;
module_param(window, int, 0444);
MODULE_PARM_DESC(window, "gradient window size (power of two <= 256)");
module_param(backoff_beta, uint, 0644);
MODULE_PARM_DESC(backoff_beta, "backoff beta (0-1024)");
module_param(backoff_factor, uint, 0644);
MODULE_PARM_DESC(backoff_factor, "backoff probability scale factor");
module_param(hystart_detect, uint, 0644);
MODULE_PARM_DESC(hystart_detect, "use Hybrid Slow start "
"(0: disabled, 1: ACK train, 2: delay threshold, 3: both)");
module_param(use_ineff, uint, 0644);
MODULE_PARM_DESC(use_ineff, "use ineffectual backoff detection (threshold)");
module_param(use_shadow, bool, 0644);
MODULE_PARM_DESC(use_shadow, "use shadow window heuristic");
module_param(use_tolerance, bool, 0644);
MODULE_PARM_DESC(use_tolerance, "use loss tolerance heuristic");
struct cdg_minmax {
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
union {
struct {
s32 min;
s32 max;
};
u64 v64;
};
};
enum cdg_state {
CDG_UNKNOWN = 0,
CDG_NONFULL = 1,
CDG_FULL = 2,
CDG_BACKOFF = 3,
};
struct cdg {
struct cdg_minmax rtt;
struct cdg_minmax rtt_prev;
struct cdg_minmax *gradients;
struct cdg_minmax gsum;
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
bool gfilled;
u8 tail;
u8 state;
u8 delack;
u32 rtt_seq;
u32 shadow_wnd;
u16 backoff_cnt;
u16 sample_cnt;
s32 delay_min;
u32 last_ack;
u32 round_start;
};
/**
* nexp_u32 - negative base-e exponential
* @ux: x in units of micro
*
* Returns exp(ux * -1e-6) * U32_MAX.
*/
static u32 __pure nexp_u32(u32 ux)
{
static const u16 v[] = {
/* exp(-x)*65536-1 for x = 0, 0.000256, 0.000512, ... */
65535,
65518, 65501, 65468, 65401, 65267, 65001, 64470, 63422,
61378, 57484, 50423, 38795, 22965, 8047, 987, 14,
};
u32 msb = ux >> 8;
u32 res;
int i;
/* Cut off when ux >= 2^24 (actual result is <= 222/U32_MAX). */
if (msb > U16_MAX)
return 0;
/* Scale first eight bits linearly: */
res = U32_MAX - (ux & 0xff) * (U32_MAX / 1000000);
/* Obtain e^(x + y + ...) by computing e^x * e^y * ...: */
for (i = 1; msb; i++, msb >>= 1) {
u32 y = v[i & -(msb & 1)] + U32_C(1);
res = ((u64)res * y) >> 16;
}
return res;
}
/* Based on the HyStart algorithm (by Ha et al.) that is implemented in
* tcp_cubic. Differences/experimental changes:
* o Using Hayes' delayed ACK filter.
* o Using a usec clock for the ACK train.
* o Reset ACK train when application limited.
* o Invoked at any cwnd (i.e. also when cwnd < 16).
* o Invoked only when cwnd < ssthresh (i.e. not when cwnd == ssthresh).
*/
static void tcp_cdg_hystart_update(struct sock *sk)
{
struct cdg *ca = inet_csk_ca(sk);
struct tcp_sock *tp = tcp_sk(sk);
ca->delay_min = min_not_zero(ca->delay_min, ca->rtt.min);
if (ca->delay_min == 0)
return;
if (hystart_detect & HYSTART_ACK_TRAIN) {
u32 now_us = tp->tcp_mstamp;
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
if (ca->last_ack == 0 || !tcp_is_cwnd_limited(sk)) {
ca->last_ack = now_us;
ca->round_start = now_us;
} else if (before(now_us, ca->last_ack + 3000)) {
u32 base_owd = max(ca->delay_min / 2U, 125U);
ca->last_ack = now_us;
if (after(now_us, ca->round_start + base_owd)) {
NET_INC_STATS(sock_net(sk),
LINUX_MIB_TCPHYSTARTTRAINDETECT);
NET_ADD_STATS(sock_net(sk),
LINUX_MIB_TCPHYSTARTTRAINCWND,
tcp_snd_cwnd(tp));
tp->snd_ssthresh = tcp_snd_cwnd(tp);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
return;
}
}
}
if (hystart_detect & HYSTART_DELAY) {
if (ca->sample_cnt < 8) {
ca->sample_cnt++;
} else {
s32 thresh = max(ca->delay_min + ca->delay_min / 8U,
125U);
if (ca->rtt.min > thresh) {
NET_INC_STATS(sock_net(sk),
LINUX_MIB_TCPHYSTARTDELAYDETECT);
NET_ADD_STATS(sock_net(sk),
LINUX_MIB_TCPHYSTARTDELAYCWND,
tcp_snd_cwnd(tp));
tp->snd_ssthresh = tcp_snd_cwnd(tp);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
}
}
}
}
static s32 tcp_cdg_grad(struct cdg *ca)
{
s32 gmin = ca->rtt.min - ca->rtt_prev.min;
s32 gmax = ca->rtt.max - ca->rtt_prev.max;
s32 grad;
if (ca->gradients) {
ca->gsum.min += gmin - ca->gradients[ca->tail].min;
ca->gsum.max += gmax - ca->gradients[ca->tail].max;
ca->gradients[ca->tail].min = gmin;
ca->gradients[ca->tail].max = gmax;
ca->tail = (ca->tail + 1) & (window - 1);
gmin = ca->gsum.min;
gmax = ca->gsum.max;
}
/* We keep sums to ignore gradients during cwnd reductions;
* the paper's smoothed gradients otherwise simplify to:
* (rtt_latest - rtt_oldest) / window.
*
* We also drop division by window here.
*/
grad = gmin > 0 ? gmin : gmax;
/* Extrapolate missing values in gradient window: */
if (!ca->gfilled) {
if (!ca->gradients && window > 1)
grad *= window; /* Memory allocation failed. */
else if (ca->tail == 0)
ca->gfilled = true;
else
grad = (grad * window) / (int)ca->tail;
}
/* Backoff was effectual: */
if (gmin <= -32 || gmax <= -32)
ca->backoff_cnt = 0;
if (use_tolerance) {
/* Reduce small variations to zero: */
gmin = DIV_ROUND_CLOSEST(gmin, 64);
gmax = DIV_ROUND_CLOSEST(gmax, 64);
if (gmin > 0 && gmax <= 0)
ca->state = CDG_FULL;
else if ((gmin > 0 && gmax > 0) || gmax < 0)
ca->state = CDG_NONFULL;
}
return grad;
}
static bool tcp_cdg_backoff(struct sock *sk, u32 grad)
{
struct cdg *ca = inet_csk_ca(sk);
struct tcp_sock *tp = tcp_sk(sk);
if (get_random_u32() <= nexp_u32(grad * backoff_factor))
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
return false;
if (use_ineff) {
ca->backoff_cnt++;
if (ca->backoff_cnt > use_ineff)
return false;
}
ca->shadow_wnd = max(ca->shadow_wnd, tcp_snd_cwnd(tp));
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
ca->state = CDG_BACKOFF;
tcp_enter_cwr(sk);
return true;
}
/* Not called in CWR or Recovery state. */
static void tcp_cdg_cong_avoid(struct sock *sk, u32 ack, u32 acked)
{
struct cdg *ca = inet_csk_ca(sk);
struct tcp_sock *tp = tcp_sk(sk);
u32 prior_snd_cwnd;
u32 incr;
if (tcp_in_slow_start(tp) && hystart_detect)
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
tcp_cdg_hystart_update(sk);
if (after(ack, ca->rtt_seq) && ca->rtt.v64) {
s32 grad = 0;
if (ca->rtt_prev.v64)
grad = tcp_cdg_grad(ca);
ca->rtt_seq = tp->snd_nxt;
ca->rtt_prev = ca->rtt;
ca->rtt.v64 = 0;
ca->last_ack = 0;
ca->sample_cnt = 0;
if (grad > 0 && tcp_cdg_backoff(sk, grad))
return;
}
if (!tcp_is_cwnd_limited(sk)) {
ca->shadow_wnd = min(ca->shadow_wnd, tcp_snd_cwnd(tp));
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
return;
}
prior_snd_cwnd = tcp_snd_cwnd(tp);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
tcp_reno_cong_avoid(sk, ack, acked);
incr = tcp_snd_cwnd(tp) - prior_snd_cwnd;
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
ca->shadow_wnd = max(ca->shadow_wnd, ca->shadow_wnd + incr);
}
static void tcp_cdg_acked(struct sock *sk, const struct ack_sample *sample)
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
{
struct cdg *ca = inet_csk_ca(sk);
struct tcp_sock *tp = tcp_sk(sk);
if (sample->rtt_us <= 0)
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
return;
/* A heuristic for filtering delayed ACKs, adapted from:
* D.A. Hayes. "Timing enhancements to the FreeBSD kernel to support
* delay and rate based TCP mechanisms." TR 100219A. CAIA, 2010.
*/
if (tp->sacked_out == 0) {
if (sample->pkts_acked == 1 && ca->delack) {
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
/* A delayed ACK is only used for the minimum if it is
* provenly lower than an existing non-zero minimum.
*/
ca->rtt.min = min(ca->rtt.min, sample->rtt_us);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
ca->delack--;
return;
} else if (sample->pkts_acked > 1 && ca->delack < 5) {
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
ca->delack++;
}
}
ca->rtt.min = min_not_zero(ca->rtt.min, sample->rtt_us);
ca->rtt.max = max(ca->rtt.max, sample->rtt_us);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
}
static u32 tcp_cdg_ssthresh(struct sock *sk)
{
struct cdg *ca = inet_csk_ca(sk);
struct tcp_sock *tp = tcp_sk(sk);
if (ca->state == CDG_BACKOFF)
return max(2U, (tcp_snd_cwnd(tp) * min(1024U, backoff_beta)) >> 10);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
if (ca->state == CDG_NONFULL && use_tolerance)
return tcp_snd_cwnd(tp);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
ca->shadow_wnd = min(ca->shadow_wnd >> 1, tcp_snd_cwnd(tp));
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
if (use_shadow)
return max3(2U, ca->shadow_wnd, tcp_snd_cwnd(tp) >> 1);
return max(2U, tcp_snd_cwnd(tp) >> 1);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
}
static void tcp_cdg_cwnd_event(struct sock *sk, const enum tcp_ca_event ev)
{
struct cdg *ca = inet_csk_ca(sk);
struct tcp_sock *tp = tcp_sk(sk);
struct cdg_minmax *gradients;
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
switch (ev) {
case CA_EVENT_CWND_RESTART:
gradients = ca->gradients;
if (gradients)
memset(gradients, 0, window * sizeof(gradients[0]));
memset(ca, 0, sizeof(*ca));
ca->gradients = gradients;
ca->rtt_seq = tp->snd_nxt;
ca->shadow_wnd = tcp_snd_cwnd(tp);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
break;
case CA_EVENT_COMPLETE_CWR:
ca->state = CDG_UNKNOWN;
ca->rtt_seq = tp->snd_nxt;
ca->rtt_prev = ca->rtt;
ca->rtt.v64 = 0;
break;
default:
break;
}
}
static void tcp_cdg_init(struct sock *sk)
{
struct cdg *ca = inet_csk_ca(sk);
struct tcp_sock *tp = tcp_sk(sk);
tcp: cdg: allow tcp_cdg_release() to be called multiple times Apparently, mptcp is able to call tcp_disconnect() on an already disconnected flow. This is generally fine, unless current congestion control is CDG, because it might trigger a double-free [1] Instead of fixing MPTCP, and future bugs, we can make tcp_disconnect() more resilient. [1] BUG: KASAN: double-free in slab_free mm/slub.c:3539 [inline] BUG: KASAN: double-free in kfree+0xe2/0x580 mm/slub.c:4567 CPU: 0 PID: 3645 Comm: kworker/0:7 Not tainted 6.0.0-syzkaller-02734-g0326074ff465 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/22/2022 Workqueue: events mptcp_worker Call Trace: <TASK> __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106 print_address_description mm/kasan/report.c:317 [inline] print_report.cold+0x2ba/0x719 mm/kasan/report.c:433 kasan_report_invalid_free+0x81/0x190 mm/kasan/report.c:462 ____kasan_slab_free+0x18b/0x1c0 mm/kasan/common.c:356 kasan_slab_free include/linux/kasan.h:200 [inline] slab_free_hook mm/slub.c:1759 [inline] slab_free_freelist_hook+0x8b/0x1c0 mm/slub.c:1785 slab_free mm/slub.c:3539 [inline] kfree+0xe2/0x580 mm/slub.c:4567 tcp_disconnect+0x980/0x1e20 net/ipv4/tcp.c:3145 __mptcp_close_ssk+0x5ca/0x7e0 net/mptcp/protocol.c:2327 mptcp_do_fastclose net/mptcp/protocol.c:2592 [inline] mptcp_worker+0x78c/0xff0 net/mptcp/protocol.c:2627 process_one_work+0x991/0x1610 kernel/workqueue.c:2289 worker_thread+0x665/0x1080 kernel/workqueue.c:2436 kthread+0x2e4/0x3a0 kernel/kthread.c:376 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:306 </TASK> Allocated by task 3671: kasan_save_stack+0x1e/0x40 mm/kasan/common.c:38 kasan_set_track mm/kasan/common.c:45 [inline] set_alloc_info mm/kasan/common.c:437 [inline] ____kasan_kmalloc mm/kasan/common.c:516 [inline] ____kasan_kmalloc mm/kasan/common.c:475 [inline] __kasan_kmalloc+0xa9/0xd0 mm/kasan/common.c:525 kmalloc_array include/linux/slab.h:640 [inline] kcalloc include/linux/slab.h:671 [inline] tcp_cdg_init+0x10d/0x170 net/ipv4/tcp_cdg.c:380 tcp_init_congestion_control+0xab/0x550 net/ipv4/tcp_cong.c:193 tcp_reinit_congestion_control net/ipv4/tcp_cong.c:217 [inline] tcp_set_congestion_control+0x96c/0xaa0 net/ipv4/tcp_cong.c:391 do_tcp_setsockopt+0x505/0x2320 net/ipv4/tcp.c:3513 tcp_setsockopt+0xd4/0x100 net/ipv4/tcp.c:3801 mptcp_setsockopt+0x35f/0x2570 net/mptcp/sockopt.c:844 __sys_setsockopt+0x2d6/0x690 net/socket.c:2252 __do_sys_setsockopt net/socket.c:2263 [inline] __se_sys_setsockopt net/socket.c:2260 [inline] __x64_sys_setsockopt+0xba/0x150 net/socket.c:2260 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd Freed by task 16: kasan_save_stack+0x1e/0x40 mm/kasan/common.c:38 kasan_set_track+0x21/0x30 mm/kasan/common.c:45 kasan_set_free_info+0x20/0x30 mm/kasan/generic.c:370 ____kasan_slab_free mm/kasan/common.c:367 [inline] ____kasan_slab_free+0x166/0x1c0 mm/kasan/common.c:329 kasan_slab_free include/linux/kasan.h:200 [inline] slab_free_hook mm/slub.c:1759 [inline] slab_free_freelist_hook+0x8b/0x1c0 mm/slub.c:1785 slab_free mm/slub.c:3539 [inline] kfree+0xe2/0x580 mm/slub.c:4567 tcp_cleanup_congestion_control+0x70/0x120 net/ipv4/tcp_cong.c:226 tcp_v4_destroy_sock+0xdd/0x750 net/ipv4/tcp_ipv4.c:2254 tcp_v6_destroy_sock+0x11/0x20 net/ipv6/tcp_ipv6.c:1969 inet_csk_destroy_sock+0x196/0x440 net/ipv4/inet_connection_sock.c:1157 tcp_done+0x23b/0x340 net/ipv4/tcp.c:4649 tcp_rcv_state_process+0x40e7/0x4990 net/ipv4/tcp_input.c:6624 tcp_v6_do_rcv+0x3fc/0x13c0 net/ipv6/tcp_ipv6.c:1525 tcp_v6_rcv+0x2e8e/0x3830 net/ipv6/tcp_ipv6.c:1759 ip6_protocol_deliver_rcu+0x2db/0x1950 net/ipv6/ip6_input.c:439 ip6_input_finish+0x14c/0x2c0 net/ipv6/ip6_input.c:484 NF_HOOK include/linux/netfilter.h:302 [inline] NF_HOOK include/linux/netfilter.h:296 [inline] ip6_input+0x9c/0xd0 net/ipv6/ip6_input.c:493 dst_input include/net/dst.h:455 [inline] ip6_rcv_finish+0x193/0x2c0 net/ipv6/ip6_input.c:79 ip_sabotage_in net/bridge/br_netfilter_hooks.c:874 [inline] ip_sabotage_in+0x1fa/0x260 net/bridge/br_netfilter_hooks.c:865 nf_hook_entry_hookfn include/linux/netfilter.h:142 [inline] nf_hook_slow+0xc5/0x1f0 net/netfilter/core.c:614 nf_hook.constprop.0+0x3ac/0x650 include/linux/netfilter.h:257 NF_HOOK include/linux/netfilter.h:300 [inline] ipv6_rcv+0x9e/0x380 net/ipv6/ip6_input.c:309 __netif_receive_skb_one_core+0x114/0x180 net/core/dev.c:5485 __netif_receive_skb+0x1f/0x1c0 net/core/dev.c:5599 netif_receive_skb_internal net/core/dev.c:5685 [inline] netif_receive_skb+0x12f/0x8d0 net/core/dev.c:5744 NF_HOOK include/linux/netfilter.h:302 [inline] NF_HOOK include/linux/netfilter.h:296 [inline] br_pass_frame_up+0x303/0x410 net/bridge/br_input.c:68 br_handle_frame_finish+0x909/0x1aa0 net/bridge/br_input.c:199 br_nf_hook_thresh+0x2f8/0x3d0 net/bridge/br_netfilter_hooks.c:1041 br_nf_pre_routing_finish_ipv6+0x695/0xef0 net/bridge/br_netfilter_ipv6.c:207 NF_HOOK include/linux/netfilter.h:302 [inline] br_nf_pre_routing_ipv6+0x417/0x7c0 net/bridge/br_netfilter_ipv6.c:237 br_nf_pre_routing+0x1496/0x1fe0 net/bridge/br_netfilter_hooks.c:507 nf_hook_entry_hookfn include/linux/netfilter.h:142 [inline] nf_hook_bridge_pre net/bridge/br_input.c:255 [inline] br_handle_frame+0x9c9/0x12d0 net/bridge/br_input.c:399 __netif_receive_skb_core+0x9fe/0x38f0 net/core/dev.c:5379 __netif_receive_skb_one_core+0xae/0x180 net/core/dev.c:5483 __netif_receive_skb+0x1f/0x1c0 net/core/dev.c:5599 process_backlog+0x3a0/0x7c0 net/core/dev.c:5927 __napi_poll+0xb3/0x6d0 net/core/dev.c:6494 napi_poll net/core/dev.c:6561 [inline] net_rx_action+0x9c1/0xd90 net/core/dev.c:6672 __do_softirq+0x1d0/0x9c8 kernel/softirq.c:571 Fixes: 2b0a8c9eee81 ("tcp: add CDG congestion control") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-10-11 15:07:48 -07:00
ca->gradients = NULL;
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
/* We silently fall back to window = 1 if allocation fails. */
if (window > 1)
ca->gradients = kcalloc(window, sizeof(ca->gradients[0]),
GFP_NOWAIT | __GFP_NOWARN);
ca->rtt_seq = tp->snd_nxt;
ca->shadow_wnd = tcp_snd_cwnd(tp);
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
}
static void tcp_cdg_release(struct sock *sk)
{
struct cdg *ca = inet_csk_ca(sk);
kfree(ca->gradients);
tcp: cdg: allow tcp_cdg_release() to be called multiple times Apparently, mptcp is able to call tcp_disconnect() on an already disconnected flow. This is generally fine, unless current congestion control is CDG, because it might trigger a double-free [1] Instead of fixing MPTCP, and future bugs, we can make tcp_disconnect() more resilient. [1] BUG: KASAN: double-free in slab_free mm/slub.c:3539 [inline] BUG: KASAN: double-free in kfree+0xe2/0x580 mm/slub.c:4567 CPU: 0 PID: 3645 Comm: kworker/0:7 Not tainted 6.0.0-syzkaller-02734-g0326074ff465 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/22/2022 Workqueue: events mptcp_worker Call Trace: <TASK> __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106 print_address_description mm/kasan/report.c:317 [inline] print_report.cold+0x2ba/0x719 mm/kasan/report.c:433 kasan_report_invalid_free+0x81/0x190 mm/kasan/report.c:462 ____kasan_slab_free+0x18b/0x1c0 mm/kasan/common.c:356 kasan_slab_free include/linux/kasan.h:200 [inline] slab_free_hook mm/slub.c:1759 [inline] slab_free_freelist_hook+0x8b/0x1c0 mm/slub.c:1785 slab_free mm/slub.c:3539 [inline] kfree+0xe2/0x580 mm/slub.c:4567 tcp_disconnect+0x980/0x1e20 net/ipv4/tcp.c:3145 __mptcp_close_ssk+0x5ca/0x7e0 net/mptcp/protocol.c:2327 mptcp_do_fastclose net/mptcp/protocol.c:2592 [inline] mptcp_worker+0x78c/0xff0 net/mptcp/protocol.c:2627 process_one_work+0x991/0x1610 kernel/workqueue.c:2289 worker_thread+0x665/0x1080 kernel/workqueue.c:2436 kthread+0x2e4/0x3a0 kernel/kthread.c:376 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:306 </TASK> Allocated by task 3671: kasan_save_stack+0x1e/0x40 mm/kasan/common.c:38 kasan_set_track mm/kasan/common.c:45 [inline] set_alloc_info mm/kasan/common.c:437 [inline] ____kasan_kmalloc mm/kasan/common.c:516 [inline] ____kasan_kmalloc mm/kasan/common.c:475 [inline] __kasan_kmalloc+0xa9/0xd0 mm/kasan/common.c:525 kmalloc_array include/linux/slab.h:640 [inline] kcalloc include/linux/slab.h:671 [inline] tcp_cdg_init+0x10d/0x170 net/ipv4/tcp_cdg.c:380 tcp_init_congestion_control+0xab/0x550 net/ipv4/tcp_cong.c:193 tcp_reinit_congestion_control net/ipv4/tcp_cong.c:217 [inline] tcp_set_congestion_control+0x96c/0xaa0 net/ipv4/tcp_cong.c:391 do_tcp_setsockopt+0x505/0x2320 net/ipv4/tcp.c:3513 tcp_setsockopt+0xd4/0x100 net/ipv4/tcp.c:3801 mptcp_setsockopt+0x35f/0x2570 net/mptcp/sockopt.c:844 __sys_setsockopt+0x2d6/0x690 net/socket.c:2252 __do_sys_setsockopt net/socket.c:2263 [inline] __se_sys_setsockopt net/socket.c:2260 [inline] __x64_sys_setsockopt+0xba/0x150 net/socket.c:2260 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd Freed by task 16: kasan_save_stack+0x1e/0x40 mm/kasan/common.c:38 kasan_set_track+0x21/0x30 mm/kasan/common.c:45 kasan_set_free_info+0x20/0x30 mm/kasan/generic.c:370 ____kasan_slab_free mm/kasan/common.c:367 [inline] ____kasan_slab_free+0x166/0x1c0 mm/kasan/common.c:329 kasan_slab_free include/linux/kasan.h:200 [inline] slab_free_hook mm/slub.c:1759 [inline] slab_free_freelist_hook+0x8b/0x1c0 mm/slub.c:1785 slab_free mm/slub.c:3539 [inline] kfree+0xe2/0x580 mm/slub.c:4567 tcp_cleanup_congestion_control+0x70/0x120 net/ipv4/tcp_cong.c:226 tcp_v4_destroy_sock+0xdd/0x750 net/ipv4/tcp_ipv4.c:2254 tcp_v6_destroy_sock+0x11/0x20 net/ipv6/tcp_ipv6.c:1969 inet_csk_destroy_sock+0x196/0x440 net/ipv4/inet_connection_sock.c:1157 tcp_done+0x23b/0x340 net/ipv4/tcp.c:4649 tcp_rcv_state_process+0x40e7/0x4990 net/ipv4/tcp_input.c:6624 tcp_v6_do_rcv+0x3fc/0x13c0 net/ipv6/tcp_ipv6.c:1525 tcp_v6_rcv+0x2e8e/0x3830 net/ipv6/tcp_ipv6.c:1759 ip6_protocol_deliver_rcu+0x2db/0x1950 net/ipv6/ip6_input.c:439 ip6_input_finish+0x14c/0x2c0 net/ipv6/ip6_input.c:484 NF_HOOK include/linux/netfilter.h:302 [inline] NF_HOOK include/linux/netfilter.h:296 [inline] ip6_input+0x9c/0xd0 net/ipv6/ip6_input.c:493 dst_input include/net/dst.h:455 [inline] ip6_rcv_finish+0x193/0x2c0 net/ipv6/ip6_input.c:79 ip_sabotage_in net/bridge/br_netfilter_hooks.c:874 [inline] ip_sabotage_in+0x1fa/0x260 net/bridge/br_netfilter_hooks.c:865 nf_hook_entry_hookfn include/linux/netfilter.h:142 [inline] nf_hook_slow+0xc5/0x1f0 net/netfilter/core.c:614 nf_hook.constprop.0+0x3ac/0x650 include/linux/netfilter.h:257 NF_HOOK include/linux/netfilter.h:300 [inline] ipv6_rcv+0x9e/0x380 net/ipv6/ip6_input.c:309 __netif_receive_skb_one_core+0x114/0x180 net/core/dev.c:5485 __netif_receive_skb+0x1f/0x1c0 net/core/dev.c:5599 netif_receive_skb_internal net/core/dev.c:5685 [inline] netif_receive_skb+0x12f/0x8d0 net/core/dev.c:5744 NF_HOOK include/linux/netfilter.h:302 [inline] NF_HOOK include/linux/netfilter.h:296 [inline] br_pass_frame_up+0x303/0x410 net/bridge/br_input.c:68 br_handle_frame_finish+0x909/0x1aa0 net/bridge/br_input.c:199 br_nf_hook_thresh+0x2f8/0x3d0 net/bridge/br_netfilter_hooks.c:1041 br_nf_pre_routing_finish_ipv6+0x695/0xef0 net/bridge/br_netfilter_ipv6.c:207 NF_HOOK include/linux/netfilter.h:302 [inline] br_nf_pre_routing_ipv6+0x417/0x7c0 net/bridge/br_netfilter_ipv6.c:237 br_nf_pre_routing+0x1496/0x1fe0 net/bridge/br_netfilter_hooks.c:507 nf_hook_entry_hookfn include/linux/netfilter.h:142 [inline] nf_hook_bridge_pre net/bridge/br_input.c:255 [inline] br_handle_frame+0x9c9/0x12d0 net/bridge/br_input.c:399 __netif_receive_skb_core+0x9fe/0x38f0 net/core/dev.c:5379 __netif_receive_skb_one_core+0xae/0x180 net/core/dev.c:5483 __netif_receive_skb+0x1f/0x1c0 net/core/dev.c:5599 process_backlog+0x3a0/0x7c0 net/core/dev.c:5927 __napi_poll+0xb3/0x6d0 net/core/dev.c:6494 napi_poll net/core/dev.c:6561 [inline] net_rx_action+0x9c1/0xd90 net/core/dev.c:6672 __do_softirq+0x1d0/0x9c8 kernel/softirq.c:571 Fixes: 2b0a8c9eee81 ("tcp: add CDG congestion control") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2022-10-11 15:07:48 -07:00
ca->gradients = NULL;
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
}
static struct tcp_congestion_ops tcp_cdg __read_mostly = {
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
.cong_avoid = tcp_cdg_cong_avoid,
.cwnd_event = tcp_cdg_cwnd_event,
.pkts_acked = tcp_cdg_acked,
.undo_cwnd = tcp_reno_undo_cwnd,
tcp: add CDG congestion control CAIA Delay-Gradient (CDG) is a TCP congestion control that modifies the TCP sender in order to [1]: o Use the delay gradient as a congestion signal. o Back off with an average probability that is independent of the RTT. o Coexist with flows that use loss-based congestion control, i.e., flows that are unresponsive to the delay signal. o Tolerate packet loss unrelated to congestion. (Disabled by default.) Its FreeBSD implementation was presented for the ICCRG in July 2012; slides are available at http://www.ietf.org/proceedings/84/iccrg.html Running the experiment scenarios in [1] suggests that our implementation achieves more goodput compared with FreeBSD 10.0 senders, although it also causes more queueing delay for a given backoff factor. The loss tolerance heuristic is disabled by default due to safety concerns for its use in the Internet [2, p. 45-46]. We use a variant of the Hybrid Slow start algorithm in tcp_cubic to reduce the probability of slow start overshoot. [1] D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using delay gradients." In Networking 2011, pages 328-341. Springer, 2011. [2] K.K. Jonassen. "Implementing CAIA Delay-Gradient in Linux." MSc thesis. Department of Informatics, University of Oslo, 2015. Cc: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Stephen Hemminger <stephen@networkplumber.org> Cc: Neal Cardwell <ncardwell@google.com> Cc: David Hayes <davihay@ifi.uio.no> Cc: Andreas Petlund <apetlund@simula.no> Cc: Dave Taht <dave.taht@bufferbloat.net> Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu> Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no> Acked-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-10 10:08:17 -07:00
.ssthresh = tcp_cdg_ssthresh,
.release = tcp_cdg_release,
.init = tcp_cdg_init,
.owner = THIS_MODULE,
.name = "cdg",
};
static int __init tcp_cdg_register(void)
{
if (backoff_beta > 1024 || window < 1 || window > 256)
return -ERANGE;
if (!is_power_of_2(window))
return -EINVAL;
BUILD_BUG_ON(sizeof(struct cdg) > ICSK_CA_PRIV_SIZE);
tcp_register_congestion_control(&tcp_cdg);
return 0;
}
static void __exit tcp_cdg_unregister(void)
{
tcp_unregister_congestion_control(&tcp_cdg);
}
module_init(tcp_cdg_register);
module_exit(tcp_cdg_unregister);
MODULE_AUTHOR("Kenneth Klette Jonassen");
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("TCP CDG");