Page MenuHomeFreeBSD
Differential D28357 Diff 82985 sys/netinet/tcp_ratelimit.c

Changeset View

Standalone View

View Options

sys/netinet/tcp_ratelimit.c

Context not available.
#ifdef INET6 #ifdef INET6
#include <netinet6/tcp6_var.h> #include <netinet6/tcp6_var.h>
#endif #endif
#include <netinet/tcp_hpts.h>
#include <netinet/tcp_log_buf.h>
#include <netinet/tcp_ratelimit.h> #include <netinet/tcp_ratelimit.h>
#ifndef USECS_IN_SECOND #ifndef USECS_IN_SECOND
#define USECS_IN_SECOND 1000000 #define USECS_IN_SECOND 1000000
Context not available.
180500, /* 1.44Mpbs - rate 2 common rate */ 180500, /* 1.44Mpbs - rate 2 common rate */
375000, /* 3Mbps - rate 3 */ 375000, /* 3Mbps - rate 3 */
625000, /* 5Mbps - rate 4 */ 625000, /* 5Mbps - rate 4 */
875000, /* 7Mbps - rate 5 */ 1250000, /* 10Mbps - rate 5 */
1125000, /* 9Mbps - rate 6 */ 1875000, /* 15Mbps - rate 6 */
1375000, /* 11Mbps - rate 7 */ 2500000, /* 20Mbps - rate 7 */
1625000, /* 13Mbps - rate 8 */ 3125000, /* 25Mbps - rate 8 */
2625000, /* 21Mbps - rate 9 */ 3750000, /* 30Mbps - rate 9 */
3875000, /* 31Mbps - rate 10 */ 4375000, /* 35Mbps - rate 10 */
5125000, /* 41Meg - rate 11 */ 5000000, /* 40Meg - rate 11 */
12500000, /* 100Mbps - rate 12 */ 6250000, /* 50Mbps - rate 12 */
25000000, /* 200Mbps - rate 13 */ 12500000, /* 100Mbps - rate 13 */
50000000, /* 400Mbps - rate 14 */ 25000000, /* 200Mbps - rate 14 */
63750000, /* 51Mbps - rate 15 */ 50000000, /* 400Mbps - rate 15 */
100000000, /* 800Mbps - rate 16 */ 100000000, /* 800Mbps - rate 16 */
1875000, /* 15Mbps - rate 17 */ 5625000, /* 45Mbps - rate 17 */
2125000, /* 17Mbps - rate 18 */ 6875000, /* 55Mbps - rate 19 */
2375000, /* 19Mbps - rate 19 */ 7500000, /* 60Mbps - rate 20 */
2875000, /* 23Mbps - rate 20 */ 8125000, /* 65Mbps - rate 21 */
3125000, /* 25Mbps - rate 21 */ 8750000, /* 70Mbps - rate 22 */
3375000, /* 27Mbps - rate 22 */ 9375000, /* 75Mbps - rate 23 */
3625000, /* 29Mbps - rate 23 */ 10000000, /* 80Mbps - rate 24 */
4125000, /* 33Mbps - rate 24 */ 10625000, /* 85Mbps - rate 25 */
4375000, /* 35Mbps - rate 25 */ 11250000, /* 90Mbps - rate 26 */
4625000, /* 37Mbps - rate 26 */ 11875000, /* 95Mbps - rate 27 */
4875000, /* 39Mbps - rate 27 */ 12500000, /* 100Mbps - rate 28 */
5375000, /* 43Mbps - rate 28 */ 13750000, /* 110Mbps - rate 29 */
5625000, /* 45Mbps - rate 29 */ 15000000, /* 120Mbps - rate 30 */
5875000, /* 47Mbps - rate 30 */ 16250000, /* 130Mbps - rate 31 */
6125000, /* 49Mbps - rate 31 */ 17500000, /* 140Mbps - rate 32 */
6625000, /* 53Mbps - rate 32 */ 18750000, /* 150Mbps - rate 33 */
6875000, /* 55Mbps - rate 33 */ 20000000, /* 160Mbps - rate 34 */
7125000, /* 57Mbps - rate 34 */ 21250000, /* 170Mbps - rate 35 */
7375000, /* 59Mbps - rate 35 */ 22500000, /* 180Mbps - rate 36 */
7625000, /* 61Mbps - rate 36 */ 23750000, /* 190Mbps - rate 37 */
7875000, /* 63Mbps - rate 37 */ 26250000, /* 210Mbps - rate 38 */
8125000, /* 65Mbps - rate 38 */ 27500000, /* 220Mbps - rate 39 */
8375000, /* 67Mbps - rate 39 */ 28750000, /* 230Mbps - rate 40 */
8625000, /* 69Mbps - rate 40 */ 30000000, /* 240Mbps - rate 41 */
8875000, /* 71Mbps - rate 41 */ 31250000, /* 250Mbps - rate 42 */
9125000, /* 73Mbps - rate 42 */ 34375000, /* 275Mbps - rate 43 */
9375000, /* 75Mbps - rate 43 */ 37500000, /* 300Mbps - rate 44 */
9625000, /* 77Mbps - rate 44 */ 40625000, /* 325Mbps - rate 45 */
9875000, /* 79Mbps - rate 45 */ 43750000, /* 350Mbps - rate 46 */
10125000, /* 81Mbps - rate 46 */ 46875000, /* 375Mbps - rate 47 */
10375000, /* 83Mbps - rate 47 */ 53125000, /* 425Mbps - rate 48 */
10625000, /* 85Mbps - rate 48 */ 56250000, /* 450Mbps - rate 49 */
10875000, /* 87Mbps - rate 49 */ 59375000, /* 475Mbps - rate 50 */
11125000, /* 89Mbps - rate 50 */ 62500000, /* 500Mbps - rate 51 */
11375000, /* 91Mbps - rate 51 */ 68750000, /* 550Mbps - rate 52 */
11625000, /* 93Mbps - rate 52 */ 75000000, /* 600Mbps - rate 53 */
11875000, /* 95Mbps - rate 53 */ 81250000, /* 650Mbps - rate 54 */
13125000, /* 105Mbps - rate 54 */ 87500000, /* 700Mbps - rate 55 */
13750000, /* 110Mbps - rate 55 */ 93750000, /* 750Mbps - rate 56 */
14375000, /* 115Mbps - rate 56 */ 106250000, /* 850Mbps - rate 57 */
15000000, /* 120Mbps - rate 57 */ 112500000, /* 900Mbps - rate 58 */
15625000, /* 125Mbps - rate 58 */ 125000000, /* 1Gbps - rate 59 */
16250000, /* 130Mbps - rate 59 */ 156250000, /* 1.25Gps - rate 60 */
16875000, /* 135Mbps - rate 60 */ 187500000, /* 1.5Gps - rate 61 */
17500000, /* 140Mbps - rate 61 */ 218750000, /* 1.75Gps - rate 62 */
18125000, /* 145Mbps - rate 62 */ 250000000, /* 2Gbps - rate 63 */
18750000, /* 150Mbps - rate 64 */ 281250000, /* 2.25Gps - rate 64 */
20000000, /* 160Mbps - rate 65 */ 312500000, /* 2.5Gbps - rate 65 */
21250000, /* 170Mbps - rate 66 */ 343750000, /* 2.75Gbps - rate 66 */
22500000, /* 180Mbps - rate 67 */ 375000000, /* 3Gbps - rate 67 */
23750000, /* 190Mbps - rate 68 */ 500000000, /* 4Gbps - rate 68 */
26250000, /* 210Mbps - rate 69 */ 625000000, /* 5Gbps - rate 69 */
27500000, /* 220Mbps - rate 70 */ 750000000, /* 6Gbps - rate 70 */
28750000, /* 230Mbps - rate 71 */ 875000000, /* 7Gbps - rate 71 */
30000000, /* 240Mbps - rate 72 */ 1000000000, /* 8Gbps - rate 72 */
31250000, /* 250Mbps - rate 73 */ 1125000000, /* 9Gbps - rate 73 */
34375000, /* 275Mbps - rate 74 */ 1250000000, /* 10Gbps - rate 74 */
37500000, /* 300Mbps - rate 75 */ 1875000000, /* 15Gbps - rate 75 */
40625000, /* 325Mbps - rate 76 */ 2500000000 /* 20Gbps - rate 76 */
43750000, /* 350Mbps - rate 77 */
46875000, /* 375Mbps - rate 78 */
53125000, /* 425Mbps - rate 79 */
56250000, /* 450Mbps - rate 80 */
59375000, /* 475Mbps - rate 81 */
62500000, /* 500Mbps - rate 82 */
68750000, /* 550Mbps - rate 83 */
75000000, /* 600Mbps - rate 84 */
81250000, /* 650Mbps - rate 85 */
87500000, /* 700Mbps - rate 86 */
93750000, /* 750Mbps - rate 87 */
106250000, /* 850Mbps - rate 88 */
112500000, /* 900Mbps - rate 89 */
125000000, /* 1Gbps - rate 90 */
156250000, /* 1.25Gps - rate 91 */
187500000, /* 1.5Gps - rate 92 */
218750000, /* 1.75Gps - rate 93 */
250000000, /* 2Gbps - rate 94 */
281250000, /* 2.25Gps - rate 95 */
312500000, /* 2.5Gbps - rate 96 */
343750000, /* 2.75Gbps - rate 97 */
375000000, /* 3Gbps - rate 98 */
500000000, /* 4Gbps - rate 99 */
625000000, /* 5Gbps - rate 100 */
750000000, /* 6Gbps - rate 101 */
875000000, /* 7Gbps - rate 102 */
1000000000, /* 8Gbps - rate 103 */
1125000000, /* 9Gbps - rate 104 */
1250000000, /* 10Gbps - rate 105 */
1875000000, /* 15Gbps - rate 106 */
2500000000 /* 20Gbps - rate 107 */
}; };
#define MAX_HDWR_RATES (sizeof(desired_rates)/sizeof(uint64_t)) #define MAX_HDWR_RATES (sizeof(desired_rates)/sizeof(uint64_t))
Context not available.
static struct mtx rs_mtx; static struct mtx rs_mtx;
uint32_t rs_number_alive; uint32_t rs_number_alive;
uint32_t rs_number_dead; uint32_t rs_number_dead;
static uint32_t rs_floor_mss = 0;
static uint32_t wait_time_floor = 8000; /* 8 ms */
static uint32_t rs_hw_floor_mss = 16;
static uint32_t num_of_waits_allowed = 1; /* How many time blocks are we willing to wait */
SYSCTL_NODE(_net_inet_tcp, OID_AUTO, rl, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, SYSCTL_NODE(_net_inet_tcp, OID_AUTO, rl, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
"TCP Ratelimit stats"); "TCP Ratelimit stats");
Context not available.
SYSCTL_UINT(_net_inet_tcp_rl, OID_AUTO, dead, CTLFLAG_RW, SYSCTL_UINT(_net_inet_tcp_rl, OID_AUTO, dead, CTLFLAG_RW,
&rs_number_dead, 0, &rs_number_dead, 0,
"Number of interfaces departing from ratelimiting"); "Number of interfaces departing from ratelimiting");
SYSCTL_UINT(_net_inet_tcp_rl, OID_AUTO, floor_mss, CTLFLAG_RW,
&rs_floor_mss, 0,
"Number of MSS that will override the normal minimums (0 means don't enforce)");
SYSCTL_UINT(_net_inet_tcp_rl, OID_AUTO, wait_floor, CTLFLAG_RW,
&wait_time_floor, 2000,
"Has b/w increases what is the wait floor we are willing to wait at the end?");
SYSCTL_UINT(_net_inet_tcp_rl, OID_AUTO, time_blocks, CTLFLAG_RW,
&num_of_waits_allowed, 1,
"How many time blocks on the end should software pacing be willing to wait?");
SYSCTL_UINT(_net_inet_tcp_rl, OID_AUTO, hw_floor_mss, CTLFLAG_RW,
&rs_hw_floor_mss, 16,
"Number of mss that are a minum for hardware pacing?");
static void static void
rl_add_syctl_entries(struct sysctl_oid *rl_sysctl_root, struct tcp_rate_set *rs) rl_add_syctl_entries(struct sysctl_oid *rl_sysctl_root, struct tcp_rate_set *rs)
Context not available.
OID_AUTO, "rate", CTLFLAG_RD, OID_AUTO, "rate", CTLFLAG_RD,
&rs->rs_rlt[i].rate, 0, &rs->rs_rlt[i].rate, 0,
"Rate in bytes per second"); "Rate in bytes per second");
SYSCTL_ADD_U64(&rs->sysctl_ctx,
SYSCTL_CHILDREN(rl_rate_num),
OID_AUTO, "using", CTLFLAG_RD,
&rs->rs_rlt[i].using, 0,
"Number of flows using");
SYSCTL_ADD_U64(&rs->sysctl_ctx,
SYSCTL_CHILDREN(rl_rate_num),
OID_AUTO, "enobufs", CTLFLAG_RD,
&rs->rs_rlt[i].rs_num_enobufs, 0,
"Number of enobufs logged on this rate");
} }
} }
#endif #endif
Context not available.
} }
#ifdef INET #ifdef INET
#ifdef NETFLIX_STATS
extern counter_u64_t rate_limit_new;
extern counter_u64_t rate_limit_chg;
extern counter_u64_t rate_limit_set_ok; extern counter_u64_t rate_limit_set_ok;
extern counter_u64_t rate_limit_active; extern counter_u64_t rate_limit_active;
extern counter_u64_t rate_limit_alloc_fail; extern counter_u64_t rate_limit_alloc_fail;
#endif #endif
#endif
static int static int
rl_attach_txrtlmt(struct ifnet *ifp, rl_attach_txrtlmt(struct ifnet *ifp,
Context not available.
error = m_snd_tag_alloc(ifp, &params, tag); error = m_snd_tag_alloc(ifp, &params, tag);
#ifdef INET #ifdef INET
#ifdef NETFLIX_STATS
if (error == 0) { if (error == 0) {
counter_u64_add(rate_limit_set_ok, 1); counter_u64_add(rate_limit_set_ok, 1);
counter_u64_add(rate_limit_active, 1); counter_u64_add(rate_limit_active, 1);
} else if (error != EOPNOTSUPP) } else if (error != EOPNOTSUPP)
counter_u64_add(rate_limit_alloc_fail, 1); counter_u64_add(rate_limit_alloc_fail, 1);
#endif
#endif #endif
return (error); return (error);
} }
Context not available.
int i; int i;
struct if_ratelimit_query_results rl; struct if_ratelimit_query_results rl;
struct sysctl_oid *rl_sysctl_root; struct sysctl_oid *rl_sysctl_root;
struct epoch_tracker et;
/* /*
* We expect to enter with the * We expect to enter with the
* mutex locked. * mutex locked.
Context not available.
CTLFLAG_RW | CTLFLAG_MPSAFE, 0, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
""); "");
rl_add_syctl_entries(rl_sysctl_root, rs); rl_add_syctl_entries(rl_sysctl_root, rs);
NET_EPOCH_ENTER(et);
mtx_lock(&rs_mtx); mtx_lock(&rs_mtx);
CK_LIST_INSERT_HEAD(&int_rs, rs, next); CK_LIST_INSERT_HEAD(&int_rs, rs, next);
mtx_unlock(&rs_mtx); mtx_unlock(&rs_mtx);
NET_EPOCH_EXIT(et);
return (rs); return (rs);
} else if ((rl.flags & RT_IS_INDIRECT) == RT_IS_INDIRECT) { } else if ((rl.flags & RT_IS_INDIRECT) == RT_IS_INDIRECT) {
memset(rs, 0, sizeof(struct tcp_rate_set)); memset(rs, 0, sizeof(struct tcp_rate_set));
Context not available.
CTLFLAG_RW | CTLFLAG_MPSAFE, 0, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
""); "");
rl_add_syctl_entries(rl_sysctl_root, rs); rl_add_syctl_entries(rl_sysctl_root, rs);
NET_EPOCH_ENTER(et);
mtx_lock(&rs_mtx); mtx_lock(&rs_mtx);
CK_LIST_INSERT_HEAD(&int_rs, rs, next); CK_LIST_INSERT_HEAD(&int_rs, rs, next);
mtx_unlock(&rs_mtx); mtx_unlock(&rs_mtx);
NET_EPOCH_EXIT(et);
return (rs); return (rs);
} else if ((rl.flags & RT_IS_FIXED_TABLE) == RT_IS_FIXED_TABLE) { } else if ((rl.flags & RT_IS_FIXED_TABLE) == RT_IS_FIXED_TABLE) {
/* Mellanox C4 likely */ /* Mellanox C4 likely */
Context not available.
*/ */
rs->rs_rlt[i].ptbl = rs; rs->rs_rlt[i].ptbl = rs;
rs->rs_rlt[i].tag = NULL; rs->rs_rlt[i].tag = NULL;
rs->rs_rlt[i].using = 0;
rs->rs_rlt[i].rs_num_enobufs = 0;
/* /*
* Calculate the time between. * Calculate the time between.
*/ */
Context not available.
CTLFLAG_RW | CTLFLAG_MPSAFE, 0, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
""); "");
rl_add_syctl_entries(rl_sysctl_root, rs); rl_add_syctl_entries(rl_sysctl_root, rs);
NET_EPOCH_ENTER(et);
mtx_lock(&rs_mtx); mtx_lock(&rs_mtx);
CK_LIST_INSERT_HEAD(&int_rs, rs, next); CK_LIST_INSERT_HEAD(&int_rs, rs, next);
mtx_unlock(&rs_mtx); mtx_unlock(&rs_mtx);
NET_EPOCH_EXIT(et);
return (rs); return (rs);
} }
/*
* For an explanation of why the argument is volatile please
* look at the comments around rt_setup_rate().
*/
static const struct tcp_hwrate_limit_table * static const struct tcp_hwrate_limit_table *
tcp_int_find_suitable_rate(const struct tcp_rate_set *rs, tcp_int_find_suitable_rate(const volatile struct tcp_rate_set *rs,
uint64_t bytes_per_sec, uint32_t flags) uint64_t bytes_per_sec, uint32_t flags, uint64_t *lower_rate)
{ {
struct tcp_hwrate_limit_table *arte = NULL, *rte = NULL; struct tcp_hwrate_limit_table *arte = NULL, *rte = NULL;
uint64_t mbits_per_sec, ind_calc; uint64_t mbits_per_sec, ind_calc, previous_rate = 0;
int i; int i;
mbits_per_sec = (bytes_per_sec * 8); mbits_per_sec = (bytes_per_sec * 8);
Context not available.
* Smaller than 1Meg, only * Smaller than 1Meg, only
* 3 entries can match it. * 3 entries can match it.
*/ */
previous_rate = 0;
for(i = rs->rs_lowest_valid; i < 3; i++) { for(i = rs->rs_lowest_valid; i < 3; i++) {
if (bytes_per_sec <= rs->rs_rlt[i].rate) { if (bytes_per_sec <= rs->rs_rlt[i].rate) {
rte = &rs->rs_rlt[i]; rte = &rs->rs_rlt[i];
Context not available.
} else if (rs->rs_rlt[i].flags & HDWRPACE_INITED) { } else if (rs->rs_rlt[i].flags & HDWRPACE_INITED) {
arte = &rs->rs_rlt[i]; arte = &rs->rs_rlt[i];
} }
previous_rate = rs->rs_rlt[i].rate;
} }
goto done; goto done;
} else if ((mbits_per_sec > RS_ONE_GIGABIT_PERSEC) && } else if ((mbits_per_sec > RS_ONE_GIGABIT_PERSEC) &&
Context not available.
rte = &rs->rs_rlt[(ALL_HARDWARE_RATES-1)]; rte = &rs->rs_rlt[(ALL_HARDWARE_RATES-1)];
else else
arte = &rs->rs_rlt[(ALL_HARDWARE_RATES-1)]; arte = &rs->rs_rlt[(ALL_HARDWARE_RATES-1)];
previous_rate = rs->rs_rlt[(ALL_HARDWARE_RATES-2)].rate;
goto done; goto done;
} }
/* /*
Context not available.
ind_calc = ALL_HARDWARE_RATES-1; ind_calc = ALL_HARDWARE_RATES-1;
} }
if ((ind_calc >= rs->rs_lowest_valid) && if ((ind_calc >= rs->rs_lowest_valid) &&
(ind_calc <= rs->rs_highest_valid)) (ind_calc <= rs->rs_highest_valid)) {
rte = &rs->rs_rlt[ind_calc]; rte = &rs->rs_rlt[ind_calc];
if (ind_calc >= 1)
previous_rate = rs->rs_rlt[(ind_calc-1)].rate;
}
} else if (flags & RS_PACING_EXACT_MATCH) { } else if (flags & RS_PACING_EXACT_MATCH) {
if ((mbits_per_sec < RS_ONE_MEGABIT_PERSEC) && if ((mbits_per_sec < RS_ONE_MEGABIT_PERSEC) &&
(rs->rs_lowest_valid <= 2)){ (rs->rs_lowest_valid <= 2)){
Context not available.
for (i=2; i>=rs->rs_lowest_valid; i--) { for (i=2; i>=rs->rs_lowest_valid; i--) {
if (bytes_per_sec < rs->rs_rlt[i].rate) { if (bytes_per_sec < rs->rs_rlt[i].rate) {
rte = &rs->rs_rlt[i]; rte = &rs->rs_rlt[i];
if (i >= 1) {
previous_rate = rs->rs_rlt[(i-1)].rate;
}
break; break;
} else if ((flags & RS_PACING_GEQ) && } else if ((flags & RS_PACING_GEQ) &&
(bytes_per_sec == rs->rs_rlt[i].rate)) { (bytes_per_sec == rs->rs_rlt[i].rate)) {
rte = &rs->rs_rlt[i]; rte = &rs->rs_rlt[i];
if (i >= 1) {
previous_rate = rs->rs_rlt[(i-1)].rate;
}
break; break;
} else { } else {
arte = &rs->rs_rlt[i]; /* new alternate */ arte = &rs->rs_rlt[i]; /* new alternate */
Context not available.
/* The top rate is an alternative */ /* The top rate is an alternative */
arte = &rs->rs_rlt[(ALL_HARDWARE_RATES-1)]; arte = &rs->rs_rlt[(ALL_HARDWARE_RATES-1)];
} }
previous_rate = rs->rs_rlt[(ALL_HARDWARE_RATES-2)].rate;
} else { } else {
/* Its in our range 1Meg - 1Gig */ /* Its in our range 1Meg - 1Gig */
if (flags & RS_PACING_GEQ) { if (flags & RS_PACING_GEQ) {
Context not available.
ind_calc = (ALL_HARDWARE_RATES-1); ind_calc = (ALL_HARDWARE_RATES-1);
} }
rte = &rs->rs_rlt[ind_calc]; rte = &rs->rs_rlt[ind_calc];
if (ind_calc >= 1)
previous_rate = rs->rs_rlt[(ind_calc-1)].rate;
} }
goto done; goto done;
} }
Context not available.
/* This should not happen */ /* This should not happen */
ind_calc = ALL_HARDWARE_RATES-1; ind_calc = ALL_HARDWARE_RATES-1;
} }
if (rs->rs_rlt[ind_calc].flags & HDWRPACE_INITED) if (rs->rs_rlt[ind_calc].flags & HDWRPACE_INITED) {
rte = &rs->rs_rlt[ind_calc]; rte = &rs->rs_rlt[ind_calc];
if (ind_calc >= 1)
previous_rate = rs->rs_rlt[(ind_calc-1)].rate;
}
} }
} }
done: done:
Context not available.
/* We can use the substitute */ /* We can use the substitute */
rte = arte; rte = arte;
} }
if (lower_rate)
*lower_rate = previous_rate;
return (rte); return (rte);
} }
/*
* For an explanation of why the argument is volatile please
* look at the comments around rt_setup_rate().
*/
static const struct tcp_hwrate_limit_table * static const struct tcp_hwrate_limit_table *
tcp_find_suitable_rate(const struct tcp_rate_set *rs, uint64_t bytes_per_sec, uint32_t flags) tcp_find_suitable_rate(const volatile struct tcp_rate_set *rs, uint64_t bytes_per_sec, uint32_t flags, uint64_t *lower_rate)
{ {
/** /**
* Hunt the rate table with the restrictions in flags and find a * Hunt the rate table with the restrictions in flags and find a
Context not available.
*/ */
int i, matched; int i, matched;
struct tcp_hwrate_limit_table *rte = NULL; struct tcp_hwrate_limit_table *rte = NULL;
uint64_t previous_rate = 0;
if ((rs->rs_flags & RS_INT_TBL) && if ((rs->rs_flags & RS_INT_TBL) &&
(rs->rs_rate_cnt >= ALL_HARDWARE_RATES)) { (rs->rs_rate_cnt >= ALL_HARDWARE_RATES)) {
Context not available.
* from 1Meg - 1000Meg in 1Meg increments. * from 1Meg - 1000Meg in 1Meg increments.
* Use an alternate method to "lookup". * Use an alternate method to "lookup".
*/ */
return (tcp_int_find_suitable_rate(rs, bytes_per_sec, flags)); return (tcp_int_find_suitable_rate(rs, bytes_per_sec, flags, lower_rate));
} }
if ((flags & RS_PACING_LT) || if ((flags & RS_PACING_LT) ||
(flags & RS_PACING_EXACT_MATCH)) { (flags & RS_PACING_EXACT_MATCH)) {
Context not available.
(bytes_per_sec == rs->rs_rlt[i].rate)) { (bytes_per_sec == rs->rs_rlt[i].rate)) {
rte = &rs->rs_rlt[i]; rte = &rs->rs_rlt[i];
matched = 1; matched = 1;
if (lower_rate != NULL)
*lower_rate = previous_rate;
break; break;
} else if ((flags & RS_PACING_LT) && } else if ((flags & RS_PACING_LT) &&
(bytes_per_sec <= rs->rs_rlt[i].rate)) { (bytes_per_sec <= rs->rs_rlt[i].rate)) {
rte = &rs->rs_rlt[i]; rte = &rs->rs_rlt[i];
matched = 1; matched = 1;
if (lower_rate != NULL)
*lower_rate = previous_rate;
break; break;
} }
previous_rate = rs->rs_rlt[i].rate;
if (bytes_per_sec > rs->rs_rlt[i].rate) if (bytes_per_sec > rs->rs_rlt[i].rate)
break; break;
} }
Context not available.
* We found a table entry that is smaller, * We found a table entry that is smaller,
* stop there will be none greater or equal. * stop there will be none greater or equal.
*/ */
if (lower_rate != NULL)
*lower_rate = rs->rs_rlt[i].rate;
break; break;
} }
} }
Context not available.
rt_find_real_interface(struct ifnet *ifp, struct inpcb *inp, int *error) rt_find_real_interface(struct ifnet *ifp, struct inpcb *inp, int *error)
{ {
struct ifnet *tifp; struct ifnet *tifp;
struct m_snd_tag *tag; struct m_snd_tag *tag, *ntag;
union if_snd_tag_alloc_params params = { union if_snd_tag_alloc_params params = {
.rate_limit.hdr.type = IF_SND_TAG_TYPE_RATE_LIMIT, .rate_limit.hdr.type = IF_SND_TAG_TYPE_RATE_LIMIT,
.rate_limit.hdr.flowid = 1, .rate_limit.hdr.flowid = inp->inp_flowid,
.rate_limit.hdr.numa_domain = inp->inp_numa_domain, .rate_limit.hdr.numa_domain = inp->inp_numa_domain,
.rate_limit.max_rate = COMMON_RATE, .rate_limit.max_rate = COMMON_RATE,
.rate_limit.flags = M_NOWAIT, .rate_limit.flags = M_NOWAIT,
Context not available.
*error = err; *error = err;
return (NULL); return (NULL);
} }
tifp = tag->ifp; ntag = tag;
while(ntag->ifp->if_next_snd_tag != NULL) {
ntag = ntag->ifp->if_next_snd_tag(ntag);
}
tifp = ntag->ifp;
m_snd_tag_rele(tag); m_snd_tag_rele(tag);
return (tifp); return (tifp);
} }
static void
rl_increment_using(const struct tcp_hwrate_limit_table *rte)
{
struct tcp_hwrate_limit_table *decon_rte;
decon_rte = __DECONST(struct tcp_hwrate_limit_table *, rte);
atomic_add_long(&decon_rte->using, 1);
}
static void
rl_decrement_using(const struct tcp_hwrate_limit_table *rte)
{
struct tcp_hwrate_limit_table *decon_rte;
decon_rte = __DECONST(struct tcp_hwrate_limit_table *, rte);
atomic_subtract_long(&decon_rte->using, 1);
}
void
tcp_rl_log_enobuf(const struct tcp_hwrate_limit_table *rte)
{
struct tcp_hwrate_limit_table *decon_rte;
decon_rte = __DECONST(struct tcp_hwrate_limit_table *, rte);
atomic_add_long(&decon_rte->rs_num_enobufs, 1);
}
/*
* Do NOT take the __noinline out of the
* find_rs_for_ifp() function. If you do the inline
* of it for the rt_setup_rate() will show you a
* compiler bug. For some reason the compiler thinks
* the list can never be empty. The consequence of
* this will be a crash when we dereference NULL
* if an ifp is removed just has a hw rate limit
* is attempted. If you are working on the compiler
* and want to "test" this go ahead and take the noinline
* out otherwise let sleeping dogs ly until such time
* as we get a compiler fix 10/2/20 -- RRS
*/
static __noinline struct tcp_rate_set *
find_rs_for_ifp(struct ifnet *ifp)
{
struct tcp_rate_set *rs;
CK_LIST_FOREACH(rs, &int_rs, next) {
if ((rs->rs_ifp == ifp) &&
(rs->rs_if_dunit == ifp->if_dunit)) {
/* Ok we found it */
return (rs);
}
}
return (NULL);
}
static const struct tcp_hwrate_limit_table * static const struct tcp_hwrate_limit_table *
rt_setup_rate(struct inpcb *inp, struct ifnet *ifp, uint64_t bytes_per_sec, rt_setup_rate(struct inpcb *inp, struct ifnet *ifp, uint64_t bytes_per_sec,
uint32_t flags, int *error) uint32_t flags, int *error, uint64_t *lower_rate)
{ {
/* First lets find the interface if it exists */ /* First lets find the interface if it exists */
const struct tcp_hwrate_limit_table *rte; const struct tcp_hwrate_limit_table *rte;
struct tcp_rate_set *rs; /*
* So why is rs volatile? This is to defeat a
* compiler bug where in the compiler is convinced
* that rs can never be NULL (which is not true). Because
* of its conviction it nicely optimizes out the if ((rs == NULL
* below which means if you get a NULL back you dereference it.
*/
volatile struct tcp_rate_set *rs;
struct epoch_tracker et; struct epoch_tracker et;
struct ifnet *oifp = ifp;
int err; int err;
NET_EPOCH_ENTER(et); NET_EPOCH_ENTER(et);
use_real_interface: use_real_interface:
CK_LIST_FOREACH(rs, &int_rs, next) { rs = find_rs_for_ifp(ifp);
/*
* Note we don't look with the lock since we either see a
* new entry or will get one when we try to add it.
*/
if (rs->rs_flags & RS_IS_DEAD) {
/* The dead are not looked at */
continue;
}
if ((rs->rs_ifp == ifp) &&
(rs->rs_if_dunit == ifp->if_dunit)) {
/* Ok we found it */
break;
}
}
if ((rs == NULL) || if ((rs == NULL) ||
(rs->rs_flags & RS_INTF_NO_SUP) || (rs->rs_flags & RS_INTF_NO_SUP) ||
(rs->rs_flags & RS_IS_DEAD)) { (rs->rs_flags & RS_IS_DEAD)) {
Context not available.
* might be arguable, but its impossible * might be arguable, but its impossible
* to tell from the departing case. * to tell from the departing case.
*/ */
if (rs->rs_disable && error) if (error)
*error = ENODEV; *error = ENODEV;
NET_EPOCH_EXIT(et); NET_EPOCH_EXIT(et);
return (NULL); return (NULL);
} }
if ((rs == NULL) || (rs->rs_disable != 0)) { if ((rs == NULL) || (rs->rs_disable != 0)) {
if (rs->rs_disable && error) if (error)
*error = ENOSPC; *error = ENOSPC;
NET_EPOCH_EXIT(et); NET_EPOCH_EXIT(et);
return (NULL); return (NULL);
Context not available.
NET_EPOCH_EXIT(et); NET_EPOCH_EXIT(et);
return (NULL); return (NULL);
} }
KASSERT((tifp != ifp),
("Lookup failure ifp:%p inp:%p rt_find_real_interface() returns the same interface tifp:%p?\n",
ifp, inp, tifp));
ifp = tifp;
goto use_real_interface; goto use_real_interface;
} }
if (rs->rs_flow_limit && if (rs->rs_flow_limit &&
Context not available.
NET_EPOCH_EXIT(et); NET_EPOCH_EXIT(et);
return (NULL); return (NULL);
} }
rte = tcp_find_suitable_rate(rs, bytes_per_sec, flags); rte = tcp_find_suitable_rate(rs, bytes_per_sec, flags, lower_rate);
if (rte) { if (rte) {
err = in_pcbattach_txrtlmt(inp, rs->rs_ifp, err = in_pcbattach_txrtlmt(inp, oifp,
inp->inp_flowtype, inp->inp_flowtype,
inp->inp_flowid, inp->inp_flowid,
rte->rate, rte->rate,
Context not available.
if (error) if (error)
*error = err; *error = err;
rte = NULL; rte = NULL;
} else {
KASSERT((inp->inp_snd_tag != NULL) ,
("Setup rate has no snd_tag inp:%p rte:%p rate:%lu rs:%p",
inp, rte, rte->rate, rs));
#ifdef NETFLIX_STATS
counter_u64_add(rate_limit_new, 1);
#endif
} }
} }
if (rte) { if (rte) {
Context not available.
{ {
int error; int error;
struct tcp_rate_set *rs; struct tcp_rate_set *rs;
struct epoch_tracker et;
if (((ifp->if_capenable & IFCAP_TXRTLMT) == 0) || if (((ifp->if_capenable & IFCAP_TXRTLMT) == 0) ||
(link_state != LINK_STATE_UP)) { (link_state != LINK_STATE_UP)) {
Context not available.
*/ */
return; return;
} }
NET_EPOCH_ENTER(et);
mtx_lock(&rs_mtx); mtx_lock(&rs_mtx);
CK_LIST_FOREACH(rs, &int_rs, next) { rs = find_rs_for_ifp(ifp);
if ((rs->rs_ifp == ifp) && if (rs) {
(rs->rs_if_dunit == ifp->if_dunit)) { /* We already have initialized this guy */
/* We already have initialized this guy */ mtx_unlock(&rs_mtx);
mtx_unlock(&rs_mtx); NET_EPOCH_EXIT(et);
return; return;
}
} }
mtx_unlock(&rs_mtx); mtx_unlock(&rs_mtx);
NET_EPOCH_EXIT(et);
rt_setup_new_rs(ifp, &error); rt_setup_new_rs(ifp, &error);
} }
static void static void
tcp_rl_ifnet_departure(void *arg __unused, struct ifnet *ifp) tcp_rl_ifnet_departure(void *arg __unused, struct ifnet *ifp)
{ {
struct tcp_rate_set *rs, *nrs; struct tcp_rate_set *rs;
struct epoch_tracker et;
int i; int i;
NET_EPOCH_ENTER(et);
mtx_lock(&rs_mtx); mtx_lock(&rs_mtx);
CK_LIST_FOREACH_SAFE(rs, &int_rs, next, nrs) { rs = find_rs_for_ifp(ifp);
if ((rs->rs_ifp == ifp) && if (rs) {
(rs->rs_if_dunit == ifp->if_dunit)) { CK_LIST_REMOVE(rs, next);
CK_LIST_REMOVE(rs, next); rs_number_alive--;
rs_number_alive--; rs->rs_flags |= RS_IS_DEAD;
rs->rs_flags |= RS_IS_DEAD; for (i = 0; i < rs->rs_rate_cnt; i++) {
for (i = 0; i < rs->rs_rate_cnt; i++) { if (rs->rs_rlt[i].flags & HDWRPACE_TAGPRESENT) {
if (rs->rs_rlt[i].flags & HDWRPACE_TAGPRESENT) { in_pcbdetach_tag(rs->rs_rlt[i].tag);
in_pcbdetach_tag(rs->rs_rlt[i].tag); rs->rs_rlt[i].tag = NULL;
rs->rs_rlt[i].tag = NULL;
}
rs->rs_rlt[i].flags = HDWRPACE_IFPDEPARTED;
} }
if (rs->rs_flows_using == 0) rs->rs_rlt[i].flags = HDWRPACE_IFPDEPARTED;
rs_defer_destroy(rs);
break;
} }
if (rs->rs_flows_using == 0)
rs_defer_destroy(rs);
} }
mtx_unlock(&rs_mtx); mtx_unlock(&rs_mtx);
NET_EPOCH_EXIT(et);
} }
static void static void
tcp_rl_shutdown(void *arg __unused, int howto __unused) tcp_rl_shutdown(void *arg __unused, int howto __unused)
{ {
struct tcp_rate_set *rs, *nrs; struct tcp_rate_set *rs, *nrs;
struct epoch_tracker et;
int i; int i;
NET_EPOCH_ENTER(et);
mtx_lock(&rs_mtx); mtx_lock(&rs_mtx);
CK_LIST_FOREACH_SAFE(rs, &int_rs, next, nrs) { CK_LIST_FOREACH_SAFE(rs, &int_rs, next, nrs) {
CK_LIST_REMOVE(rs, next); CK_LIST_REMOVE(rs, next);
Context not available.
rs_defer_destroy(rs); rs_defer_destroy(rs);
} }
mtx_unlock(&rs_mtx); mtx_unlock(&rs_mtx);
NET_EPOCH_EXIT(et);
} }
const struct tcp_hwrate_limit_table * const struct tcp_hwrate_limit_table *
tcp_set_pacing_rate(struct tcpcb *tp, struct ifnet *ifp, tcp_set_pacing_rate(struct tcpcb *tp, struct ifnet *ifp,
uint64_t bytes_per_sec, int flags, int *error) uint64_t bytes_per_sec, int flags, int *error, uint64_t *lower_rate)
{ {
const struct tcp_hwrate_limit_table *rte; const struct tcp_hwrate_limit_table *rte;
#ifdef KERN_TLS #ifdef KERN_TLS
Context not available.
} }
} }
#endif #endif
rte = rt_setup_rate(tp->t_inpcb, ifp, bytes_per_sec, flags, error); rte = rt_setup_rate(tp->t_inpcb, ifp, bytes_per_sec, flags, error, lower_rate);
if (rte)
rl_increment_using(rte);
#ifdef KERN_TLS #ifdef KERN_TLS
if (rte != NULL && tls != NULL && tls->snd_tag != NULL) { if (rte != NULL && tls != NULL && tls->snd_tag != NULL) {
/* /*
Context not available.
*error = EINVAL; *error = EINVAL;
rte = NULL; rte = NULL;
} }
tp->t_pacing_rate = rte->rate; if (rte != NULL) {
*error = 0; tp->t_pacing_rate = rte->rate;
*error = 0;
}
return (rte); return (rte);
} }
const struct tcp_hwrate_limit_table * const struct tcp_hwrate_limit_table *
tcp_chg_pacing_rate(const struct tcp_hwrate_limit_table *crte, tcp_chg_pacing_rate(const struct tcp_hwrate_limit_table *crte,
struct tcpcb *tp, struct ifnet *ifp, struct tcpcb *tp, struct ifnet *ifp,
uint64_t bytes_per_sec, int flags, int *error) uint64_t bytes_per_sec, int flags, int *error, uint64_t *lower_rate)
{ {
const struct tcp_hwrate_limit_table *nrte; const struct tcp_hwrate_limit_table *nrte;
const struct tcp_rate_set *rs; const struct tcp_rate_set *rs;
#ifdef KERN_TLS #ifdef KERN_TLS
struct ktls_session *tls = NULL; struct ktls_session *tls = NULL;
#endif #endif
int is_indirect = 0;
int err; int err;
INP_WLOCK_ASSERT(tp->t_inpcb); INP_WLOCK_ASSERT(tp->t_inpcb);
Context not available.
if ((rs->rs_flags & RS_IS_DEAD) || if ((rs->rs_flags & RS_IS_DEAD) ||
(crte->flags & HDWRPACE_IFPDEPARTED)) { (crte->flags & HDWRPACE_IFPDEPARTED)) {
/* Release the rate, and try anew */ /* Release the rate, and try anew */
re_rate:
tcp_rel_pacing_rate(crte, tp); tcp_rel_pacing_rate(crte, tp);
nrte = tcp_set_pacing_rate(tp, ifp, nrte = tcp_set_pacing_rate(tp, ifp,
bytes_per_sec, flags, error); bytes_per_sec, flags, error, lower_rate);
return (nrte); return (nrte);
} }
if ((rs->rs_flags & RT_IS_INDIRECT ) == RT_IS_INDIRECT) nrte = tcp_find_suitable_rate(rs, bytes_per_sec, flags, lower_rate);
is_indirect = 1;
else
is_indirect = 0;
if ((is_indirect == 0) &&
((ifp != rs->rs_ifp) ||
(ifp->if_dunit != rs->rs_if_dunit))) {
/*
* Something changed, the user is not pointing to the same
* ifp? Maybe a route updated on this guy?
*/
goto re_rate;
} else if (is_indirect) {
/*
* For indirect we have to dig in and find the real interface.
*/
struct ifnet *rifp;
rifp = rt_find_real_interface(ifp, tp->t_inpcb, error);
if (rifp == NULL) {
/* Can't find it? */
goto re_rate;
}
if ((rifp != rs->rs_ifp) ||
(ifp->if_dunit != rs->rs_if_dunit)) {
goto re_rate;
}
}
nrte = tcp_find_suitable_rate(rs, bytes_per_sec, flags);
if (nrte == crte) { if (nrte == crte) {
/* No change */ /* No change */
if (error) if (error)
Context not available.
} }
if (nrte == NULL) { if (nrte == NULL) {
/* Release the old rate */ /* Release the old rate */
if (error)
*error = ENOENT;
tcp_rel_pacing_rate(crte, tp); tcp_rel_pacing_rate(crte, tp);
return (NULL); return (NULL);
} }
rl_decrement_using(crte);
rl_increment_using(nrte);
/* Change rates to our new entry */ /* Change rates to our new entry */
#ifdef KERN_TLS #ifdef KERN_TLS
if (tls != NULL) if (tls != NULL)
Context not available.
#endif #endif
err = in_pcbmodify_txrtlmt(tp->t_inpcb, nrte->rate); err = in_pcbmodify_txrtlmt(tp->t_inpcb, nrte->rate);
if (err) { if (err) {
rl_decrement_using(nrte);
/* Do we still have a snd-tag attached? */
if (tp->t_inpcb->inp_snd_tag)
in_pcbdetach_txrtlmt(tp->t_inpcb);
if (error) if (error)
*error = err; *error = err;
return (NULL); return (NULL);
} }
#ifdef NETFLIX_STATS
else
counter_u64_add(rate_limit_chg, 1);
#endif
if (error) if (error)
*error = 0; *error = 0;
tp->t_pacing_rate = nrte->rate; tp->t_pacing_rate = nrte->rate;
Context not available.
* in order to release our refcount. * in order to release our refcount.
*/ */
rs = __DECONST(struct tcp_rate_set *, crs); rs = __DECONST(struct tcp_rate_set *, crs);
rl_decrement_using(crte);
pre = atomic_fetchadd_64(&rs->rs_flows_using, -1); pre = atomic_fetchadd_64(&rs->rs_flows_using, -1);
if (pre == 1) { if (pre == 1) {
struct epoch_tracker et;
NET_EPOCH_ENTER(et);
mtx_lock(&rs_mtx); mtx_lock(&rs_mtx);
/* /*
* Is it dead? * Is it dead?
Context not available.
if (rs->rs_flags & RS_IS_DEAD) if (rs->rs_flags & RS_IS_DEAD)
rs_defer_destroy(rs); rs_defer_destroy(rs);
mtx_unlock(&rs_mtx); mtx_unlock(&rs_mtx);
NET_EPOCH_EXIT(et);
} }
/* /*
Context not available.
#define FIVE_HUNDRED_MBPS 62500000 /* 500Mbps in bytes per second */ #define FIVE_HUNDRED_MBPS 62500000 /* 500Mbps in bytes per second */
#define MAX_MSS_SENT 43 /* 43 mss = 43 x 1500 = 64,500 bytes */ #define MAX_MSS_SENT 43 /* 43 mss = 43 x 1500 = 64,500 bytes */
static void
tcp_log_pacing_size(struct tcpcb *tp, uint64_t bw, uint32_t segsiz, uint32_t new_tso,
uint64_t hw_rate, uint32_t time_between, uint32_t calc_time_between,
uint32_t segs, uint32_t res_div, uint16_t mult, uint8_t mod)
{
if (tp->t_logstate != TCP_LOG_STATE_OFF) {
union tcp_log_stackspecific log;
struct timeval tv;
uint32_t cts;
memset(&log, 0, sizeof(log));
cts = tcp_get_usecs(&tv);
log.u_bbr.flex1 = segsiz;
log.u_bbr.flex2 = new_tso;
log.u_bbr.flex3 = time_between;
log.u_bbr.flex4 = calc_time_between;
log.u_bbr.flex5 = segs;
log.u_bbr.flex6 = res_div;
log.u_bbr.flex7 = mult;
log.u_bbr.flex8 = mod;
log.u_bbr.timeStamp = tcp_get_usecs(&tv);
log.u_bbr.cur_del_rate = bw;
log.u_bbr.delRate = hw_rate;
TCP_LOG_EVENTP(tp, NULL,
&tp->t_inpcb->inp_socket->so_rcv,
&tp->t_inpcb->inp_socket->so_snd,
TCP_HDWR_PACE_SIZE, 0,
0, &log, false, &tv);
}
}
uint32_t uint32_t
tcp_get_pacing_burst_size (uint64_t bw, uint32_t segsiz, int can_use_1mss, tcp_get_pacing_burst_size (struct tcpcb *tp, uint64_t bw, uint32_t segsiz, int can_use_1mss,
const struct tcp_hwrate_limit_table *te, int *err) const struct tcp_hwrate_limit_table *te, int *err)
{ {
/* /*
Context not available.
min_tso_segs = 1; min_tso_segs = 1;
else else
min_tso_segs = 2; min_tso_segs = 2;
if (new_tso < min_tso_segs) if (rs_floor_mss && (new_tso < rs_floor_mss))
new_tso = rs_floor_mss;
else if (new_tso < min_tso_segs)
new_tso = min_tso_segs; new_tso = min_tso_segs;
if (new_tso > MAX_MSS_SENT) if (new_tso > MAX_MSS_SENT)
new_tso = MAX_MSS_SENT; new_tso = MAX_MSS_SENT;
new_tso *= segsiz; new_tso *= segsiz;
tcp_log_pacing_size(tp, bw, segsiz, new_tso,
0, 0, 0, 0, 0, 0, 1);
/* /*
* If we are not doing hardware pacing * If we are not doing hardware pacing
* then we are done. * then we are done.
Context not available.
* max (43 segments). * max (43 segments).
*/ */
if (te->rate > FIVE_HUNDRED_MBPS) if (te->rate > FIVE_HUNDRED_MBPS)
return (segsiz * MAX_MSS_SENT); goto max;
if (te->rate == bw) { if (te->rate == bw) {
/* We are pacing at exactly the hdwr rate */ /* We are pacing at exactly the hdwr rate */
max:
tcp_log_pacing_size(tp, bw, segsiz, new_tso,
te->rate, te->time_between, (uint32_t)0,
(segsiz * MAX_MSS_SENT), 0, 0, 3);
return (segsiz * MAX_MSS_SENT); return (segsiz * MAX_MSS_SENT);
} }
lentim = ETHERNET_SEGMENT_SIZE * USECS_IN_SECOND; lentim = ETHERNET_SEGMENT_SIZE * USECS_IN_SECOND;
res = lentim / bw; res = lentim / bw;
if (res > te->time_between) { if (res > te->time_between) {
uint32_t delta, segs; uint32_t delta, segs, res_div;
res_div = ((res * num_of_waits_allowed) + wait_time_floor);
delta = res - te->time_between; delta = res - te->time_between;
segs = (res + delta - 1)/delta; segs = (res_div + delta - 1)/delta;
if (te->rate > ONE_HUNDRED_MBPS)
segs *= 2;
if (segs < min_tso_segs) if (segs < min_tso_segs)
segs = min_tso_segs; segs = min_tso_segs;
if (segs < rs_hw_floor_mss)
segs = rs_hw_floor_mss;
if (segs > MAX_MSS_SENT) if (segs > MAX_MSS_SENT)
segs = MAX_MSS_SENT; segs = MAX_MSS_SENT;
segs *= segsiz; segs *= segsiz;
tcp_log_pacing_size(tp, bw, segsiz, new_tso,
te->rate, te->time_between, (uint32_t)res,
segs, res_div, 1, 3);
if (err) if (err)
*err = 0; *err = 0;
if (segs < new_tso) { if (segs < new_tso) {
Context not available.
* hardware. Send back the non-hardware * hardware. Send back the non-hardware
* rate. * rate.
*/ */
tcp_log_pacing_size(tp, bw, segsiz, new_tso,
te->rate, te->time_between, (uint32_t)res,
0, 0, 0, 4);
if (err) if (err)
*err = -1; *err = -1;
return (new_tso); return (new_tso);
} }
} }
uint64_t
tcp_hw_highest_rate_ifp(struct ifnet *ifp, struct inpcb *inp)
{
struct epoch_tracker et;
struct tcp_rate_set *rs;
uint64_t rate_ret;
NET_EPOCH_ENTER(et);
use_next_interface:
rs = find_rs_for_ifp(ifp);
if (rs == NULL) {
/* This interface does not do ratelimiting */
rate_ret = 0;
} else if (rs->rs_flags & RS_IS_DEFF) {
/* We need to find the real interface */
struct ifnet *tifp;
tifp = rt_find_real_interface(ifp, inp, NULL);
if (tifp == NULL) {
NET_EPOCH_EXIT(et);
return (0);
}
ifp = tifp;
goto use_next_interface;
} else {
/* Lets return the highest rate this guy has */
rate_ret = rs->rs_rlt[rs->rs_highest_valid].rate;
}
NET_EPOCH_EXIT(et);
return(rate_ret);
}
static eventhandler_tag rl_ifnet_departs; static eventhandler_tag rl_ifnet_departs;
static eventhandler_tag rl_ifnet_arrives; static eventhandler_tag rl_ifnet_arrives;
static eventhandler_tag rl_shutdown_start; static eventhandler_tag rl_shutdown_start;
Context not available.