diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c index b9f429ec0db4..670a43fe2a00 100644 --- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c +++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c @@ -323,7 +323,7 @@ mlxsw_sp_qdisc_get_tc_stats(struct mlxsw_sp_port *mlxsw_sp_port, static int mlxsw_sp_tclass_congestion_enable(struct mlxsw_sp_port *mlxsw_sp_port, int tclass_num, u32 min, u32 max, - u32 probability, bool is_ecn) + u32 probability, bool is_wred, bool is_ecn) { char cwtpm_cmd[MLXSW_REG_CWTPM_LEN]; char cwtp_cmd[MLXSW_REG_CWTP_LEN]; @@ -341,7 +341,7 @@ mlxsw_sp_tclass_congestion_enable(struct mlxsw_sp_port *mlxsw_sp_port, return err; mlxsw_reg_cwtpm_pack(cwtpm_cmd, mlxsw_sp_port->local_port, tclass_num, - MLXSW_REG_CWTP_DEFAULT_PROFILE, true, is_ecn); + MLXSW_REG_CWTP_DEFAULT_PROFILE, is_wred, is_ecn); return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(cwtpm), cwtpm_cmd); } @@ -445,8 +445,9 @@ mlxsw_sp_qdisc_red_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle, prob = DIV_ROUND_UP(prob, 1 << 16); min = mlxsw_sp_bytes_cells(mlxsw_sp, p->min); max = mlxsw_sp_bytes_cells(mlxsw_sp, p->max); - return mlxsw_sp_tclass_congestion_enable(mlxsw_sp_port, tclass_num, min, - max, prob, p->is_ecn); + return mlxsw_sp_tclass_congestion_enable(mlxsw_sp_port, tclass_num, + min, max, prob, + !p->is_nodrop, p->is_ecn); } static void diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h index dbc89452f90b..1db8b27d4515 100644 --- a/include/net/pkt_cls.h +++ b/include/net/pkt_cls.h @@ -740,6 +740,7 @@ struct tc_red_qopt_offload_params { u32 limit; bool is_ecn; bool is_harddrop; + bool is_nodrop; struct gnet_stats_queue *qstats; }; diff --git a/include/net/red.h b/include/net/red.h index 9665582c4687..fc455445f4b2 100644 --- a/include/net/red.h +++ b/include/net/red.h @@ -179,6 +179,44 @@ static inline bool red_check_params(u32 qth_min, u32 qth_max, u8 Wlog) return true; } +static inline int red_get_flags(unsigned char qopt_flags, + unsigned char historic_mask, + struct nlattr *flags_attr, + unsigned char supported_mask, + struct nla_bitfield32 *p_flags, + unsigned char *p_userbits, + struct netlink_ext_ack *extack) +{ + struct nla_bitfield32 flags; + + if (qopt_flags && flags_attr) { + NL_SET_ERR_MSG_MOD(extack, "flags should be passed either through qopt, or through a dedicated attribute"); + return -EINVAL; + } + + if (flags_attr) { + flags = nla_get_bitfield32(flags_attr); + } else { + flags.selector = historic_mask; + flags.value = qopt_flags & historic_mask; + } + + *p_flags = flags; + *p_userbits = qopt_flags & ~historic_mask; + return 0; +} + +static inline int red_validate_flags(unsigned char flags, + struct netlink_ext_ack *extack) +{ + if ((flags & TC_RED_NODROP) && !(flags & TC_RED_ECN)) { + NL_SET_ERR_MSG_MOD(extack, "nodrop mode is only meaningful with ECN"); + return -EINVAL; + } + + return 0; +} + static inline void red_set_parms(struct red_parms *p, u32 qth_min, u32 qth_max, u8 Wlog, u8 Plog, u8 Scell_log, u8 *stab, u32 max_P) diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h index bbe791b24168..ea39287d59c8 100644 --- a/include/uapi/linux/pkt_sched.h +++ b/include/uapi/linux/pkt_sched.h @@ -256,6 +256,7 @@ enum { TCA_RED_PARMS, TCA_RED_STAB, TCA_RED_MAX_P, + TCA_RED_FLAGS, /* bitfield32 */ __TCA_RED_MAX, }; @@ -268,12 +269,28 @@ struct tc_red_qopt { unsigned char Wlog; /* log(W) */ unsigned char Plog; /* log(P_max/(qth_max-qth_min)) */ unsigned char Scell_log; /* cell size for idle damping */ + + /* This field can be used for flags that a RED-like qdisc has + * historically supported. E.g. when configuring RED, it can be used for + * ECN, HARDDROP and ADAPTATIVE. For SFQ it can be used for ECN, + * HARDDROP. Etc. Because this field has not been validated, and is + * copied back on dump, any bits besides those to which a given qdisc + * has assigned a historical meaning need to be considered for free use + * by userspace tools. + * + * Any further flags need to be passed differently, e.g. through an + * attribute (such as TCA_RED_FLAGS above). Such attribute should allow + * passing both recent and historic flags in one value. + */ unsigned char flags; #define TC_RED_ECN 1 #define TC_RED_HARDDROP 2 #define TC_RED_ADAPTATIVE 4 +#define TC_RED_NODROP 8 }; +#define TC_RED_HISTORIC_FLAGS (TC_RED_ECN | TC_RED_HARDDROP | TC_RED_ADAPTATIVE) + struct tc_red_xstats { __u32 early; /* Early drops */ __u32 pdrop; /* Drops due to queue limits */ diff --git a/net/sched/sch_red.c b/net/sched/sch_red.c index 1695421333e3..3ef0a4f7399b 100644 --- a/net/sched/sch_red.c +++ b/net/sched/sch_red.c @@ -35,7 +35,11 @@ struct red_sched_data { u32 limit; /* HARD maximal queue length */ + unsigned char flags; + /* Non-flags in tc_red_qopt.flags. */ + unsigned char userbits; + struct timer_list adapt_timer; struct Qdisc *sch; struct red_parms parms; @@ -44,6 +48,8 @@ struct red_sched_data { struct Qdisc *qdisc; }; +static const u32 red_supported_flags = TC_RED_HISTORIC_FLAGS | TC_RED_NODROP; + static inline int red_use_ecn(struct red_sched_data *q) { return q->flags & TC_RED_ECN; @@ -54,6 +60,11 @@ static inline int red_use_harddrop(struct red_sched_data *q) return q->flags & TC_RED_HARDDROP; } +static int red_use_nodrop(struct red_sched_data *q) +{ + return q->flags & TC_RED_NODROP; +} + static int red_enqueue(struct sk_buff *skb, struct Qdisc *sch, struct sk_buff **to_free) { @@ -74,23 +85,36 @@ static int red_enqueue(struct sk_buff *skb, struct Qdisc *sch, case RED_PROB_MARK: qdisc_qstats_overlimit(sch); - if (!red_use_ecn(q) || !INET_ECN_set_ce(skb)) { + if (!red_use_ecn(q)) { q->stats.prob_drop++; goto congestion_drop; } - q->stats.prob_mark++; + if (INET_ECN_set_ce(skb)) { + q->stats.prob_mark++; + } else if (!red_use_nodrop(q)) { + q->stats.prob_drop++; + goto congestion_drop; + } + + /* Non-ECT packet in ECN nodrop mode: queue it. */ break; case RED_HARD_MARK: qdisc_qstats_overlimit(sch); - if (red_use_harddrop(q) || !red_use_ecn(q) || - !INET_ECN_set_ce(skb)) { + if (red_use_harddrop(q) || !red_use_ecn(q)) { q->stats.forced_drop++; goto congestion_drop; } - q->stats.forced_mark++; + if (INET_ECN_set_ce(skb)) { + q->stats.forced_mark++; + } else if (!red_use_nodrop(q)) { + q->stats.forced_drop++; + goto congestion_drop; + } + + /* Non-ECT packet in ECN nodrop mode: queue it. */ break; } @@ -165,6 +189,7 @@ static int red_offload(struct Qdisc *sch, bool enable) opt.set.limit = q->limit; opt.set.is_ecn = red_use_ecn(q); opt.set.is_harddrop = red_use_harddrop(q); + opt.set.is_nodrop = red_use_nodrop(q); opt.set.qstats = &sch->qstats; } else { opt.command = TC_RED_DESTROY; @@ -183,9 +208,12 @@ static void red_destroy(struct Qdisc *sch) } static const struct nla_policy red_policy[TCA_RED_MAX + 1] = { + [TCA_RED_UNSPEC] = { .strict_start_type = TCA_RED_FLAGS }, [TCA_RED_PARMS] = { .len = sizeof(struct tc_red_qopt) }, [TCA_RED_STAB] = { .len = RED_STAB_SIZE }, [TCA_RED_MAX_P] = { .type = NLA_U32 }, + [TCA_RED_FLAGS] = { .type = NLA_BITFIELD32, + .validation_data = &red_supported_flags }, }; static int red_change(struct Qdisc *sch, struct nlattr *opt, @@ -194,7 +222,10 @@ static int red_change(struct Qdisc *sch, struct nlattr *opt, struct Qdisc *old_child = NULL, *child = NULL; struct red_sched_data *q = qdisc_priv(sch); struct nlattr *tb[TCA_RED_MAX + 1]; + struct nla_bitfield32 flags_bf; struct tc_red_qopt *ctl; + unsigned char userbits; + unsigned char flags; int err; u32 max_P; @@ -216,6 +247,12 @@ static int red_change(struct Qdisc *sch, struct nlattr *opt, if (!red_check_params(ctl->qth_min, ctl->qth_max, ctl->Wlog)) return -EINVAL; + err = red_get_flags(ctl->flags, TC_RED_HISTORIC_FLAGS, + tb[TCA_RED_FLAGS], red_supported_flags, + &flags_bf, &userbits, extack); + if (err) + return err; + if (ctl->limit > 0) { child = fifo_create_dflt(sch, &bfifo_qdisc_ops, ctl->limit, extack); @@ -227,7 +264,14 @@ static int red_change(struct Qdisc *sch, struct nlattr *opt, } sch_tree_lock(sch); - q->flags = ctl->flags; + + flags = (q->flags & ~flags_bf.selector) | flags_bf.value; + err = red_validate_flags(flags, extack); + if (err) + goto unlock_out; + + q->flags = flags; + q->userbits = userbits; q->limit = ctl->limit; if (child) { qdisc_tree_flush_backlog(q->qdisc); @@ -256,6 +300,12 @@ static int red_change(struct Qdisc *sch, struct nlattr *opt, if (old_child) qdisc_put(old_child); return 0; + +unlock_out: + sch_tree_unlock(sch); + if (child) + qdisc_put(child); + return err; } static inline void red_adaptative_timer(struct timer_list *t) @@ -299,10 +349,15 @@ static int red_dump_offload_stats(struct Qdisc *sch) static int red_dump(struct Qdisc *sch, struct sk_buff *skb) { struct red_sched_data *q = qdisc_priv(sch); + struct nla_bitfield32 flags_bf = { + .selector = red_supported_flags, + .value = q->flags, + }; struct nlattr *opts = NULL; struct tc_red_qopt opt = { .limit = q->limit, - .flags = q->flags, + .flags = (q->flags & TC_RED_HISTORIC_FLAGS) | + q->userbits, .qth_min = q->parms.qth_min >> q->parms.Wlog, .qth_max = q->parms.qth_max >> q->parms.Wlog, .Wlog = q->parms.Wlog, @@ -319,7 +374,8 @@ static int red_dump(struct Qdisc *sch, struct sk_buff *skb) if (opts == NULL) goto nla_put_failure; if (nla_put(skb, TCA_RED_PARMS, sizeof(opt), &opt) || - nla_put_u32(skb, TCA_RED_MAX_P, q->parms.max_P)) + nla_put_u32(skb, TCA_RED_MAX_P, q->parms.max_P) || + nla_put(skb, TCA_RED_FLAGS, sizeof(flags_bf), &flags_bf)) goto nla_put_failure; return nla_nest_end(skb, opts); diff --git a/tools/testing/selftests/drivers/net/mlxsw/sch_red_core.sh b/tools/testing/selftests/drivers/net/mlxsw/sch_red_core.sh index 8f833678ac4d..0d347d48c112 100644 --- a/tools/testing/selftests/drivers/net/mlxsw/sch_red_core.sh +++ b/tools/testing/selftests/drivers/net/mlxsw/sch_red_core.sh @@ -389,17 +389,14 @@ check_marking() ((pct $cond)) } -do_ecn_test() +ecn_test_common() { + local name=$1; shift local vlan=$1; shift local limit=$1; shift local backlog local pct - # Main stream. - start_tcp_traffic $h1.$vlan $(ipaddr 1 $vlan) $(ipaddr 3 $vlan) \ - $h3_mac tos=0x01 - # Build the below-the-limit backlog using UDP. We could use TCP just # fine, but this way we get a proof that UDP is accepted when queue # length is below the limit. The main stream is using TCP, and if the @@ -409,7 +406,7 @@ do_ecn_test() check_err $? "Could not build the requested backlog" pct=$(check_marking $vlan "== 0") check_err $? "backlog $backlog / $limit Got $pct% marked packets, expected == 0." - log_test "TC $((vlan - 10)): ECN backlog < limit" + log_test "TC $((vlan - 10)): $name backlog < limit" # Now push TCP, because non-TCP traffic would be early-dropped after the # backlog crosses the limit, and we want to make sure that the backlog @@ -419,7 +416,20 @@ do_ecn_test() check_err $? "Could not build the requested backlog" pct=$(check_marking $vlan ">= 95") check_err $? "backlog $backlog / $limit Got $pct% marked packets, expected >= 95." - log_test "TC $((vlan - 10)): ECN backlog > limit" + log_test "TC $((vlan - 10)): $name backlog > limit" +} + +do_ecn_test() +{ + local vlan=$1; shift + local limit=$1; shift + local name=ECN + + start_tcp_traffic $h1.$vlan $(ipaddr 1 $vlan) $(ipaddr 3 $vlan) \ + $h3_mac tos=0x01 + sleep 1 + + ecn_test_common "$name" $vlan $limit # Up there we saw that UDP gets accepted when backlog is below the # limit. Now that it is above, it should all get dropped, and backlog @@ -427,7 +437,31 @@ do_ecn_test() RET=0 build_backlog $vlan $((2 * limit)) udp >/dev/null check_fail $? "UDP traffic went into backlog instead of being early-dropped" - log_test "TC $((vlan - 10)): ECN backlog > limit: UDP early-dropped" + log_test "TC $((vlan - 10)): $name backlog > limit: UDP early-dropped" + + stop_traffic + sleep 1 +} + +do_ecn_nodrop_test() +{ + local vlan=$1; shift + local limit=$1; shift + local name="ECN nodrop" + + start_tcp_traffic $h1.$vlan $(ipaddr 1 $vlan) $(ipaddr 3 $vlan) \ + $h3_mac tos=0x01 + sleep 1 + + ecn_test_common "$name" $vlan $limit + + # Up there we saw that UDP gets accepted when backlog is below the + # limit. Now that it is above, in nodrop mode, make sure it goes to + # backlog as well. + RET=0 + build_backlog $vlan $((2 * limit)) udp >/dev/null + check_err $? "UDP traffic was early-dropped instead of getting into backlog" + log_test "TC $((vlan - 10)): $name backlog > limit: UDP not dropped" stop_traffic sleep 1 diff --git a/tools/testing/selftests/drivers/net/mlxsw/sch_red_ets.sh b/tools/testing/selftests/drivers/net/mlxsw/sch_red_ets.sh index af83efe9ccf1..1c36c576613b 100755 --- a/tools/testing/selftests/drivers/net/mlxsw/sch_red_ets.sh +++ b/tools/testing/selftests/drivers/net/mlxsw/sch_red_ets.sh @@ -4,6 +4,7 @@ ALL_TESTS=" ping_ipv4 ecn_test + ecn_nodrop_test red_test mc_backlog_test " @@ -50,6 +51,16 @@ ecn_test() uninstall_qdisc } +ecn_nodrop_test() +{ + install_qdisc ecn nodrop + + do_ecn_nodrop_test 10 $BACKLOG1 + do_ecn_nodrop_test 11 $BACKLOG2 + + uninstall_qdisc +} + red_test() { install_qdisc diff --git a/tools/testing/selftests/drivers/net/mlxsw/sch_red_root.sh b/tools/testing/selftests/drivers/net/mlxsw/sch_red_root.sh index b2217493a88e..558667ea11ec 100755 --- a/tools/testing/selftests/drivers/net/mlxsw/sch_red_root.sh +++ b/tools/testing/selftests/drivers/net/mlxsw/sch_red_root.sh @@ -4,6 +4,7 @@ ALL_TESTS=" ping_ipv4 ecn_test + ecn_nodrop_test red_test mc_backlog_test " @@ -33,6 +34,13 @@ ecn_test() uninstall_qdisc } +ecn_nodrop_test() +{ + install_qdisc ecn nodrop + do_ecn_nodrop_test 10 $BACKLOG + uninstall_qdisc +} + red_test() { install_qdisc diff --git a/tools/testing/selftests/tc-testing/tc-tests/qdiscs/red.json b/tools/testing/selftests/tc-testing/tc-tests/qdiscs/red.json new file mode 100644 index 000000000000..0703a2a255eb --- /dev/null +++ b/tools/testing/selftests/tc-testing/tc-tests/qdiscs/red.json @@ -0,0 +1,185 @@ +[ + { + "id": "8b6e", + "name": "Create RED with no flags", + "category": [ + "qdisc", + "red" + ], + "plugins": { + "requires": "nsPlugin" + }, + "setup": [ + "$IP link add dev $DUMMY type dummy || /bin/true" + ], + "cmdUnderTest": "$TC qdisc add dev $DUMMY handle 1: root red limit 1M avpkt 1500 min 100K max 300K", + "expExitCode": "0", + "verifyCmd": "$TC qdisc show dev $DUMMY", + "matchPattern": "qdisc red 1: root .* limit 1Mb min 100Kb max 300Kb $", + "matchCount": "1", + "teardown": [ + "$TC qdisc del dev $DUMMY handle 1: root", + "$IP link del dev $DUMMY type dummy" + ] + }, + { + "id": "342e", + "name": "Create RED with adaptive flag", + "category": [ + "qdisc", + "red" + ], + "plugins": { + "requires": "nsPlugin" + }, + "setup": [ + "$IP link add dev $DUMMY type dummy || /bin/true" + ], + "cmdUnderTest": "$TC qdisc add dev $DUMMY handle 1: root red adaptive limit 1M avpkt 1500 min 100K max 300K", + "expExitCode": "0", + "verifyCmd": "$TC qdisc show dev $DUMMY", + "matchPattern": "qdisc red 1: root .* limit 1Mb min 100Kb max 300Kb adaptive $", + "matchCount": "1", + "teardown": [ + "$TC qdisc del dev $DUMMY handle 1: root", + "$IP link del dev $DUMMY type dummy" + ] + }, + { + "id": "2d4b", + "name": "Create RED with ECN flag", + "category": [ + "qdisc", + "red" + ], + "plugins": { + "requires": "nsPlugin" + }, + "setup": [ + "$IP link add dev $DUMMY type dummy || /bin/true" + ], + "cmdUnderTest": "$TC qdisc add dev $DUMMY handle 1: root red ecn limit 1M avpkt 1500 min 100K max 300K", + "expExitCode": "0", + "verifyCmd": "$TC qdisc show dev $DUMMY", + "matchPattern": "qdisc red 1: root .* limit 1Mb min 100Kb max 300Kb ecn $", + "matchCount": "1", + "teardown": [ + "$TC qdisc del dev $DUMMY handle 1: root", + "$IP link del dev $DUMMY type dummy" + ] + }, + { + "id": "650f", + "name": "Create RED with flags ECN, adaptive", + "category": [ + "qdisc", + "red" + ], + "plugins": { + "requires": "nsPlugin" + }, + "setup": [ + "$IP link add dev $DUMMY type dummy || /bin/true" + ], + "cmdUnderTest": "$TC qdisc add dev $DUMMY handle 1: root red ecn adaptive limit 1M avpkt 1500 min 100K max 300K", + "expExitCode": "0", + "verifyCmd": "$TC qdisc show dev $DUMMY", + "matchPattern": "qdisc red 1: root .* limit 1Mb min 100Kb max 300Kb ecn adaptive $", + "matchCount": "1", + "teardown": [ + "$TC qdisc del dev $DUMMY handle 1: root", + "$IP link del dev $DUMMY type dummy" + ] + }, + { + "id": "5f15", + "name": "Create RED with flags ECN, harddrop", + "category": [ + "qdisc", + "red" + ], + "plugins": { + "requires": "nsPlugin" + }, + "setup": [ + "$IP link add dev $DUMMY type dummy || /bin/true" + ], + "cmdUnderTest": "$TC qdisc add dev $DUMMY handle 1: root red ecn harddrop limit 1M avpkt 1500 min 100K max 300K", + "expExitCode": "0", + "verifyCmd": "$TC qdisc show dev $DUMMY", + "matchPattern": "qdisc red 1: root .* limit 1Mb min 100Kb max 300Kb ecn harddrop $", + "matchCount": "1", + "teardown": [ + "$TC qdisc del dev $DUMMY handle 1: root", + "$IP link del dev $DUMMY type dummy" + ] + }, + { + "id": "53e8", + "name": "Create RED with flags ECN, nodrop", + "category": [ + "qdisc", + "red" + ], + "plugins": { + "requires": "nsPlugin" + }, + "setup": [ + "$IP link add dev $DUMMY type dummy || /bin/true" + ], + "cmdUnderTest": "$TC qdisc add dev $DUMMY handle 1: root red ecn nodrop limit 1M avpkt 1500 min 100K max 300K", + "expExitCode": "0", + "verifyCmd": "$TC qdisc show dev $DUMMY", + "matchPattern": "qdisc red 1: root .* limit 1Mb min 100Kb max 300Kb ecn nodrop $", + "matchCount": "1", + "teardown": [ + "$TC qdisc del dev $DUMMY handle 1: root", + "$IP link del dev $DUMMY type dummy" + ] + }, + { + "id": "d091", + "name": "Fail to create RED with only nodrop flag", + "category": [ + "qdisc", + "red" + ], + "plugins": { + "requires": "nsPlugin" + }, + "setup": [ + "$IP link add dev $DUMMY type dummy || /bin/true" + ], + "cmdUnderTest": "$TC qdisc add dev $DUMMY handle 1: root red nodrop limit 1M avpkt 1500 min 100K max 300K", + "expExitCode": "2", + "verifyCmd": "$TC qdisc show dev $DUMMY", + "matchPattern": "qdisc red", + "matchCount": "0", + "teardown": [ + "$IP link del dev $DUMMY type dummy" + ] + }, + { + "id": "af8e", + "name": "Create RED with flags ECN, nodrop, harddrop", + "category": [ + "qdisc", + "red" + ], + "plugins": { + "requires": "nsPlugin" + }, + "setup": [ + "$IP link add dev $DUMMY type dummy || /bin/true" + ], + "cmdUnderTest": "$TC qdisc add dev $DUMMY handle 1: root red ecn harddrop nodrop limit 1M avpkt 1500 min 100K max 300K", + "expExitCode": "0", + "verifyCmd": "$TC qdisc show dev $DUMMY", + "matchPattern": "qdisc red 1: root .* limit 1Mb min 100Kb max 300Kb ecn harddrop nodrop $", + "matchCount": "1", + "teardown": [ + "$TC qdisc del dev $DUMMY handle 1: root", + "$IP link del dev $DUMMY type dummy" + ] + } +]