Merge branch 'mlxsw-multi-level-qdisc-offload'

Ido Schimmel says:

====================
mlxsw: Multi-level qdisc offload

Petr says:

Currently, mlxsw admits for offload a suitable root qdisc, and its
children. Thus up to two levels of hierarchy are offloaded. Often, this is
enough: one can configure TCs with RED and TCs with a shaper on, and can
even see counters for each TC by looking at a qdisc at a sufficiently
shallow position.

While simple, the system has obvious shortcomings. It is not possible to
configure both RED and shaping on one TC. It is not possible to place a
PRIO below root TBF, which would then be offloaded as port shaper. FIFOs
are only offloaded at root or directly below, which is confusing to users,
because RED and TBF of course have their own FIFO.

This patch set lifts assumptions that prevent offloading multi-level qdisc
trees.

In patch #1, offload of a graft operation is added to TBF. Grafts are
issued as another qdisc is linked to the qdisc in question, and give
drivers a chance to react to the linking. The absence of this event was not
a major issue so far, because TBF was not considered classful, which
changes with this patchset.

The codebase currently assumes that ETS and PRIO are the only classful
qdiscs. The following patches gradually lift this assumption.

In patch #2, calculation of traffic class and priomap of a qdisc is fixed.

Patch #3 fixes handling of future FIFOs. Child FIFO qdiscs may be created
and notified before their parent qdisc exists and therefore need special
handling.

Patches #4, #5 and #6 unify, respectively, child destruction, child
grafting, and cleanup of statistics.

Patch #7 adds a function that validates whether a given qdisc topology is
offloadable.

Finally in patch #8, TBF and RED become classful. At this point, FIFO
qdiscs grafted to an offloaded qdisc should always be offloaded.

Patch #9 adds a selftest to verify some offloadable and unoffloadable qdisc
trees.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/pkt_cls.h

@@ -977,6 +977,7 @@ enum tc_tbf_command {
 	TC_TBF_REPLACE,
 	TC_TBF_DESTROY,
 	TC_TBF_STATS,
+	TC_TBF_GRAFT,
 };
 
 struct tc_tbf_qopt_offload_replace_params {
@@ -992,6 +993,7 @@ struct tc_tbf_qopt_offload {
 	union {
 		struct tc_tbf_qopt_offload_replace_params replace_params;
 		struct tc_qopt_offload_stats stats;
+		u32 child_handle;
 	};
 };
 

net/sched/sch_tbf.c

@@ -184,6 +184,20 @@ static int tbf_offload_dump(struct Qdisc *sch)
 	return qdisc_offload_dump_helper(sch, TC_SETUP_QDISC_TBF, &qopt);
 }
 
+static void tbf_offload_graft(struct Qdisc *sch, struct Qdisc *new,
+			      struct Qdisc *old, struct netlink_ext_ack *extack)
+{
+	struct tc_tbf_qopt_offload graft_offload = {
+		.handle		= sch->handle,
+		.parent		= sch->parent,
+		.child_handle	= new->handle,
+		.command	= TC_TBF_GRAFT,
+	};
+
+	qdisc_offload_graft_helper(qdisc_dev(sch), sch, new, old,
+				   TC_SETUP_QDISC_TBF, &graft_offload, extack);
+}
+
 /* GSO packet is too big, segment it so that tbf can transmit
  * each segment in time
  */
@@ -547,6 +561,8 @@ static int tbf_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new,
 		new = &noop_qdisc;
 
 	*old = qdisc_replace(sch, new, &q->qdisc);
+
+	tbf_offload_graft(sch, new, *old, extack);
 	return 0;
 }
 

tools/testing/selftests/drivers/net/mlxsw/sch_offload.sh

+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+#
+# Test qdisc offload indication
+
+
+ALL_TESTS="
+	test_root
+	test_etsprio
+"
+NUM_NETIFS=1
+lib_dir=$(dirname $0)/../../../net/forwarding
+source $lib_dir/lib.sh
+
+check_not_offloaded()
+{
+	local handle=$1; shift
+	local h
+	local offloaded
+
+	h=$(qdisc_stats_get $h1 "$handle" .handle)
+	[[ $h == '"'$handle'"' ]]
+	check_err $? "Qdisc with handle $handle does not exist"
+
+	offloaded=$(qdisc_stats_get $h1 "$handle" .offloaded)
+	[[ $offloaded == true ]]
+	check_fail $? "Qdisc with handle $handle offloaded, but should not be"
+}
+
+check_all_offloaded()
+{
+	local handle=$1; shift
+
+	if [[ ! -z $handle ]]; then
+		local offloaded=$(qdisc_stats_get $h1 "$handle" .offloaded)
+		[[ $offloaded == true ]]
+		check_err $? "Qdisc with handle $handle not offloaded"
+	fi
+
+	local unoffloaded=$(tc q sh dev $h1 invisible |
+				grep -v offloaded |
+				sed s/root/parent\ root/ |
+				cut -d' ' -f 5)
+	[[ -z $unoffloaded ]]
+	check_err $? "Qdiscs with following parents not offloaded: $unoffloaded"
+
+	pre_cleanup
+}
+
+with_ets()
+{
+	local handle=$1; shift
+	local locus=$1; shift
+
+	tc qdisc add dev $h1 $locus handle $handle \
+	   ets bands 8 priomap 7 6 5 4 3 2 1 0
+	"$@"
+	tc qdisc del dev $h1 $locus
+}
+
+with_prio()
+{
+	local handle=$1; shift
+	local locus=$1; shift
+
+	tc qdisc add dev $h1 $locus handle $handle \
+	   prio bands 8 priomap 7 6 5 4 3 2 1 0
+	"$@"
+	tc qdisc del dev $h1 $locus
+}
+
+with_red()
+{
+	local handle=$1; shift
+	local locus=$1; shift
+
+	tc qdisc add dev $h1 $locus handle $handle \
+	   red limit 1000000 min 200000 max 300000 probability 0.5 avpkt 1500
+	"$@"
+	tc qdisc del dev $h1 $locus
+}
+
+with_tbf()
+{
+	local handle=$1; shift
+	local locus=$1; shift
+
+	tc qdisc add dev $h1 $locus handle $handle \
+	   tbf rate 400Mbit burst 128K limit 1M
+	"$@"
+	tc qdisc del dev $h1 $locus
+}
+
+with_pfifo()
+{
+	local handle=$1; shift
+	local locus=$1; shift
+
+	tc qdisc add dev $h1 $locus handle $handle pfifo limit 100K
+	"$@"
+	tc qdisc del dev $h1 $locus
+}
+
+with_bfifo()
+{
+	local handle=$1; shift
+	local locus=$1; shift
+
+	tc qdisc add dev $h1 $locus handle $handle bfifo limit 100K
+	"$@"
+	tc qdisc del dev $h1 $locus
+}
+
+with_drr()
+{
+	local handle=$1; shift
+	local locus=$1; shift
+
+	tc qdisc add dev $h1 $locus handle $handle drr
+	"$@"
+	tc qdisc del dev $h1 $locus
+}
+
+with_qdiscs()
+{
+	local handle=$1; shift
+	local parent=$1; shift
+	local kind=$1; shift
+	local next_handle=$((handle * 2))
+	local locus;
+
+	if [[ $kind == "--" ]]; then
+		local cmd=$1; shift
+		$cmd $(printf %x: $parent) "$@"
+	else
+		if ((parent == 0)); then
+			locus=root
+		else
+			locus=$(printf "parent %x:1" $parent)
+		fi
+
+		with_$kind $(printf %x: $handle) "$locus" \
+			with_qdiscs $next_handle $handle "$@"
+	fi
+}
+
+get_name()
+{
+	local parent=$1; shift
+	local name=$(echo "" "${@^^}" | tr ' ' -)
+
+	if ((parent != 0)); then
+		kind=$(qdisc_stats_get $h1 $parent: .kind)
+		kind=${kind%\"}
+		kind=${kind#\"}
+		name="-${kind^^}$name"
+	fi
+
+	echo root$name
+}
+
+do_test_offloaded()
+{
+	local handle=$1; shift
+	local parent=$1; shift
+
+	RET=0
+	with_qdiscs $handle $parent "$@" -- check_all_offloaded
+	log_test $(get_name $parent "$@")" offloaded"
+}
+
+do_test_nooffload()
+{
+	local handle=$1; shift
+	local parent=$1; shift
+
+	local name=$(echo "${@^^}" | tr ' ' -)
+	local kind
+
+	RET=0
+	with_qdiscs $handle $parent "$@" -- check_not_offloaded
+	log_test $(get_name $parent "$@")" not offloaded"
+}
+
+do_test_combinations()
+{
+	local handle=$1; shift
+	local parent=$1; shift
+
+	local cont
+	local leaf
+	local fifo
+
+	for cont in "" ets prio; do
+		for leaf in "" red tbf "red tbf" "tbf red"; do
+			for fifo in "" pfifo bfifo; do
+				if [[ -z "$cont$leaf$fifo" ]]; then
+					continue
+				fi
+				do_test_offloaded $handle $parent \
+						  $cont $leaf $fifo
+			done
+		done
+	done
+
+	for cont in ets prio; do
+		for leaf in red tbf; do
+			do_test_nooffload $handle $parent $cont red tbf $leaf
+			do_test_nooffload $handle $parent $cont tbf red $leaf
+		done
+		for leaf in "red red" "tbf tbf"; do
+			do_test_nooffload $handle $parent $cont $leaf
+		done
+	done
+
+	do_test_nooffload $handle $parent drr
+}
+
+test_root()
+{
+	do_test_combinations 1 0
+}
+
+do_test_etsprio()
+{
+	local parent=$1; shift
+	local tbfpfx=$1; shift
+	local cont
+
+	for cont in ets prio; do
+		RET=0
+		with_$cont 8: "$parent" \
+			with_red 11: "parent 8:1" \
+			with_red 12: "parent 8:2" \
+			with_tbf 13: "parent 8:3" \
+			with_tbf 14: "parent 8:4" \
+			check_all_offloaded
+		log_test "root$tbfpfx-ETS-{RED,TBF} offloaded"
+
+		RET=0
+		with_$cont 8: "$parent" \
+			with_red 81: "parent 8:1" \
+				with_tbf 811: "parent 81:1" \
+			with_tbf 84: "parent 8:4" \
+				with_red 841: "parent 84:1" \
+			check_all_offloaded
+		log_test "root$tbfpfx-ETS-{RED-TBF,TBF-RED} offloaded"
+
+		RET=0
+		with_$cont 8: "$parent" \
+			with_red 81: "parent 8:1" \
+				with_tbf 811: "parent 81:1" \
+					with_bfifo 8111: "parent 811:1" \
+			with_tbf 82: "parent 8:2" \
+				with_red 821: "parent 82:1" \
+					with_bfifo 8211: "parent 821:1" \
+			check_all_offloaded
+		log_test "root$tbfpfx-ETS-{RED-TBF-bFIFO,TBF-RED-bFIFO} offloaded"
+	done
+}
+
+test_etsprio()
+{
+	do_test_etsprio root ""
+}
+
+cleanup()
+{
+	tc qdisc del dev $h1 root &>/dev/null
+}
+
+trap cleanup EXIT
+h1=${NETIFS[p1]}
+tests_run
+
+exit $EXIT_STATUS