[PATCH iproute2-next v2 0/4] Support qevents

[PATCH iproute2-next v2 0/4] Support qevents

[Petr Machata]

To allow configuring user-defined actions as a result of inner workings of
a qdisc, a concept of qevents was recently introduced to the kernel.
Qevents are attach points for TC blocks, where filters can be put that are
executed as the packet hits well-defined points in the qdisc algorithms.
The attached blocks can be shared, in a manner similar to clsact ingress
and egress blocks, arbitrary classifiers with arbitrary actions can be put
on them, etc.

For example:

# tc qdisc add dev eth0 root handle 1: \
	red limit 500K avpkt 1K qevent early_drop block 10
# tc filter add block 10 \
	matchall action mirred egress mirror dev eth1

This patch set introduces the corresponding iproute2 support. Patch #1 adds
the new netlink attribute enumerators. Patch #2 adds a set of helpers to
implement qevents, and #3 adds a generic documentation to tc.8. Patch #4
then adds two new qevents to the RED qdisc: mark and early_drop.

Changes from v1 to v2:
- Patch #3:
    - s/early/early_drop/ in the example.

Changes from RFC to v1:
- Rename "tail" qevent to "tail_drop".
- Adapt to the new 100-column standard.
- Add a selftest

Petr Machata (4):
  uapi: pkt_sched: Add two new RED attributes
  tc: Add helpers to support qevent handling
  man: tc: Describe qevents
  tc: q_red: Add support for qevents "mark" and "early_drop"

 include/uapi/linux/pkt_sched.h |   2 +
 man/man8/tc-red.8              |  18 ++-
 man/man8/tc.8                  |  19 ++++
 tc/Makefile                    |   1 +
 tc/q_red.c                     |  30 ++++-
 tc/tc_qevent.c                 | 202 +++++++++++++++++++++++++++++++++
 tc/tc_qevent.h                 |  49 ++++++++
 7 files changed, 317 insertions(+), 4 deletions(-)
 create mode 100644 tc/tc_qevent.c
 create mode 100644 tc/tc_qevent.h

-- 
2.20.1

[PATCH iproute2-next v2 1/4] uapi: pkt_sched: Add two new RED attributes

[Petr Machata]

Signed-off-by: Petr Machata <petrm@mellanox.com>
---
 include/uapi/linux/pkt_sched.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
index a95f3ae7..9e7c2c60 100644
--- a/include/uapi/linux/pkt_sched.h
+++ b/include/uapi/linux/pkt_sched.h
@@ -257,6 +257,8 @@ enum {
 	TCA_RED_STAB,
 	TCA_RED_MAX_P,
 	TCA_RED_FLAGS,		/* bitfield32 */
+	TCA_RED_EARLY_DROP_BLOCK, /* u32 */
+	TCA_RED_MARK_BLOCK,	/* u32 */
 	__TCA_RED_MAX,
 };
 
-- 
2.20.1

[PATCH iproute2-next v2 2/4] tc: Add helpers to support qevent handling

[Petr Machata]

Introduce a set of helpers to make it easy to add support for qevents into
qdisc.

The idea behind this is that qevent types will be generally reused between
qdiscs, rather than each having a completely idiosyncratic set of qevents.
The qevent module holds functions for parsing, dumping and formatting of
these common qevent types, and for dispatch to the appropriate set of
handlers based on the qevent name.

Signed-off-by: Petr Machata <petrm@mellanox.com>
---
 tc/Makefile    |   1 +
 tc/tc_qevent.c | 202 +++++++++++++++++++++++++++++++++++++++++++++++++
 tc/tc_qevent.h |  49 ++++++++++++
 3 files changed, 252 insertions(+)
 create mode 100644 tc/tc_qevent.c
 create mode 100644 tc/tc_qevent.h

diff --git a/tc/Makefile b/tc/Makefile
index 79c9c1dd..5a517af2 100644
--- a/tc/Makefile
+++ b/tc/Makefile
@@ -122,6 +122,7 @@ TCLIB += tc_red.o
 TCLIB += tc_cbq.o
 TCLIB += tc_estimator.o
 TCLIB += tc_stab.o
+TCLIB += tc_qevent.o
 
 CFLAGS += -DCONFIG_GACT -DCONFIG_GACT_PROB
 ifneq ($(IPT_LIB_DIR),)
diff --git a/tc/tc_qevent.c b/tc/tc_qevent.c
new file mode 100644
index 00000000..1f8e6506
--- /dev/null
+++ b/tc/tc_qevent.c
@@ -0,0 +1,202 @@
+// SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause
+
+/*
+ * Helpers for handling qevents.
+ */
+
+#include <stdio.h>
+#include <string.h>
+
+#include "tc_qevent.h"
+#include "utils.h"
+
+void qevents_init(struct qevent_util *qevents)
+{
+	if (!qevents)
+		return;
+
+	for (; qevents->id; qevents++)
+		memset(qevents->data, 0, qevents->data_size);
+}
+
+int qevent_parse(struct qevent_util *qevents, int *p_argc, char ***p_argv)
+{
+	char **argv = *p_argv;
+	int argc = *p_argc;
+	const char *name = *argv;
+	int err;
+
+	if (!qevents)
+		goto out;
+
+	for (; qevents->id; qevents++) {
+		if (strcmp(name, qevents->id) == 0) {
+			NEXT_ARG();
+			err = qevents->parse_qevent(qevents, &argc, &argv);
+			if (err)
+				return err;
+
+			*p_argc = argc;
+			*p_argv = argv;
+			return 0;
+		}
+	}
+
+out:
+	fprintf(stderr, "Unknown qevent `%s'\n", name);
+	return -1;
+}
+
+int qevents_read(struct qevent_util *qevents, struct rtattr **tb)
+{
+	int err;
+
+	if (!qevents)
+		return 0;
+
+	for (; qevents->id; qevents++) {
+		if (tb[qevents->attr]) {
+			err = qevents->read_qevent(qevents, tb);
+			if (err)
+				return err;
+		}
+	}
+
+	return 0;
+}
+
+void qevents_print(struct qevent_util *qevents, FILE *f)
+{
+	int first = true;
+
+	if (!qevents)
+		return;
+
+	for (; qevents->id; qevents++) {
+		struct qevent_base *qeb = qevents->data;
+
+		if (qeb->block_idx) {
+			if (first) {
+				open_json_array(PRINT_JSON, "qevents");
+				first = false;
+			}
+
+			open_json_object(NULL);
+			print_string(PRINT_ANY, "kind", " qevent %s", qevents->id);
+			qevents->print_qevent(qevents, f);
+			close_json_object();
+		}
+	}
+
+	if (!first)
+		close_json_array(PRINT_ANY, "");
+}
+
+int qevents_dump(struct qevent_util *qevents, struct nlmsghdr *n)
+{
+	int err;
+
+	if (!qevents)
+		return 0;
+
+	for (; qevents->id; qevents++) {
+		struct qevent_base *qeb = qevents->data;
+
+		if (qeb->block_idx) {
+			err = qevents->dump_qevent(qevents, n);
+			if (err)
+				return err;
+		}
+	}
+
+	return 0;
+}
+
+static int parse_block_idx(const char *arg, struct qevent_base *qeb)
+{
+	if (qeb->block_idx) {
+		fprintf(stderr, "Qevent block index already specified\n");
+		return -1;
+	}
+
+	if (get_unsigned(&qeb->block_idx, arg, 10) || !qeb->block_idx) {
+		fprintf(stderr, "Illegal qevent block index\n");
+		return -1;
+	}
+
+	return 0;
+}
+
+static int read_block_idx(struct rtattr *attr, struct qevent_base *qeb)
+{
+	if (qeb->block_idx) {
+		fprintf(stderr, "Qevent block index already specified\n");
+		return -1;
+	}
+
+	qeb->block_idx = rta_getattr_u32(attr);
+	if (!qeb->block_idx) {
+		fprintf(stderr, "Illegal qevent block index\n");
+		return -1;
+	}
+
+	return 0;
+}
+
+static void print_block_idx(FILE *f, __u32 block_idx)
+{
+	print_uint(PRINT_ANY, "block", " block %u", block_idx);
+}
+
+int qevent_parse_plain(struct qevent_util *qu, int *p_argc, char ***p_argv)
+{
+	struct qevent_plain *qe = qu->data;
+	char **argv = *p_argv;
+	int argc = *p_argc;
+
+	if (qe->base.block_idx) {
+		fprintf(stderr, "Duplicate qevent\n");
+		return -1;
+	}
+
+	while (argc > 0) {
+		if (strcmp(*argv, "block") == 0) {
+			NEXT_ARG();
+			if (parse_block_idx(*argv, &qe->base))
+				return -1;
+		} else {
+			break;
+		}
+		NEXT_ARG_FWD();
+	}
+
+	if (!qe->base.block_idx) {
+		fprintf(stderr, "Unspecified qevent block index\n");
+		return -1;
+	}
+
+	*p_argc = argc;
+	*p_argv = argv;
+	return 0;
+}
+
+int qevent_read_plain(struct qevent_util *qu, struct rtattr **tb)
+{
+	struct qevent_plain *qe = qu->data;
+
+	return read_block_idx(tb[qu->attr], &qe->base);
+}
+
+void qevent_print_plain(struct qevent_util *qu, FILE *f)
+{
+	struct qevent_plain *qe = qu->data;
+
+	print_block_idx(f, qe->base.block_idx);
+}
+
+int qevent_dump_plain(struct qevent_util *qu, struct nlmsghdr *n)
+{
+	struct qevent_plain *qe = qu->data;
+
+	return addattr32(n, 1024, qu->attr, qe->base.block_idx);
+}
diff --git a/tc/tc_qevent.h b/tc/tc_qevent.h
new file mode 100644
index 00000000..574e7cff
--- /dev/null
+++ b/tc/tc_qevent.h
@@ -0,0 +1,49 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _TC_QEVENT_H_
+#define _TC_QEVENT_H_
+
+#include <linux/types.h>
+#include <libnetlink.h>
+
+struct qevent_base {
+	__u32 block_idx;
+};
+
+struct qevent_util {
+	const char *id;
+	int (*parse_qevent)(struct qevent_util *qu, int *argc, char ***argv);
+	int (*read_qevent)(struct qevent_util *qu, struct rtattr **tb);
+	void (*print_qevent)(struct qevent_util *qu, FILE *f);
+	int (*dump_qevent)(struct qevent_util *qu, struct nlmsghdr *n);
+	size_t data_size;
+	void *data;
+	int attr;
+};
+
+#define QEVENT(_name, _form, _data, _attr)				\
+	{								\
+		.id = _name,						\
+		.parse_qevent = qevent_parse_##_form,			\
+		.read_qevent = qevent_read_##_form,			\
+		.print_qevent = qevent_print_##_form,			\
+		.dump_qevent = qevent_dump_##_form,			\
+		.data_size = sizeof(struct qevent_##_form),		\
+		.data = _data,						\
+		.attr = _attr,						\
+	}
+
+void qevents_init(struct qevent_util *qevents);
+int qevent_parse(struct qevent_util *qevents, int *p_argc, char ***p_argv);
+int qevents_read(struct qevent_util *qevents, struct rtattr **tb);
+int qevents_dump(struct qevent_util *qevents, struct nlmsghdr *n);
+void qevents_print(struct qevent_util *qevents, FILE *f);
+
+struct qevent_plain {
+	struct qevent_base base;
+};
+int qevent_parse_plain(struct qevent_util *qu, int *p_argc, char ***p_argv);
+int qevent_read_plain(struct qevent_util *qu, struct rtattr **tb);
+void qevent_print_plain(struct qevent_util *qu, FILE *f);
+int qevent_dump_plain(struct qevent_util *qu, struct nlmsghdr *n);
+
+#endif
-- 
2.20.1

[PATCH iproute2-next v2 3/4] man: tc: Describe qevents

[Petr Machata]

Add some general remarks about qevents.

Signed-off-by: Petr Machata <petrm@mellanox.com>
---

Notes:
    v2:
    - s/early/early_drop/ in the example.

 man/man8/tc.8 | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/man/man8/tc.8 b/man/man8/tc.8
index e8e0cd0f..eba73dbf 100644
--- a/man/man8/tc.8
+++ b/man/man8/tc.8
@@ -254,6 +254,25 @@ Traffic control filter that matches every packet. See
 .BR tc-matchall (8)
 for details.
 
+.SH QEVENTS
+Qdiscs may invoke user-configured actions when certain interesting events
+take place in the qdisc. Each qevent can either be unused, or can have a
+block attached to it. To this block are then attached filters using the "tc
+block BLOCK_IDX" syntax. The block is executed when the qevent associated
+with the attachment point takes place. For example, packet could be
+dropped, or delayed, etc., depending on the qdisc and the qevent in
+question.
+
+For example:
+.PP
+.RS
+tc qdisc add dev eth0 root handle 1: red limit 500K avpkt 1K \\
+   qevent early_drop block 10
+.RE
+.RS
+tc filter add block 10 matchall action mirred egress mirror dev eth1
+.RE
+
 .SH CLASSLESS QDISCS
 The classless qdiscs are:
 .TP
-- 
2.20.1

[PATCH iproute2-next v2 4/4] tc: q_red: Add support for qevents "mark" and "early_drop"

[Petr Machata]

The "early_drop" qevent matches packets that have been early-dropped. The
"mark" qevent matches packets that have been ECN-marked.

Signed-off-by: Petr Machata <petrm@mellanox.com>
---
 man/man8/tc-red.8 | 18 +++++++++++++++++-
 tc/q_red.c        | 30 +++++++++++++++++++++++++++---
 2 files changed, 44 insertions(+), 4 deletions(-)

diff --git a/man/man8/tc-red.8 b/man/man8/tc-red.8
index b5aaa986..662e4d8b 100644
--- a/man/man8/tc-red.8
+++ b/man/man8/tc-red.8
@@ -17,7 +17,11 @@ packets
 rate
 .B ] [ probability
 chance
-.B ] [ adaptive ]
+.B ] [ adaptive ] [ qevent early_drop block
+index
+.B ] [ qevent mark block
+index
+.B ]
 
 .SH DESCRIPTION
 Random Early Detection is a classless qdisc which manages its queue size
@@ -134,6 +138,18 @@ Goal of Adaptive RED is to make 'probability' dynamic value between 1% and 50% t
 .B (max - min) / 2
 .fi
 
+.SH QEVENTS
+See tc (8) for some general notes about qevents. The RED qdisc supports the
+following qevents:
+
+.TP
+early_drop
+The associated block is executed when packets are early-dropped. This includes
+non-ECT packets in ECN mode.
+.TP
+mark
+The associated block is executed when packets are marked in ECN mode.
+
 .SH EXAMPLE
 
 .P
diff --git a/tc/q_red.c b/tc/q_red.c
index 53181c82..97856f03 100644
--- a/tc/q_red.c
+++ b/tc/q_red.c
@@ -22,6 +22,7 @@
 
 #include "utils.h"
 #include "tc_util.h"
+#include "tc_qevent.h"
 
 #include "tc_red.h"
 
@@ -30,11 +31,20 @@ static void explain(void)
 	fprintf(stderr,
 		"Usage: ... red	limit BYTES [min BYTES] [max BYTES] avpkt BYTES [burst PACKETS]\n"
 		"		[adaptive] [probability PROBABILITY] [bandwidth KBPS]\n"
-		"		[ecn] [harddrop] [nodrop]\n");
+		"		[ecn] [harddrop] [nodrop]\n"
+		"		[qevent early_drop block IDX] [qevent mark block IDX]\n");
 }
 
 #define RED_SUPPORTED_FLAGS (TC_RED_HISTORIC_FLAGS | TC_RED_NODROP)
 
+static struct qevent_plain qe_early_drop = {};
+static struct qevent_plain qe_mark = {};
+static struct qevent_util qevents[] = {
+	QEVENT("early_drop", plain, &qe_early_drop, TCA_RED_EARLY_DROP_BLOCK),
+	QEVENT("mark", plain, &qe_mark, TCA_RED_MARK_BLOCK),
+	{},
+};
+
 static int red_parse_opt(struct qdisc_util *qu, int argc, char **argv,
 			 struct nlmsghdr *n, const char *dev)
 {
@@ -51,6 +61,8 @@ static int red_parse_opt(struct qdisc_util *qu, int argc, char **argv,
 	__u32 max_P;
 	struct rtattr *tail;
 
+	qevents_init(qevents);
+
 	while (argc > 0) {
 		if (strcmp(*argv, "limit") == 0) {
 			NEXT_ARG();
@@ -109,6 +121,11 @@ static int red_parse_opt(struct qdisc_util *qu, int argc, char **argv,
 			flags_bf.value |= TC_RED_ADAPTATIVE;
 		} else if (strcmp(*argv, "adaptive") == 0) {
 			flags_bf.value |= TC_RED_ADAPTATIVE;
+		} else if (matches(*argv, "qevent") == 0) {
+			NEXT_ARG();
+			if (qevent_parse(qevents, &argc, &argv))
+				return -1;
+			continue;
 		} else if (strcmp(*argv, "help") == 0) {
 			explain();
 			return -1;
@@ -162,6 +179,8 @@ static int red_parse_opt(struct qdisc_util *qu, int argc, char **argv,
 	max_P = probability * pow(2, 32);
 	addattr_l(n, 1024, TCA_RED_MAX_P, &max_P, sizeof(max_P));
 	addattr_l(n, 1024, TCA_RED_FLAGS, &flags_bf, sizeof(flags_bf));
+	if (qevents_dump(qevents, n))
+		return -1;
 	addattr_nest_end(n, tail);
 	return 0;
 }
@@ -203,12 +222,12 @@ static int red_print_opt(struct qdisc_util *qu, FILE *f, struct rtattr *opt)
 	print_uint(PRINT_JSON, "min", NULL, qopt->qth_min);
 	print_string(PRINT_FP, NULL, "min %s ", sprint_size(qopt->qth_min, b2));
 	print_uint(PRINT_JSON, "max", NULL, qopt->qth_max);
-	print_string(PRINT_FP, NULL, "max %s ", sprint_size(qopt->qth_max, b3));
+	print_string(PRINT_FP, NULL, "max %s", sprint_size(qopt->qth_max, b3));
 
 	tc_red_print_flags(qopt->flags);
 
 	if (show_details) {
-		print_uint(PRINT_ANY, "ewma", "ewma %u ", qopt->Wlog);
+		print_uint(PRINT_ANY, "ewma", " ewma %u ", qopt->Wlog);
 		if (max_P)
 			print_float(PRINT_ANY, "probability",
 				    "probability %lg ", max_P / pow(2, 32));
@@ -217,6 +236,11 @@ static int red_print_opt(struct qdisc_util *qu, FILE *f, struct rtattr *opt)
 		print_uint(PRINT_ANY, "Scell_log", "Scell_log %u",
 			   qopt->Scell_log);
 	}
+
+	qevents_init(qevents);
+	if (qevents_read(qevents, tb))
+		return -1;
+	qevents_print(qevents, f);
 	return 0;
 }
 
-- 
2.20.1

Re: [PATCH iproute2-next v2 0/4] Support qevents

[David Ahern]

On 6/30/20 4:14 AM, Petr Machata wrote:
> To allow configuring user-defined actions as a result of inner workings of
> a qdisc, a concept of qevents was recently introduced to the kernel.
> Qevents are attach points for TC blocks, where filters can be put that are
> executed as the packet hits well-defined points in the qdisc algorithms.
> The attached blocks can be shared, in a manner similar to clsact ingress
> and egress blocks, arbitrary classifiers with arbitrary actions can be put
> on them, etc.
> 
> For example:
> 
> # tc qdisc add dev eth0 root handle 1: \
> 	red limit 500K avpkt 1K qevent early_drop block 10
> # tc filter add block 10 \
> 	matchall action mirred egress mirror dev eth1
> 
> This patch set introduces the corresponding iproute2 support. Patch #1 adds
> the new netlink attribute enumerators. Patch #2 adds a set of helpers to
> implement qevents, and #3 adds a generic documentation to tc.8. Patch #4
> then adds two new qevents to the RED qdisc: mark and early_drop.
> 

applied to iproute2-next. Thanks