[Codel] [PATCH iproute2] codel: Controlled Delay AQM

[Eric Dumazet]

 include/linux/pkt_sched.h |   19 ++++
 tc/Makefile               |    1 
 tc/q_codel.c              |  155 ++++++++++++++++++++++++++++++++++++
 3 files changed, 175 insertions(+)


diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
index 410b33d..fbece83 100644
--- a/include/linux/pkt_sched.h
+++ b/include/linux/pkt_sched.h
@@ -654,4 +654,23 @@ struct tc_qfq_stats {
 	__u32 lmax;
 };
 
+/* CODEL */
+
+enum {
+	TCA_CODEL_UNSPEC,
+	TCA_CODEL_TARGET,
+	TCA_CODEL_LIMIT,
+	TCA_CODEL_MINBYTES,
+	TCA_CODEL_INTERVAL,
+	__TCA_CODEL_MAX
+};
+
+#define TCA_CODEL_MAX	(__TCA_CODEL_MAX - 1)
+
+struct tc_codel_xstats {
+	__u32	count;
+	__u32	delay; /* time elapsed since next packet was queued (in us) */
+	__u32	drop_next;
+};
+
 #endif
diff --git a/tc/Makefile b/tc/Makefile
index be8cd5a..8a7cc8d 100644
--- a/tc/Makefile
+++ b/tc/Makefile
@@ -47,6 +47,7 @@ TCMODULES += em_cmp.o
 TCMODULES += em_u32.o
 TCMODULES += em_meta.o
 TCMODULES += q_mqprio.o
+TCMODULES += q_codel.o
 
 TCSO :=
 ifeq ($(TC_CONFIG_ATM),y)
diff --git a/tc/q_codel.c b/tc/q_codel.c
new file mode 100644
index 0000000..ec23ebb
--- /dev/null
+++ b/tc/q_codel.c
@@ -0,0 +1,155 @@
+/*
+ * q_codel.c		Codel.
+ *
+ *	This program is free software; you can redistribute it and/or
+ *	modify it under the terms of the GNU General Public License
+ *	as published by the Free Software Foundation; either version
+ *	2 of the License, or (at your option) any later version.
+ *
+ * Authors:	Eric Dumazet <edumazet at google.com>
+ *		Dave Taht <dave.taht at bufferbloat.net>
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <syslog.h>
+#include <fcntl.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <arpa/inet.h>
+#include <string.h>
+
+#include "utils.h"
+#include "tc_util.h"
+
+static void explain(void)
+{
+	fprintf(stderr, "Usage: ... codel [ limit PACKETS ] [ target TIME]\n");
+	fprintf(stderr, "                 [ interval TIME ] [ minbytes BYTES ]\n");
+}
+
+static int codel_parse_opt(struct qdisc_util *qu, int argc, char **argv,
+			   struct nlmsghdr *n)
+{
+	unsigned limit = 0;
+	unsigned target = 0;
+	unsigned interval = 0;
+	unsigned minbytes = 0;
+	struct rtattr *tail;
+
+	while (argc > 0) {
+		if (strcmp(*argv, "limit") == 0) {
+			NEXT_ARG();
+			if (get_unsigned(&limit, *argv, 0)) {
+				fprintf(stderr, "Illegal \"limit\"\n");
+				return -1;
+			}
+		} else if (strcmp(*argv, "minbytes") == 0) {
+			NEXT_ARG();
+			if (get_unsigned(&minbytes, *argv, 0)) {
+				fprintf(stderr, "Illegal \"minbytes\"\n");
+				return -1;
+			}
+		} else if (strcmp(*argv, "target") == 0) {
+			NEXT_ARG();
+			if (get_time(&target, *argv)) {
+				fprintf(stderr, "Illegal \"target\"\n");
+				return -1;
+			}
+		} else if (strcmp(*argv, "interval") == 0) {
+			NEXT_ARG();
+			if (get_time(&interval, *argv)) {
+				fprintf(stderr, "Illegal \"interval\"\n");
+				return -1;
+			}
+		} else if (strcmp(*argv, "help") == 0) {
+			explain();
+			return -1;
+		} else {
+			fprintf(stderr, "What is \"%s\"?\n", *argv);
+			explain();
+			return -1;
+		}
+		argc--; argv++;
+	}
+
+	tail = NLMSG_TAIL(n);
+	addattr_l(n, 1024, TCA_OPTIONS, NULL, 0);
+	if (limit)
+		addattr_l(n, 1024, TCA_CODEL_LIMIT, &limit, sizeof(limit));
+	if (minbytes)
+		addattr_l(n, 1024, TCA_CODEL_MINBYTES, &minbytes, sizeof(minbytes));
+	if (interval)
+		addattr_l(n, 1024, TCA_CODEL_INTERVAL, &interval, sizeof(interval));
+	if (target)
+		addattr_l(n, 1024, TCA_CODEL_TARGET, &target, sizeof(target));
+	tail->rta_len = (void *) NLMSG_TAIL(n) - (void *) tail;
+	return 0;
+}
+
+static int codel_print_opt(struct qdisc_util *qu, FILE *f, struct rtattr *opt)
+{
+	struct rtattr *tb[TCA_CODEL_MAX + 1];
+	unsigned limit;
+	unsigned interval;
+	unsigned target;
+	unsigned minbytes;
+	SPRINT_BUF(b1);
+
+	if (opt == NULL)
+		return 0;
+
+	parse_rtattr_nested(tb, TCA_CODEL_MAX, opt);
+
+	if (tb[TCA_CODEL_LIMIT] &&
+	    RTA_PAYLOAD(tb[TCA_CODEL_LIMIT]) >= sizeof(__u32)) {
+		limit = rta_getattr_u32(tb[TCA_CODEL_LIMIT]);
+		fprintf(f, "limit %up ", limit);
+	}
+	if (tb[TCA_CODEL_MINBYTES] &&
+	    RTA_PAYLOAD(tb[TCA_CODEL_MINBYTES]) >= sizeof(__u32)) {
+		minbytes = rta_getattr_u32(tb[TCA_CODEL_MINBYTES]);
+		fprintf(f, "minbytes %u ", minbytes);
+	}
+	if (tb[TCA_CODEL_TARGET] &&
+	    RTA_PAYLOAD(tb[TCA_CODEL_TARGET]) >= sizeof(__u32)) {
+		target = rta_getattr_u32(tb[TCA_CODEL_TARGET]);
+		fprintf(f, "target %s ", sprint_time(target, b1));
+	}
+	if (tb[TCA_CODEL_INTERVAL] &&
+	    RTA_PAYLOAD(tb[TCA_CODEL_INTERVAL]) >= sizeof(__u32)) {
+		interval = rta_getattr_u32(tb[TCA_CODEL_INTERVAL]);
+		fprintf(f, "interval %s ", sprint_time(interval, b1));
+	}
+
+	return 0;
+}
+
+static int codel_print_xstats(struct qdisc_util *qu, FILE *f,
+			      struct rtattr *xstats)
+{
+	struct tc_codel_xstats *st;
+	SPRINT_BUF(b1);
+
+	if (xstats == NULL)
+		return 0;
+
+	if (RTA_PAYLOAD(xstats) < sizeof(*st))
+		return -1;
+
+	st = RTA_DATA(xstats);
+	fprintf(f, "  count %u delay %s",
+		st->count, sprint_time(st->delay, b1));
+	if (st->drop_next)
+		fprintf(f, " drop_next %s", sprint_time(st->drop_next, b1));
+	return 0;
+
+}
+
+struct qdisc_util codel_qdisc_util = {
+	.id		= "codel",
+	.parse_qopt	= codel_parse_opt,
+	.print_qopt	= codel_print_opt,
+	.print_xstats	= codel_print_xstats,
+};