[linux-block.git] / net / sched / sch_mqprio.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 * net/sched/sch_mqprio.c
 *
 * Copyright (c) 2010 John Fastabend <john.r.fastabend@intel.com>
 */

#include <linux/types.h>
#include <linux/slab.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/module.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>
#include <net/sch_generic.h>
#include <net/pkt_cls.h>

#include "sch_mqprio_lib.h"

struct mqprio_sched {
	struct Qdisc		**qdiscs;
	u16 mode;
	u16 shaper;
	int hw_offload;
	u32 flags;
	u64 min_rate[TC_QOPT_MAX_QUEUE];
	u64 max_rate[TC_QOPT_MAX_QUEUE];
};

static int mqprio_enable_offload(struct Qdisc *sch,
				 const struct tc_mqprio_qopt *qopt,
				 struct netlink_ext_ack *extack)
{
	struct tc_mqprio_qopt_offload mqprio = {.qopt = *qopt};
	struct mqprio_sched *priv = qdisc_priv(sch);
	struct net_device *dev = qdisc_dev(sch);
	int err, i;

	switch (priv->mode) {
	case TC_MQPRIO_MODE_DCB:
		if (priv->shaper != TC_MQPRIO_SHAPER_DCB)
			return -EINVAL;
		break;
	case TC_MQPRIO_MODE_CHANNEL:
		mqprio.flags = priv->flags;
		if (priv->flags & TC_MQPRIO_F_MODE)
			mqprio.mode = priv->mode;
		if (priv->flags & TC_MQPRIO_F_SHAPER)
			mqprio.shaper = priv->shaper;
		if (priv->flags & TC_MQPRIO_F_MIN_RATE)
			for (i = 0; i < mqprio.qopt.num_tc; i++)
				mqprio.min_rate[i] = priv->min_rate[i];
		if (priv->flags & TC_MQPRIO_F_MAX_RATE)
			for (i = 0; i < mqprio.qopt.num_tc; i++)
				mqprio.max_rate[i] = priv->max_rate[i];
		break;
	default:
		return -EINVAL;
	}

	err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_MQPRIO,
					    &mqprio);
	if (err)
		return err;

	priv->hw_offload = mqprio.qopt.hw;

	return 0;
}

static void mqprio_disable_offload(struct Qdisc *sch)
{
	struct tc_mqprio_qopt_offload mqprio = { { 0 } };
	struct mqprio_sched *priv = qdisc_priv(sch);
	struct net_device *dev = qdisc_dev(sch);

	switch (priv->mode) {
	case TC_MQPRIO_MODE_DCB:
	case TC_MQPRIO_MODE_CHANNEL:
		dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_MQPRIO,
					      &mqprio);
		break;
	}
}

static void mqprio_destroy(struct Qdisc *sch)
{
	struct net_device *dev = qdisc_dev(sch);
	struct mqprio_sched *priv = qdisc_priv(sch);
	unsigned int ntx;

	if (priv->qdiscs) {
		for (ntx = 0;
		     ntx < dev->num_tx_queues && priv->qdiscs[ntx];
		     ntx++)
			qdisc_put(priv->qdiscs[ntx]);
		kfree(priv->qdiscs);
	}

	if (priv->hw_offload && dev->netdev_ops->ndo_setup_tc)
		mqprio_disable_offload(sch);
	else
		netdev_set_num_tc(dev, 0);
}

static int mqprio_parse_opt(struct net_device *dev, struct tc_mqprio_qopt *qopt,
			    const struct tc_mqprio_caps *caps,
			    struct netlink_ext_ack *extack)
{
	int err;

	/* Limit qopt->hw to maximum supported offload value.  Drivers have
	 * the option of overriding this later if they don't support the a
	 * given offload type.
	 */
	if (qopt->hw > TC_MQPRIO_HW_OFFLOAD_MAX)
		qopt->hw = TC_MQPRIO_HW_OFFLOAD_MAX;

	/* If hardware offload is requested, we will leave 3 options to the
	 * device driver:
	 * - populate the queue counts itself (and ignore what was requested)
	 * - validate the provided queue counts by itself (and apply them)
	 * - request queue count validation here (and apply them)
	 */
	err = mqprio_validate_qopt(dev, qopt,
				   !qopt->hw || caps->validate_queue_counts,
				   false, extack);
	if (err)
		return err;

	/* If ndo_setup_tc is not present then hardware doesn't support offload
	 * and we should return an error.
	 */
	if (qopt->hw && !dev->netdev_ops->ndo_setup_tc)
		return -EINVAL;

	return 0;
}

static const struct nla_policy mqprio_policy[TCA_MQPRIO_MAX + 1] = {
	[TCA_MQPRIO_MODE]	= { .len = sizeof(u16) },
	[TCA_MQPRIO_SHAPER]	= { .len = sizeof(u16) },
	[TCA_MQPRIO_MIN_RATE64]	= { .type = NLA_NESTED },
	[TCA_MQPRIO_MAX_RATE64]	= { .type = NLA_NESTED },
};

static int parse_attr(struct nlattr *tb[], int maxtype, struct nlattr *nla,
		      const struct nla_policy *policy, int len)
{
	int nested_len = nla_len(nla) - NLA_ALIGN(len);

	if (nested_len >= nla_attr_size(0))
		return nla_parse_deprecated(tb, maxtype,
					    nla_data(nla) + NLA_ALIGN(len),
					    nested_len, policy, NULL);

	memset(tb, 0, sizeof(struct nlattr *) * (maxtype + 1));
	return 0;
}

static int mqprio_parse_nlattr(struct Qdisc *sch, struct tc_mqprio_qopt *qopt,
			       struct nlattr *opt)
{
	struct mqprio_sched *priv = qdisc_priv(sch);
	struct nlattr *tb[TCA_MQPRIO_MAX + 1];
	struct nlattr *attr;
	int i, rem, err;

	err = parse_attr(tb, TCA_MQPRIO_MAX, opt, mqprio_policy,
			 sizeof(*qopt));
	if (err < 0)
		return err;

	if (!qopt->hw)
		return -EINVAL;

	if (tb[TCA_MQPRIO_MODE]) {
		priv->flags |= TC_MQPRIO_F_MODE;
		priv->mode = *(u16 *)nla_data(tb[TCA_MQPRIO_MODE]);
	}

	if (tb[TCA_MQPRIO_SHAPER]) {
		priv->flags |= TC_MQPRIO_F_SHAPER;
		priv->shaper = *(u16 *)nla_data(tb[TCA_MQPRIO_SHAPER]);
	}

	if (tb[TCA_MQPRIO_MIN_RATE64]) {
		if (priv->shaper != TC_MQPRIO_SHAPER_BW_RATE)
			return -EINVAL;
		i = 0;
		nla_for_each_nested(attr, tb[TCA_MQPRIO_MIN_RATE64],
				    rem) {
			if (nla_type(attr) != TCA_MQPRIO_MIN_RATE64)
				return -EINVAL;
			if (i >= qopt->num_tc)
				break;
			priv->min_rate[i] = *(u64 *)nla_data(attr);
			i++;
		}
		priv->flags |= TC_MQPRIO_F_MIN_RATE;
	}

	if (tb[TCA_MQPRIO_MAX_RATE64]) {
		if (priv->shaper != TC_MQPRIO_SHAPER_BW_RATE)
			return -EINVAL;
		i = 0;
		nla_for_each_nested(attr, tb[TCA_MQPRIO_MAX_RATE64],
				    rem) {
			if (nla_type(attr) != TCA_MQPRIO_MAX_RATE64)
				return -EINVAL;
			if (i >= qopt->num_tc)
				break;
			priv->max_rate[i] = *(u64 *)nla_data(attr);
			i++;
		}
		priv->flags |= TC_MQPRIO_F_MAX_RATE;
	}

	return 0;
}

static int mqprio_init(struct Qdisc *sch, struct nlattr *opt,
		       struct netlink_ext_ack *extack)
{
	struct net_device *dev = qdisc_dev(sch);
	struct mqprio_sched *priv = qdisc_priv(sch);
	struct netdev_queue *dev_queue;
	struct Qdisc *qdisc;
	int i, err = -EOPNOTSUPP;
	struct tc_mqprio_qopt *qopt = NULL;
	struct tc_mqprio_caps caps;
	int len;

	BUILD_BUG_ON(TC_MAX_QUEUE != TC_QOPT_MAX_QUEUE);
	BUILD_BUG_ON(TC_BITMASK != TC_QOPT_BITMASK);

	if (sch->parent != TC_H_ROOT)
		return -EOPNOTSUPP;

	if (!netif_is_multiqueue(dev))
		return -EOPNOTSUPP;

	/* make certain can allocate enough classids to handle queues */
	if (dev->num_tx_queues >= TC_H_MIN_PRIORITY)
		return -ENOMEM;

	if (!opt || nla_len(opt) < sizeof(*qopt))
		return -EINVAL;

	qdisc_offload_query_caps(dev, TC_SETUP_QDISC_MQPRIO,
				 &caps, sizeof(caps));

	qopt = nla_data(opt);
	if (mqprio_parse_opt(dev, qopt, &caps, extack))
		return -EINVAL;

	len = nla_len(opt) - NLA_ALIGN(sizeof(*qopt));
	if (len > 0) {
		err = mqprio_parse_nlattr(sch, qopt, opt);
		if (err)
			return err;
	}

	/* pre-allocate qdisc, attachment can't fail */
	priv->qdiscs = kcalloc(dev->num_tx_queues, sizeof(priv->qdiscs[0]),
			       GFP_KERNEL);
	if (!priv->qdiscs)
		return -ENOMEM;

	for (i = 0; i < dev->num_tx_queues; i++) {
		dev_queue = netdev_get_tx_queue(dev, i);
		qdisc = qdisc_create_dflt(dev_queue,
					  get_default_qdisc_ops(dev, i),
					  TC_H_MAKE(TC_H_MAJ(sch->handle),
						    TC_H_MIN(i + 1)), extack);
		if (!qdisc)
			return -ENOMEM;

		priv->qdiscs[i] = qdisc;
		qdisc->flags |= TCQ_F_ONETXQUEUE | TCQ_F_NOPARENT;
	}

	/* If the mqprio options indicate that hardware should own
	 * the queue mapping then run ndo_setup_tc otherwise use the
	 * supplied and verified mapping
	 */
	if (qopt->hw) {
		err = mqprio_enable_offload(sch, qopt, extack);
		if (err)
			return err;
	} else {
		netdev_set_num_tc(dev, qopt->num_tc);
		for (i = 0; i < qopt->num_tc; i++)
			netdev_set_tc_queue(dev, i,
					    qopt->count[i], qopt->offset[i]);
	}

	/* Always use supplied priority mappings */
	for (i = 0; i < TC_BITMASK + 1; i++)
		netdev_set_prio_tc_map(dev, i, qopt->prio_tc_map[i]);

	sch->flags |= TCQ_F_MQROOT;
	return 0;
}

static void mqprio_attach(struct Qdisc *sch)
{
	struct net_device *dev = qdisc_dev(sch);
	struct mqprio_sched *priv = qdisc_priv(sch);
	struct Qdisc *qdisc, *old;
	unsigned int ntx;

	/* Attach underlying qdisc */
	for (ntx = 0; ntx < dev->num_tx_queues; ntx++) {
		qdisc = priv->qdiscs[ntx];
		old = dev_graft_qdisc(qdisc->dev_queue, qdisc);
		if (old)
			qdisc_put(old);
		if (ntx < dev->real_num_tx_queues)
			qdisc_hash_add(qdisc, false);
	}
	kfree(priv->qdiscs);
	priv->qdiscs = NULL;
}

static struct netdev_queue *mqprio_queue_get(struct Qdisc *sch,
					     unsigned long cl)
{
	struct net_device *dev = qdisc_dev(sch);
	unsigned long ntx = cl - 1;

	if (ntx >= dev->num_tx_queues)
		return NULL;
	return netdev_get_tx_queue(dev, ntx);
}

static int mqprio_graft(struct Qdisc *sch, unsigned long cl, struct Qdisc *new,
			struct Qdisc **old, struct netlink_ext_ack *extack)
{
	struct net_device *dev = qdisc_dev(sch);
	struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);

	if (!dev_queue)
		return -EINVAL;

	if (dev->flags & IFF_UP)
		dev_deactivate(dev);

	*old = dev_graft_qdisc(dev_queue, new);

	if (new)
		new->flags |= TCQ_F_ONETXQUEUE | TCQ_F_NOPARENT;

	if (dev->flags & IFF_UP)
		dev_activate(dev);

	return 0;
}

static int dump_rates(struct mqprio_sched *priv,
		      struct tc_mqprio_qopt *opt, struct sk_buff *skb)
{
	struct nlattr *nest;
	int i;

	if (priv->flags & TC_MQPRIO_F_MIN_RATE) {
		nest = nla_nest_start_noflag(skb, TCA_MQPRIO_MIN_RATE64);
		if (!nest)
			goto nla_put_failure;

		for (i = 0; i < opt->num_tc; i++) {
			if (nla_put(skb, TCA_MQPRIO_MIN_RATE64,
				    sizeof(priv->min_rate[i]),
				    &priv->min_rate[i]))
				goto nla_put_failure;
		}
		nla_nest_end(skb, nest);
	}

	if (priv->flags & TC_MQPRIO_F_MAX_RATE) {
		nest = nla_nest_start_noflag(skb, TCA_MQPRIO_MAX_RATE64);
		if (!nest)
			goto nla_put_failure;

		for (i = 0; i < opt->num_tc; i++) {
			if (nla_put(skb, TCA_MQPRIO_MAX_RATE64,
				    sizeof(priv->max_rate[i]),
				    &priv->max_rate[i]))
				goto nla_put_failure;
		}
		nla_nest_end(skb, nest);
	}
	return 0;

nla_put_failure:
	nla_nest_cancel(skb, nest);
	return -1;
}

static int mqprio_dump(struct Qdisc *sch, struct sk_buff *skb)
{
	struct net_device *dev = qdisc_dev(sch);
	struct mqprio_sched *priv = qdisc_priv(sch);
	struct nlattr *nla = (struct nlattr *)skb_tail_pointer(skb);
	struct tc_mqprio_qopt opt = { 0 };
	struct Qdisc *qdisc;
	unsigned int ntx;

	sch->q.qlen = 0;
	gnet_stats_basic_sync_init(&sch->bstats);
	memset(&sch->qstats, 0, sizeof(sch->qstats));

	/* MQ supports lockless qdiscs. However, statistics accounting needs
	 * to account for all, none, or a mix of locked and unlocked child
	 * qdiscs. Percpu stats are added to counters in-band and locking
	 * qdisc totals are added at end.
	 */
	for (ntx = 0; ntx < dev->num_tx_queues; ntx++) {
		qdisc = netdev_get_tx_queue(dev, ntx)->qdisc_sleeping;
		spin_lock_bh(qdisc_lock(qdisc));

		gnet_stats_add_basic(&sch->bstats, qdisc->cpu_bstats,
				     &qdisc->bstats, false);
		gnet_stats_add_queue(&sch->qstats, qdisc->cpu_qstats,
				     &qdisc->qstats);
		sch->q.qlen += qdisc_qlen(qdisc);

		spin_unlock_bh(qdisc_lock(qdisc));
	}

	mqprio_qopt_reconstruct(dev, &opt);
	opt.hw = priv->hw_offload;

	if (nla_put(skb, TCA_OPTIONS, sizeof(opt), &opt))
		goto nla_put_failure;

	if ((priv->flags & TC_MQPRIO_F_MODE) &&
	    nla_put_u16(skb, TCA_MQPRIO_MODE, priv->mode))
		goto nla_put_failure;

	if ((priv->flags & TC_MQPRIO_F_SHAPER) &&
	    nla_put_u16(skb, TCA_MQPRIO_SHAPER, priv->shaper))
		goto nla_put_failure;

	if ((priv->flags & TC_MQPRIO_F_MIN_RATE ||
	     priv->flags & TC_MQPRIO_F_MAX_RATE) &&
	    (dump_rates(priv, &opt, skb) != 0))
		goto nla_put_failure;

	return nla_nest_end(skb, nla);
nla_put_failure:
	nlmsg_trim(skb, nla);
	return -1;
}

static struct Qdisc *mqprio_leaf(struct Qdisc *sch, unsigned long cl)
{
	struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);

	if (!dev_queue)
		return NULL;

	return dev_queue->qdisc_sleeping;
}

static unsigned long mqprio_find(struct Qdisc *sch, u32 classid)
{
	struct net_device *dev = qdisc_dev(sch);
	unsigned int ntx = TC_H_MIN(classid);

	/* There are essentially two regions here that have valid classid
	 * values. The first region will have a classid value of 1 through
	 * num_tx_queues. All of these are backed by actual Qdiscs.
	 */
	if (ntx < TC_H_MIN_PRIORITY)
		return (ntx <= dev->num_tx_queues) ? ntx : 0;

	/* The second region represents the hardware traffic classes. These
	 * are represented by classid values of TC_H_MIN_PRIORITY through
	 * TC_H_MIN_PRIORITY + netdev_get_num_tc - 1
	 */
	return ((ntx - TC_H_MIN_PRIORITY) < netdev_get_num_tc(dev)) ? ntx : 0;
}

static int mqprio_dump_class(struct Qdisc *sch, unsigned long cl,
			 struct sk_buff *skb, struct tcmsg *tcm)
{
	if (cl < TC_H_MIN_PRIORITY) {
		struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);
		struct net_device *dev = qdisc_dev(sch);
		int tc = netdev_txq_to_tc(dev, cl - 1);

		tcm->tcm_parent = (tc < 0) ? 0 :
			TC_H_MAKE(TC_H_MAJ(sch->handle),
				  TC_H_MIN(tc + TC_H_MIN_PRIORITY));
		tcm->tcm_info = dev_queue->qdisc_sleeping->handle;
	} else {
		tcm->tcm_parent = TC_H_ROOT;
		tcm->tcm_info = 0;
	}
	tcm->tcm_handle |= TC_H_MIN(cl);
	return 0;
}

static int mqprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
				   struct gnet_dump *d)
	__releases(d->lock)
	__acquires(d->lock)
{
	if (cl >= TC_H_MIN_PRIORITY) {
		int i;
		__u32 qlen;
		struct gnet_stats_queue qstats = {0};
		struct gnet_stats_basic_sync bstats;
		struct net_device *dev = qdisc_dev(sch);
		struct netdev_tc_txq tc = dev->tc_to_txq[cl & TC_BITMASK];

		gnet_stats_basic_sync_init(&bstats);
		/* Drop lock here it will be reclaimed before touching
		 * statistics this is required because the d->lock we
		 * hold here is the look on dev_queue->qdisc_sleeping
		 * also acquired below.
		 */
		if (d->lock)
			spin_unlock_bh(d->lock);

		for (i = tc.offset; i < tc.offset + tc.count; i++) {
			struct netdev_queue *q = netdev_get_tx_queue(dev, i);
			struct Qdisc *qdisc = rtnl_dereference(q->qdisc);

			spin_lock_bh(qdisc_lock(qdisc));

			gnet_stats_add_basic(&bstats, qdisc->cpu_bstats,
					     &qdisc->bstats, false);
			gnet_stats_add_queue(&qstats, qdisc->cpu_qstats,
					     &qdisc->qstats);
			sch->q.qlen += qdisc_qlen(qdisc);

			spin_unlock_bh(qdisc_lock(qdisc));
		}
		qlen = qdisc_qlen(sch) + qstats.qlen;

		/* Reclaim root sleeping lock before completing stats */
		if (d->lock)
			spin_lock_bh(d->lock);
		if (gnet_stats_copy_basic(d, NULL, &bstats, false) < 0 ||
		    gnet_stats_copy_queue(d, NULL, &qstats, qlen) < 0)
			return -1;
	} else {
		struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);

		sch = dev_queue->qdisc_sleeping;
		if (gnet_stats_copy_basic(d, sch->cpu_bstats,
					  &sch->bstats, true) < 0 ||
		    qdisc_qstats_copy(d, sch) < 0)
			return -1;
	}
	return 0;
}

static void mqprio_walk(struct Qdisc *sch, struct qdisc_walker *arg)
{
	struct net_device *dev = qdisc_dev(sch);
	unsigned long ntx;

	if (arg->stop)
		return;

	/* Walk hierarchy with a virtual class per tc */
	arg->count = arg->skip;
	for (ntx = arg->skip; ntx < netdev_get_num_tc(dev); ntx++) {
		if (!tc_qdisc_stats_dump(sch, ntx + TC_H_MIN_PRIORITY, arg))
			return;
	}

	/* Pad the values and skip over unused traffic classes */
	if (ntx < TC_MAX_QUEUE) {
		arg->count = TC_MAX_QUEUE;
		ntx = TC_MAX_QUEUE;
	}

	/* Reset offset, sort out remaining per-queue qdiscs */
	for (ntx -= TC_MAX_QUEUE; ntx < dev->num_tx_queues; ntx++) {
		if (arg->fn(sch, ntx + 1, arg) < 0) {
			arg->stop = 1;
			return;
		}
		arg->count++;
	}
}

static struct netdev_queue *mqprio_select_queue(struct Qdisc *sch,
						struct tcmsg *tcm)
{
	return mqprio_queue_get(sch, TC_H_MIN(tcm->tcm_parent));
}

static const struct Qdisc_class_ops mqprio_class_ops = {
	.graft		= mqprio_graft,
	.leaf		= mqprio_leaf,
	.find		= mqprio_find,
	.walk		= mqprio_walk,
	.dump		= mqprio_dump_class,
	.dump_stats	= mqprio_dump_class_stats,
	.select_queue	= mqprio_select_queue,
};

static struct Qdisc_ops mqprio_qdisc_ops __read_mostly = {
	.cl_ops		= &mqprio_class_ops,
	.id		= "mqprio",
	.priv_size	= sizeof(struct mqprio_sched),
	.init		= mqprio_init,
	.destroy	= mqprio_destroy,
	.attach		= mqprio_attach,
	.change_real_num_tx = mq_change_real_num_tx,
	.dump		= mqprio_dump,
	.owner		= THIS_MODULE,
};

static int __init mqprio_module_init(void)
{
	return register_qdisc(&mqprio_qdisc_ops);
}

static void __exit mqprio_module_exit(void)
{
	unregister_qdisc(&mqprio_qdisc_ops);
}

module_init(mqprio_module_init);
module_exit(mqprio_module_exit);

MODULE_LICENSE("GPL");
Commit	Line	Data
d2912cb1	1	// SPDX-License-Identifier: GPL-2.0-only
b8970f0b JF	2	/*
	3	* net/sched/sch_mqprio.c
	4	*
	5	* Copyright (c) 2010 John Fastabend <john.r.fastabend@intel.com>
b8970f0b JF	6	*/
	7
	8	#include <linux/types.h>
	9	#include <linux/slab.h>
	10	#include <linux/kernel.h>
	11	#include <linux/string.h>
	12	#include <linux/errno.h>
	13	#include <linux/skbuff.h>
3a9a231d	14	#include <linux/module.h>
b8970f0b JF	15	#include <net/netlink.h>
	16	#include <net/pkt_sched.h>
	17	#include <net/sch_generic.h>
4e8b86c0	18	#include <net/pkt_cls.h>
b8970f0b	19
1dfe086d VO	20	#include "sch_mqprio_lib.h"
1dfe086d VO	21
b8970f0b JF	22	struct mqprio_sched {
b8970f0b JF	23	struct Qdisc **qdiscs;
4e8b86c0 AN	24	u16 mode;
4e8b86c0 AN	25	u16 shaper;
2026fecf	26	int hw_offload;
4e8b86c0 AN	27	u32 flags;
	28	u64 min_rate[TC_QOPT_MAX_QUEUE];
	29	u64 max_rate[TC_QOPT_MAX_QUEUE];
b8970f0b JF	30	};
b8970f0b JF	31
5cfb45e2	32	static int mqprio_enable_offload(struct Qdisc *sch,
d404959f VO	33	const struct tc_mqprio_qopt *qopt,
d404959f VO	34	struct netlink_ext_ack *extack)
5cfb45e2 VO	35	{
	36	struct tc_mqprio_qopt_offload mqprio = {.qopt = *qopt};
	37	struct mqprio_sched *priv = qdisc_priv(sch);
	38	struct net_device *dev = qdisc_dev(sch);
	39	int err, i;
	40
	41	switch (priv->mode) {
	42	case TC_MQPRIO_MODE_DCB:
	43	if (priv->shaper != TC_MQPRIO_SHAPER_DCB)
	44	return -EINVAL;
	45	break;
	46	case TC_MQPRIO_MODE_CHANNEL:
	47	mqprio.flags = priv->flags;
	48	if (priv->flags & TC_MQPRIO_F_MODE)
	49	mqprio.mode = priv->mode;
	50	if (priv->flags & TC_MQPRIO_F_SHAPER)
	51	mqprio.shaper = priv->shaper;
	52	if (priv->flags & TC_MQPRIO_F_MIN_RATE)
	53	for (i = 0; i < mqprio.qopt.num_tc; i++)
	54	mqprio.min_rate[i] = priv->min_rate[i];
	55	if (priv->flags & TC_MQPRIO_F_MAX_RATE)
	56	for (i = 0; i < mqprio.qopt.num_tc; i++)
	57	mqprio.max_rate[i] = priv->max_rate[i];
	58	break;
	59	default:
	60	return -EINVAL;
	61	}
	62
	63	err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_MQPRIO,
	64	&mqprio);
	65	if (err)
	66	return err;
	67
	68	priv->hw_offload = mqprio.qopt.hw;
	69
	70	return 0;
	71	}
	72
	73	static void mqprio_disable_offload(struct Qdisc *sch)
	74	{
	75	struct tc_mqprio_qopt_offload mqprio = { { 0 } };
	76	struct mqprio_sched *priv = qdisc_priv(sch);
	77	struct net_device *dev = qdisc_dev(sch);
	78
	79	switch (priv->mode) {
	80	case TC_MQPRIO_MODE_DCB:
	81	case TC_MQPRIO_MODE_CHANNEL:
	82	dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_MQPRIO,
	83	&mqprio);
	84	break;
	85	}
	86	}
	87
b8970f0b JF	88	static void mqprio_destroy(struct Qdisc *sch)
	89	{
	90	struct net_device *dev = qdisc_dev(sch);
	91	struct mqprio_sched *priv = qdisc_priv(sch);
	92	unsigned int ntx;
	93
ac7100ba BH	94	if (priv->qdiscs) {
	95	for (ntx = 0;
	96	ntx < dev->num_tx_queues && priv->qdiscs[ntx];
	97	ntx++)
86bd446b	98	qdisc_put(priv->qdiscs[ntx]);
ac7100ba BH	99	kfree(priv->qdiscs);
ac7100ba BH	100	}
b8970f0b	101
5cfb45e2 VO	102	if (priv->hw_offload && dev->netdev_ops->ndo_setup_tc)
	103	mqprio_disable_offload(sch);
	104	else
b8970f0b	105	netdev_set_num_tc(dev, 0);
b8970f0b JF	106	}
b8970f0b JF	107
19278d76	108	static int mqprio_parse_opt(struct net_device dev, struct tc_mqprio_qopt qopt,
d404959f VO	109	const struct tc_mqprio_caps *caps,
d404959f VO	110	struct netlink_ext_ack *extack)
b8970f0b	111	{
1dfe086d	112	int err;
b8970f0b	113
2026fecf AD	114	/* Limit qopt->hw to maximum supported offload value. Drivers have
	115	* the option of overriding this later if they don't support the a
	116	* given offload type.
	117	*/
	118	if (qopt->hw > TC_MQPRIO_HW_OFFLOAD_MAX)
	119	qopt->hw = TC_MQPRIO_HW_OFFLOAD_MAX;
b8970f0b	120
19278d76 VO	121	/* If hardware offload is requested, we will leave 3 options to the
	122	* device driver:
	123	* - populate the queue counts itself (and ignore what was requested)
	124	* - validate the provided queue counts by itself (and apply them)
	125	* - request queue count validation here (and apply them)
b8970f0b	126	*/
1dfe086d VO	127	err = mqprio_validate_qopt(dev, qopt,
	128	!qopt->hw \|\| caps->validate_queue_counts,
	129	false, extack);
	130	if (err)
	131	return err;
b8970f0b	132
19278d76 VO	133	/* If ndo_setup_tc is not present then hardware doesn't support offload
	134	* and we should return an error.
	135	*/
	136	if (qopt->hw && !dev->netdev_ops->ndo_setup_tc)
	137	return -EINVAL;
	138
b8970f0b JF	139	return 0;
	140	}
	141
4e8b86c0 AN	142	static const struct nla_policy mqprio_policy[TCA_MQPRIO_MAX + 1] = {
	143	[TCA_MQPRIO_MODE] = { .len = sizeof(u16) },
	144	[TCA_MQPRIO_SHAPER] = { .len = sizeof(u16) },
	145	[TCA_MQPRIO_MIN_RATE64] = { .type = NLA_NESTED },
	146	[TCA_MQPRIO_MAX_RATE64] = { .type = NLA_NESTED },
	147	};
	148
	149	static int parse_attr(struct nlattr tb[], int maxtype, struct nlattr nla,
	150	const struct nla_policy *policy, int len)
	151	{
	152	int nested_len = nla_len(nla) - NLA_ALIGN(len);
	153
	154	if (nested_len >= nla_attr_size(0))
8cb08174 JB	155	return nla_parse_deprecated(tb, maxtype,
	156	nla_data(nla) + NLA_ALIGN(len),
	157	nested_len, policy, NULL);
4e8b86c0 AN	158
	159	memset(tb, 0, sizeof(struct nlattr ) (maxtype + 1));
	160	return 0;
	161	}
	162
feb2cf3d VO	163	static int mqprio_parse_nlattr(struct Qdisc sch, struct tc_mqprio_qopt qopt,
	164	struct nlattr *opt)
	165	{
	166	struct mqprio_sched *priv = qdisc_priv(sch);
	167	struct nlattr *tb[TCA_MQPRIO_MAX + 1];
	168	struct nlattr *attr;
	169	int i, rem, err;
	170
	171	err = parse_attr(tb, TCA_MQPRIO_MAX, opt, mqprio_policy,
	172	sizeof(*qopt));
	173	if (err < 0)
	174	return err;
	175
	176	if (!qopt->hw)
	177	return -EINVAL;
	178
	179	if (tb[TCA_MQPRIO_MODE]) {
	180	priv->flags \|= TC_MQPRIO_F_MODE;
	181	priv->mode = (u16 )nla_data(tb[TCA_MQPRIO_MODE]);
	182	}
	183
	184	if (tb[TCA_MQPRIO_SHAPER]) {
	185	priv->flags \|= TC_MQPRIO_F_SHAPER;
	186	priv->shaper = (u16 )nla_data(tb[TCA_MQPRIO_SHAPER]);
	187	}
	188
	189	if (tb[TCA_MQPRIO_MIN_RATE64]) {
	190	if (priv->shaper != TC_MQPRIO_SHAPER_BW_RATE)
	191	return -EINVAL;
	192	i = 0;
	193	nla_for_each_nested(attr, tb[TCA_MQPRIO_MIN_RATE64],
	194	rem) {
	195	if (nla_type(attr) != TCA_MQPRIO_MIN_RATE64)
	196	return -EINVAL;
	197	if (i >= qopt->num_tc)
	198	break;
	199	priv->min_rate[i] = (u64 )nla_data(attr);
	200	i++;
	201	}
	202	priv->flags \|= TC_MQPRIO_F_MIN_RATE;
	203	}
	204
	205	if (tb[TCA_MQPRIO_MAX_RATE64]) {
	206	if (priv->shaper != TC_MQPRIO_SHAPER_BW_RATE)
	207	return -EINVAL;
	208	i = 0;
	209	nla_for_each_nested(attr, tb[TCA_MQPRIO_MAX_RATE64],
	210	rem) {
	211	if (nla_type(attr) != TCA_MQPRIO_MAX_RATE64)
	212	return -EINVAL;
	213	if (i >= qopt->num_tc)
	214	break;
	215	priv->max_rate[i] = (u64 )nla_data(attr);
	216	i++;
	217	}
	218	priv->flags \|= TC_MQPRIO_F_MAX_RATE;
	219	}
	220
	221	return 0;
	222	}
	223
e63d7dfd AA	224	static int mqprio_init(struct Qdisc sch, struct nlattr opt,
e63d7dfd AA	225	struct netlink_ext_ack *extack)
b8970f0b JF	226	{
	227	struct net_device *dev = qdisc_dev(sch);
	228	struct mqprio_sched *priv = qdisc_priv(sch);
	229	struct netdev_queue *dev_queue;
	230	struct Qdisc *qdisc;
	231	int i, err = -EOPNOTSUPP;
	232	struct tc_mqprio_qopt *qopt = NULL;
19278d76	233	struct tc_mqprio_caps caps;
22ce97fe	234	int len;
b8970f0b JF	235
	236	BUILD_BUG_ON(TC_MAX_QUEUE != TC_QOPT_MAX_QUEUE);
	237	BUILD_BUG_ON(TC_BITMASK != TC_QOPT_BITMASK);
	238
	239	if (sch->parent != TC_H_ROOT)
	240	return -EOPNOTSUPP;
	241
	242	if (!netif_is_multiqueue(dev))
	243	return -EOPNOTSUPP;
	244
32302902 AD	245	/* make certain can allocate enough classids to handle queues */
	246	if (dev->num_tx_queues >= TC_H_MIN_PRIORITY)
	247	return -ENOMEM;
	248
7838f2ce	249	if (!opt \|\| nla_len(opt) < sizeof(*qopt))
b8970f0b JF	250	return -EINVAL;
b8970f0b JF	251
19278d76 VO	252	qdisc_offload_query_caps(dev, TC_SETUP_QDISC_MQPRIO,
	253	&caps, sizeof(caps));
	254
b8970f0b	255	qopt = nla_data(opt);
d404959f	256	if (mqprio_parse_opt(dev, qopt, &caps, extack))
b8970f0b JF	257	return -EINVAL;
b8970f0b JF	258
22ce97fe	259	len = nla_len(opt) - NLA_ALIGN(sizeof(*qopt));
4e8b86c0	260	if (len > 0) {
feb2cf3d VO	261	err = mqprio_parse_nlattr(sch, qopt, opt);
feb2cf3d VO	262	if (err)
4e8b86c0	263	return err;
4e8b86c0 AN	264	}
4e8b86c0 AN	265
b8970f0b JF	266	/* pre-allocate qdisc, attachment can't fail */
	267	priv->qdiscs = kcalloc(dev->num_tx_queues, sizeof(priv->qdiscs[0]),
	268	GFP_KERNEL);
87b60cfa ED	269	if (!priv->qdiscs)
87b60cfa ED	270	return -ENOMEM;
b8970f0b JF	271
	272	for (i = 0; i < dev->num_tx_queues; i++) {
	273	dev_queue = netdev_get_tx_queue(dev, i);
1f27cde3 ED	274	qdisc = qdisc_create_dflt(dev_queue,
1f27cde3 ED	275	get_default_qdisc_ops(dev, i),
b8970f0b	276	TC_H_MAKE(TC_H_MAJ(sch->handle),
a38a9882	277	TC_H_MIN(i + 1)), extack);
87b60cfa ED	278	if (!qdisc)
	279	return -ENOMEM;
	280
b8970f0b	281	priv->qdiscs[i] = qdisc;
4eaf3b84	282	qdisc->flags \|= TCQ_F_ONETXQUEUE \| TCQ_F_NOPARENT;
b8970f0b JF	283	}
	284
	285	/* If the mqprio options indicate that hardware should own
	286	* the queue mapping then run ndo_setup_tc otherwise use the
	287	* supplied and verified mapping
	288	*/
	289	if (qopt->hw) {
d404959f	290	err = mqprio_enable_offload(sch, qopt, extack);
b8970f0b	291	if (err)
87b60cfa	292	return err;
b8970f0b JF	293	} else {
	294	netdev_set_num_tc(dev, qopt->num_tc);
	295	for (i = 0; i < qopt->num_tc; i++)
	296	netdev_set_tc_queue(dev, i,
	297	qopt->count[i], qopt->offset[i]);
	298	}
	299
	300	/* Always use supplied priority mappings */
	301	for (i = 0; i < TC_BITMASK + 1; i++)
	302	netdev_set_prio_tc_map(dev, i, qopt->prio_tc_map[i]);
	303
	304	sch->flags \|= TCQ_F_MQROOT;
	305	return 0;
b8970f0b JF	306	}
	307
	308	static void mqprio_attach(struct Qdisc *sch)
	309	{
	310	struct net_device *dev = qdisc_dev(sch);
	311	struct mqprio_sched *priv = qdisc_priv(sch);
95dc1929	312	struct Qdisc qdisc, old;
b8970f0b JF	313	unsigned int ntx;
	314
	315	/* Attach underlying qdisc */
	316	for (ntx = 0; ntx < dev->num_tx_queues; ntx++) {
	317	qdisc = priv->qdiscs[ntx];
95dc1929 ED	318	old = dev_graft_qdisc(qdisc->dev_queue, qdisc);
95dc1929 ED	319	if (old)
86bd446b	320	qdisc_put(old);
95dc1929	321	if (ntx < dev->real_num_tx_queues)
49b49971	322	qdisc_hash_add(qdisc, false);
b8970f0b JF	323	}
	324	kfree(priv->qdiscs);
	325	priv->qdiscs = NULL;
	326	}
	327
	328	static struct netdev_queue mqprio_queue_get(struct Qdisc sch,
	329	unsigned long cl)
	330	{
	331	struct net_device *dev = qdisc_dev(sch);
32302902	332	unsigned long ntx = cl - 1;
b8970f0b JF	333
	334	if (ntx >= dev->num_tx_queues)
	335	return NULL;
	336	return netdev_get_tx_queue(dev, ntx);
	337	}
	338
	339	static int mqprio_graft(struct Qdisc sch, unsigned long cl, struct Qdisc new,
653d6fd6	340	struct Qdisc *old, struct netlink_ext_ack extack)
b8970f0b JF	341	{
	342	struct net_device *dev = qdisc_dev(sch);
	343	struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);
	344
	345	if (!dev_queue)
	346	return -EINVAL;
	347
	348	if (dev->flags & IFF_UP)
	349	dev_deactivate(dev);
	350
	351	*old = dev_graft_qdisc(dev_queue, new);
	352
1abbe139	353	if (new)
4eaf3b84	354	new->flags \|= TCQ_F_ONETXQUEUE \| TCQ_F_NOPARENT;
1abbe139	355
b8970f0b JF	356	if (dev->flags & IFF_UP)
	357	dev_activate(dev);
	358
	359	return 0;
	360	}
	361
4e8b86c0 AN	362	static int dump_rates(struct mqprio_sched *priv,
	363	struct tc_mqprio_qopt opt, struct sk_buff skb)
	364	{
	365	struct nlattr *nest;
	366	int i;
	367
	368	if (priv->flags & TC_MQPRIO_F_MIN_RATE) {
ae0be8de	369	nest = nla_nest_start_noflag(skb, TCA_MQPRIO_MIN_RATE64);
4e8b86c0 AN	370	if (!nest)
	371	goto nla_put_failure;
	372
	373	for (i = 0; i < opt->num_tc; i++) {
	374	if (nla_put(skb, TCA_MQPRIO_MIN_RATE64,
	375	sizeof(priv->min_rate[i]),
	376	&priv->min_rate[i]))
	377	goto nla_put_failure;
	378	}
	379	nla_nest_end(skb, nest);
	380	}
	381
	382	if (priv->flags & TC_MQPRIO_F_MAX_RATE) {
ae0be8de	383	nest = nla_nest_start_noflag(skb, TCA_MQPRIO_MAX_RATE64);
4e8b86c0 AN	384	if (!nest)
	385	goto nla_put_failure;
	386
	387	for (i = 0; i < opt->num_tc; i++) {
	388	if (nla_put(skb, TCA_MQPRIO_MAX_RATE64,
	389	sizeof(priv->max_rate[i]),
	390	&priv->max_rate[i]))
	391	goto nla_put_failure;
	392	}
	393	nla_nest_end(skb, nest);
	394	}
	395	return 0;
	396
	397	nla_put_failure:
	398	nla_nest_cancel(skb, nest);
	399	return -1;
	400	}
	401
b8970f0b JF	402	static int mqprio_dump(struct Qdisc sch, struct sk_buff skb)
	403	{
	404	struct net_device *dev = qdisc_dev(sch);
	405	struct mqprio_sched *priv = qdisc_priv(sch);
4e8b86c0	406	struct nlattr nla = (struct nlattr )skb_tail_pointer(skb);
144ce879	407	struct tc_mqprio_qopt opt = { 0 };
b8970f0b	408	struct Qdisc *qdisc;
9dd6ad67	409	unsigned int ntx;
b8970f0b JF	410
b8970f0b JF	411	sch->q.qlen = 0;
50dc9a85	412	gnet_stats_basic_sync_init(&sch->bstats);
b8970f0b JF	413	memset(&sch->qstats, 0, sizeof(sch->qstats));
b8970f0b JF	414
ce679e8d JF	415	/* MQ supports lockless qdiscs. However, statistics accounting needs
	416	* to account for all, none, or a mix of locked and unlocked child
	417	* qdiscs. Percpu stats are added to counters in-band and locking
	418	* qdisc totals are added at end.
	419	*/
	420	for (ntx = 0; ntx < dev->num_tx_queues; ntx++) {
	421	qdisc = netdev_get_tx_queue(dev, ntx)->qdisc_sleeping;
b8970f0b	422	spin_lock_bh(qdisc_lock(qdisc));
ce679e8d	423
29cbcd85 AD	424	gnet_stats_add_basic(&sch->bstats, qdisc->cpu_bstats,
29cbcd85 AD	425	&qdisc->bstats, false);
7361df46 SAS	426	gnet_stats_add_queue(&sch->qstats, qdisc->cpu_qstats,
	427	&qdisc->qstats);
	428	sch->q.qlen += qdisc_qlen(qdisc);
ce679e8d	429
b8970f0b JF	430	spin_unlock_bh(qdisc_lock(qdisc));
	431	}
	432
9dd6ad67	433	mqprio_qopt_reconstruct(dev, &opt);
2026fecf	434	opt.hw = priv->hw_offload;
b8970f0b	435
9f104c77	436	if (nla_put(skb, TCA_OPTIONS, sizeof(opt), &opt))
4e8b86c0 AN	437	goto nla_put_failure;
	438
	439	if ((priv->flags & TC_MQPRIO_F_MODE) &&
	440	nla_put_u16(skb, TCA_MQPRIO_MODE, priv->mode))
	441	goto nla_put_failure;
	442
	443	if ((priv->flags & TC_MQPRIO_F_SHAPER) &&
	444	nla_put_u16(skb, TCA_MQPRIO_SHAPER, priv->shaper))
	445	goto nla_put_failure;
	446
	447	if ((priv->flags & TC_MQPRIO_F_MIN_RATE \|\|
	448	priv->flags & TC_MQPRIO_F_MAX_RATE) &&
	449	(dump_rates(priv, &opt, skb) != 0))
1b34ec43	450	goto nla_put_failure;
b8970f0b	451
4e8b86c0	452	return nla_nest_end(skb, nla);
b8970f0b	453	nla_put_failure:
4e8b86c0	454	nlmsg_trim(skb, nla);
b8970f0b JF	455	return -1;
	456	}
	457
	458	static struct Qdisc mqprio_leaf(struct Qdisc sch, unsigned long cl)
	459	{
	460	struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);
	461
	462	if (!dev_queue)
	463	return NULL;
	464
	465	return dev_queue->qdisc_sleeping;
	466	}
	467
143976ce	468	static unsigned long mqprio_find(struct Qdisc *sch, u32 classid)
b8970f0b JF	469	{
	470	struct net_device *dev = qdisc_dev(sch);
	471	unsigned int ntx = TC_H_MIN(classid);
	472
32302902 AD	473	/* There are essentially two regions here that have valid classid
	474	* values. The first region will have a classid value of 1 through
	475	* num_tx_queues. All of these are backed by actual Qdiscs.
	476	*/
	477	if (ntx < TC_H_MIN_PRIORITY)
	478	return (ntx <= dev->num_tx_queues) ? ntx : 0;
	479
	480	/* The second region represents the hardware traffic classes. These
	481	* are represented by classid values of TC_H_MIN_PRIORITY through
	482	* TC_H_MIN_PRIORITY + netdev_get_num_tc - 1
	483	*/
	484	return ((ntx - TC_H_MIN_PRIORITY) < netdev_get_num_tc(dev)) ? ntx : 0;
b8970f0b JF	485	}
b8970f0b JF	486
b8970f0b JF	487	static int mqprio_dump_class(struct Qdisc *sch, unsigned long cl,
	488	struct sk_buff skb, struct tcmsg tcm)
	489	{
32302902 AD	490	if (cl < TC_H_MIN_PRIORITY) {
	491	struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);
	492	struct net_device *dev = qdisc_dev(sch);
	493	int tc = netdev_txq_to_tc(dev, cl - 1);
b8970f0b	494
32302902 AD	495	tcm->tcm_parent = (tc < 0) ? 0 :
	496	TC_H_MAKE(TC_H_MAJ(sch->handle),
	497	TC_H_MIN(tc + TC_H_MIN_PRIORITY));
	498	tcm->tcm_info = dev_queue->qdisc_sleeping->handle;
	499	} else {
b8970f0b JF	500	tcm->tcm_parent = TC_H_ROOT;
b8970f0b JF	501	tcm->tcm_info = 0;
b8970f0b JF	502	}
	503	tcm->tcm_handle \|= TC_H_MIN(cl);
	504	return 0;
	505	}
	506
	507	static int mqprio_dump_class_stats(struct Qdisc *sch, unsigned long cl,
ea18fd95	508	struct gnet_dump *d)
	509	__releases(d->lock)
	510	__acquires(d->lock)
b8970f0b	511	{
32302902	512	if (cl >= TC_H_MIN_PRIORITY) {
b8970f0b	513	int i;
7361df46	514	__u32 qlen;
b8970f0b	515	struct gnet_stats_queue qstats = {0};
50dc9a85	516	struct gnet_stats_basic_sync bstats;
32302902 AD	517	struct net_device *dev = qdisc_dev(sch);
32302902 AD	518	struct netdev_tc_txq tc = dev->tc_to_txq[cl & TC_BITMASK];
b8970f0b	519
50dc9a85	520	gnet_stats_basic_sync_init(&bstats);
b8970f0b JF	521	/* Drop lock here it will be reclaimed before touching
	522	* statistics this is required because the d->lock we
	523	* hold here is the look on dev_queue->qdisc_sleeping
	524	* also acquired below.
	525	*/
edb09eb1 ED	526	if (d->lock)
edb09eb1 ED	527	spin_unlock_bh(d->lock);
b8970f0b JF	528
b8970f0b JF	529	for (i = tc.offset; i < tc.offset + tc.count; i++) {
46e5da40	530	struct netdev_queue *q = netdev_get_tx_queue(dev, i);
ce679e8d	531	struct Qdisc *qdisc = rtnl_dereference(q->qdisc);
46e5da40	532
b8970f0b	533	spin_lock_bh(qdisc_lock(qdisc));
14132690	534
29cbcd85 AD	535	gnet_stats_add_basic(&bstats, qdisc->cpu_bstats,
29cbcd85 AD	536	&qdisc->bstats, false);
7361df46 SAS	537	gnet_stats_add_queue(&qstats, qdisc->cpu_qstats,
	538	&qdisc->qstats);
	539	sch->q.qlen += qdisc_qlen(qdisc);
	540
b8970f0b JF	541	spin_unlock_bh(qdisc_lock(qdisc));
b8970f0b JF	542	}
7361df46	543	qlen = qdisc_qlen(sch) + qstats.qlen;
ce679e8d	544
b8970f0b	545	/* Reclaim root sleeping lock before completing stats */
edb09eb1 ED	546	if (d->lock)
edb09eb1 ED	547	spin_lock_bh(d->lock);
29cbcd85	548	if (gnet_stats_copy_basic(d, NULL, &bstats, false) < 0 \|\|
b0ab6f92	549	gnet_stats_copy_queue(d, NULL, &qstats, qlen) < 0)
b8970f0b JF	550	return -1;
	551	} else {
	552	struct netdev_queue *dev_queue = mqprio_queue_get(sch, cl);
	553
	554	sch = dev_queue->qdisc_sleeping;
29cbcd85 AD	555	if (gnet_stats_copy_basic(d, sch->cpu_bstats,
29cbcd85 AD	556	&sch->bstats, true) < 0 \|\|
5dd431b6	557	qdisc_qstats_copy(d, sch) < 0)
b8970f0b JF	558	return -1;
	559	}
	560	return 0;
	561	}
	562
	563	static void mqprio_walk(struct Qdisc sch, struct qdisc_walker arg)
	564	{
	565	struct net_device *dev = qdisc_dev(sch);
	566	unsigned long ntx;
	567
	568	if (arg->stop)
	569	return;
	570
	571	/* Walk hierarchy with a virtual class per tc */
	572	arg->count = arg->skip;
32302902	573	for (ntx = arg->skip; ntx < netdev_get_num_tc(dev); ntx++) {
e046fa89	574	if (!tc_qdisc_stats_dump(sch, ntx + TC_H_MIN_PRIORITY, arg))
32302902	575	return;
32302902 AD	576	}
	577
	578	/* Pad the values and skip over unused traffic classes */
	579	if (ntx < TC_MAX_QUEUE) {
	580	arg->count = TC_MAX_QUEUE;
	581	ntx = TC_MAX_QUEUE;
	582	}
	583
	584	/* Reset offset, sort out remaining per-queue qdiscs */
	585	for (ntx -= TC_MAX_QUEUE; ntx < dev->num_tx_queues; ntx++) {
b8970f0b JF	586	if (arg->fn(sch, ntx + 1, arg) < 0) {
b8970f0b JF	587	arg->stop = 1;
32302902	588	return;
b8970f0b JF	589	}
	590	arg->count++;
	591	}
	592	}
	593
0f7787b4 JSP	594	static struct netdev_queue mqprio_select_queue(struct Qdisc sch,
	595	struct tcmsg *tcm)
	596	{
	597	return mqprio_queue_get(sch, TC_H_MIN(tcm->tcm_parent));
	598	}
	599
b8970f0b JF	600	static const struct Qdisc_class_ops mqprio_class_ops = {
	601	.graft = mqprio_graft,
	602	.leaf = mqprio_leaf,
143976ce	603	.find = mqprio_find,
b8970f0b JF	604	.walk = mqprio_walk,
	605	.dump = mqprio_dump_class,
	606	.dump_stats = mqprio_dump_class_stats,
0f7787b4	607	.select_queue = mqprio_select_queue,
b8970f0b JF	608	};
b8970f0b JF	609
ea18fd95	610	static struct Qdisc_ops mqprio_qdisc_ops __read_mostly = {
b8970f0b JF	611	.cl_ops = &mqprio_class_ops,
	612	.id = "mqprio",
	613	.priv_size = sizeof(struct mqprio_sched),
	614	.init = mqprio_init,
	615	.destroy = mqprio_destroy,
	616	.attach = mqprio_attach,
f7116fb4	617	.change_real_num_tx = mq_change_real_num_tx,
b8970f0b JF	618	.dump = mqprio_dump,
	619	.owner = THIS_MODULE,
	620	};
	621
	622	static int __init mqprio_module_init(void)
	623	{
	624	return register_qdisc(&mqprio_qdisc_ops);
	625	}
	626
	627	static void __exit mqprio_module_exit(void)
	628	{
	629	unregister_qdisc(&mqprio_qdisc_ops);
	630	}
	631
	632	module_init(mqprio_module_init);
	633	module_exit(mqprio_module_exit);
	634
	635	MODULE_LICENSE("GPL");