busy_poll.h 4.4 KB
Newer Older
E
Eliezer Tamir 已提交
1
/*
2
 * net busy poll support
E
Eliezer Tamir 已提交
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 * Copyright(c) 2013 Intel Corporation.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * This program is distributed in the hope it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program; if not, write to the Free Software Foundation, Inc.,
 * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * Author: Eliezer Tamir
 *
 * Contact Information:
 * e1000-devel Mailing List <e1000-devel@lists.sourceforge.net>
 */

24 25
#ifndef _LINUX_NET_BUSY_POLL_H
#define _LINUX_NET_BUSY_POLL_H
E
Eliezer Tamir 已提交
26 27 28 29

#include <linux/netdevice.h>
#include <net/ip.h>

30
#ifdef CONFIG_NET_RX_BUSY_POLL
E
Eliezer Tamir 已提交
31 32

struct napi_struct;
33 34
extern unsigned int sysctl_net_busy_read __read_mostly;
extern unsigned int sysctl_net_busy_poll __read_mostly;
E
Eliezer Tamir 已提交
35 36 37 38 39

/* return values from ndo_ll_poll */
#define LL_FLUSH_FAILED		-1
#define LL_FLUSH_BUSY		-2

40
static inline bool net_busy_loop_on(void)
41
{
42
	return sysctl_net_busy_poll;
43 44
}

45 46
/* a wrapper to make debug_smp_processor_id() happy
 * we can use sched_clock() because we don't care much about precision
47
 * we only care that the average is bounded
48 49
 */
#ifdef CONFIG_DEBUG_PREEMPT
50
static inline u64 busy_loop_us_clock(void)
51 52 53 54 55 56 57
{
	u64 rc;

	preempt_disable_notrace();
	rc = sched_clock();
	preempt_enable_no_resched_notrace();

58
	return rc >> 10;
59 60
}
#else /* CONFIG_DEBUG_PREEMPT */
61
static inline u64 busy_loop_us_clock(void)
62
{
63
	return sched_clock() >> 10;
64 65 66
}
#endif /* CONFIG_DEBUG_PREEMPT */

67
static inline unsigned long sk_busy_loop_end_time(struct sock *sk)
E
Eliezer Tamir 已提交
68
{
69
	return busy_loop_us_clock() + ACCESS_ONCE(sk->sk_ll_usec);
70
}
71

72 73
/* in poll/select we use the global sysctl_net_ll_poll value */
static inline unsigned long busy_loop_end_time(void)
74
{
75
	return busy_loop_us_clock() + ACCESS_ONCE(sysctl_net_busy_poll);
E
Eliezer Tamir 已提交
76 77
}

78
static inline bool sk_can_busy_loop(struct sock *sk)
E
Eliezer Tamir 已提交
79
{
80
	return sk->sk_ll_usec && sk->sk_napi_id &&
E
Eliezer Tamir 已提交
81 82 83
	       !need_resched() && !signal_pending(current);
}

84 85

static inline bool busy_loop_timeout(unsigned long end_time)
E
Eliezer Tamir 已提交
86
{
87
	unsigned long now = busy_loop_us_clock();
88

89
	return time_after(now, end_time);
E
Eliezer Tamir 已提交
90 91
}

92 93 94
/* when used in sock_poll() nonblock is known at compile time to be true
 * so the loop and end_time will be optimized out
 */
95
static inline bool sk_busy_loop(struct sock *sk, int nonblock)
E
Eliezer Tamir 已提交
96
{
97
	unsigned long end_time = !nonblock ? sk_busy_loop_end_time(sk) : 0;
E
Eliezer Tamir 已提交
98 99 100 101 102 103 104 105 106 107 108 109 110 111 112
	const struct net_device_ops *ops;
	struct napi_struct *napi;
	int rc = false;

	/*
	 * rcu read lock for napi hash
	 * bh so we don't race with net_rx_action
	 */
	rcu_read_lock_bh();

	napi = napi_by_id(sk->sk_napi_id);
	if (!napi)
		goto out;

	ops = napi->dev->netdev_ops;
113
	if (!ops->ndo_busy_poll)
E
Eliezer Tamir 已提交
114 115 116
		goto out;

	do {
117
		rc = ops->ndo_busy_poll(napi);
E
Eliezer Tamir 已提交
118 119 120 121 122 123 124

		if (rc == LL_FLUSH_FAILED)
			break; /* permanent failure */

		if (rc > 0)
			/* local bh are disabled so it is ok to use _BH */
			NET_ADD_STATS_BH(sock_net(sk),
125
					 LINUX_MIB_BUSYPOLLRXPACKETS, rc);
126
		cpu_relax();
E
Eliezer Tamir 已提交
127

128
	} while (!nonblock && skb_queue_empty(&sk->sk_receive_queue) &&
129
		 !need_resched() && !busy_loop_timeout(end_time));
E
Eliezer Tamir 已提交
130 131 132 133 134 135 136 137

	rc = !skb_queue_empty(&sk->sk_receive_queue);
out:
	rcu_read_unlock_bh();
	return rc;
}

/* used in the NIC receive handler to mark the skb */
138 139
static inline void skb_mark_napi_id(struct sk_buff *skb,
				    struct napi_struct *napi)
E
Eliezer Tamir 已提交
140 141 142 143 144
{
	skb->napi_id = napi->napi_id;
}

/* used in the protocol hanlder to propagate the napi_id to the socket */
145
static inline void sk_mark_napi_id(struct sock *sk, struct sk_buff *skb)
E
Eliezer Tamir 已提交
146 147 148 149
{
	sk->sk_napi_id = skb->napi_id;
}

150
#else /* CONFIG_NET_RX_BUSY_POLL */
151
static inline unsigned long net_busy_loop_on(void)
152 153 154
{
	return 0;
}
E
Eliezer Tamir 已提交
155

156
static inline unsigned long busy_loop_end_time(void)
E
Eliezer Tamir 已提交
157 158 159 160
{
	return 0;
}

161
static inline bool sk_can_busy_loop(struct sock *sk)
E
Eliezer Tamir 已提交
162 163 164 165
{
	return false;
}

166 167
static inline void skb_mark_napi_id(struct sk_buff *skb,
				    struct napi_struct *napi)
E
Eliezer Tamir 已提交
168 169 170
{
}

171
static inline void sk_mark_napi_id(struct sock *sk, struct sk_buff *skb)
E
Eliezer Tamir 已提交
172 173 174
{
}

175
static inline bool busy_loop_timeout(unsigned long end_time)
E
Eliezer Tamir 已提交
176
{
177
	return true;
E
Eliezer Tamir 已提交
178 179
}

180 181 182 183 184
static inline bool sk_busy_loop(struct sock *sk, int nonblock)
{
	return false;
}

185
#endif /* CONFIG_NET_RX_BUSY_POLL */
186
#endif /* _LINUX_NET_BUSY_POLL_H */