aboutsummaryrefslogtreecommitdiffstats
path: root/tools/perf/util/parse-events.c
blob: 35d04da38d6a156de599a5ca157c73fb5fe203ca (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316

#include "../perf.h"
#include "util.h"
#include "parse-options.h"
#include "parse-events.h"
#include "exec_cmd.h"
#include "string.h"

extern char *strcasestr(const char *haystack, const char *needle);

int					nr_counters;

struct perf_counter_attr		attrs[MAX_COUNTERS];

struct event_symbol {
	u8	type;
	u64	config;
	char	*symbol;
};

#define C(x, y) .type = PERF_TYPE_##x, .config = PERF_COUNT_##y
#define CR(x, y) .type = PERF_TYPE_##x, .config = y

static struct event_symbol event_symbols[] = {
  { C(HARDWARE, HW_CPU_CYCLES),		"cpu-cycles",		},
  { C(HARDWARE, HW_CPU_CYCLES),		"cycles",		},
  { C(HARDWARE, HW_INSTRUCTIONS),	"instructions",		},
  { C(HARDWARE, HW_CACHE_REFERENCES),	"cache-references",	},
  { C(HARDWARE, HW_CACHE_MISSES),	"cache-misses",		},
  { C(HARDWARE, HW_BRANCH_INSTRUCTIONS),"branch-instructions",	},
  { C(HARDWARE, HW_BRANCH_INSTRUCTIONS),"branches",		},
  { C(HARDWARE, HW_BRANCH_MISSES),	"branch-misses",	},
  { C(HARDWARE, HW_BUS_CYCLES),		"bus-cycles",		},

  { C(SOFTWARE, SW_CPU_CLOCK),		"cpu-clock",		},
  { C(SOFTWARE, SW_TASK_CLOCK),		"task-clock",		},
  { C(SOFTWARE, SW_PAGE_FAULTS),	"page-faults",		},
  { C(SOFTWARE, SW_PAGE_FAULTS),	"faults",		},
  { C(SOFTWARE, SW_PAGE_FAULTS_MIN),	"minor-faults",		},
  { C(SOFTWARE, SW_PAGE_FAULTS_MAJ),	"major-faults",		},
  { C(SOFTWARE, SW_CONTEXT_SWITCHES),	"context-switches",	},
  { C(SOFTWARE, SW_CONTEXT_SWITCHES),	"cs",			},
  { C(SOFTWARE, SW_CPU_MIGRATIONS),	"cpu-migrations",	},
  { C(SOFTWARE, SW_CPU_MIGRATIONS),	"migrations",		},
};

#define __PERF_COUNTER_FIELD(config, name) \
	((config & PERF_COUNTER_##name##_MASK) >> PERF_COUNTER_##name##_SHIFT)

#define PERF_COUNTER_RAW(config)	__PERF_COUNTER_FIELD(config, RAW)
#define PERF_COUNTER_CONFIG(config)	__PERF_COUNTER_FIELD(config, CONFIG)
#define PERF_COUNTER_TYPE(config)	__PERF_COUNTER_FIELD(config, TYPE)
#define PERF_COUNTER_ID(config)		__PERF_COUNTER_FIELD(config, EVENT)

static char *hw_event_names[] = {
	"cycles",
	"instructions",
	"cache-references",
	"cache-misses",
	"branches",
	"branch-misses",
	"bus-cycles",
};

static char *sw_event_names[] = {
	"cpu-clock-msecs",
	"task-clock-msecs",
	"page-faults",
	"context-switches",
	"CPU-migrations",
	"minor-faults",
	"major-faults",
};

#define MAX_ALIASES 8

static char *hw_cache [][MAX_ALIASES] = {
	{ "L1-data"		, "l1-d", "l1d"					},
	{ "L1-instruction"	, "l1-i", "l1i"					},
	{ "L2"			, "l2"						},
	{ "Data-TLB"		, "dtlb", "d-tlb"				},
	{ "Instruction-TLB"	, "itlb", "i-tlb"				},
	{ "Branch"		, "bpu" , "btb", "bpc"				},
};

static char *hw_cache_op [][MAX_ALIASES] = {
	{ "Load"		, "read"					},
	{ "Store"		, "write"					},
	{ "Prefetch"		, "speculative-read", "speculative-load"	},
};

static char *hw_cache_result [][MAX_ALIASES] = {
	{ "Reference"		, "ops", "access"				},
	{ "Miss"								},
};

char *event_name(int counter)
{
	u64 config = attrs[counter].config;
	int type = attrs[counter].type;
	static char buf[32];

	if (attrs[counter].type == PERF_TYPE_RAW) {
		sprintf(buf, "raw 0x%llx", config);
		return buf;
	}

	switch (type) {
	case PERF_TYPE_HARDWARE:
		if (config < PERF_COUNT_HW_MAX)
			return hw_event_names[config];
		return "unknown-hardware";

	case PERF_TYPE_HW_CACHE: {
		u8 cache_type, cache_op, cache_result;
		static char name[100];

		cache_type   = (config >>  0) & 0xff;
		if (cache_type > PERF_COUNT_HW_CACHE_MAX)
			return "unknown-ext-hardware-cache-type";

		cache_op     = (config >>  8) & 0xff;
		if (cache_op > PERF_COUNT_HW_CACHE_OP_MAX)
			return "unknown-ext-hardware-cache-op";

		cache_result = (config >> 16) & 0xff;
		if (cache_result > PERF_COUNT_HW_CACHE_RESULT_MAX)
			return "unknown-ext-hardware-cache-result";

		sprintf(name, "%s-Cache-%s-%ses",
			hw_cache[cache_type][0],
			hw_cache_op[cache_op][0],
			hw_cache_result[cache_result][0]);

		return name;
	}

	case PERF_TYPE_SOFTWARE:
		if (config < PERF_COUNT_SW_MAX)
			return sw_event_names[config];
		return "unknown-software";

	default:
		break;
	}

	return "unknown";
}

static int parse_aliases(const char *str, char *names[][MAX_ALIASES], int size)
{
	int i, j;

	for (i = 0; i < size; i++) {
		for (j = 0; j < MAX_ALIASES; j++) {
			if (!names[i][j])
				break;
			if (strcasestr(str, names[i][j]))
				return i;
		}
	}

	return -1;
}

static int parse_generic_hw_symbols(const char *str, struct perf_counter_attr *attr)
{
	int cache_type = -1, cache_op = 0, cache_result = 0;

	cache_type = parse_aliases(str, hw_cache, PERF_COUNT_HW_CACHE_MAX);
	/*
	 * No fallback - if we cannot get a clear cache type
	 * then bail out:
	 */
	if (cache_type == -1)
		return -EINVAL;

	cache_op = parse_aliases(str, hw_cache_op, PERF_COUNT_HW_CACHE_OP_MAX);
	/*
	 * Fall back to reads:
	 */
	if (cache_op == -1)
		cache_op = PERF_COUNT_HW_CACHE_OP_READ;

	cache_result = parse_aliases(str, hw_cache_result,
					PERF_COUNT_HW_CACHE_RESULT_MAX);
	/*
	 * Fall back to accesses:
	 */
	if (cache_result == -1)
		cache_result = PERF_COUNT_HW_CACHE_RESULT_ACCESS;

	attr->config = cache_type | (cache_op << 8) | (cache_result << 16);
	attr->type = PERF_TYPE_HW_CACHE;

	return 0;
}

/*
 * Each event can have multiple symbolic names.
 * Symbolic names are (almost) exactly matched.
 */
static int parse_event_symbols(const char *str, struct perf_counter_attr *attr)
{
	u64 config, id;
	int type;
	unsigned int i;
	const char *sep, *pstr;

	if (str[0] == 'r' && hex2u64(str + 1, &config) > 0) {
		attr->type = PERF_TYPE_RAW;
		attr->config = config;

		return 0;
	}

	pstr = str;
	sep = strchr(pstr, ':');
	if (sep) {
		type = atoi(pstr);
		pstr = sep + 1;
		id = atoi(pstr);
		sep = strchr(pstr, ':');
		if (sep) {
			pstr = sep + 1;
			if (strchr(pstr, 'k'))
				attr->exclude_user = 1;
			if (strchr(pstr, 'u'))
				attr->exclude_kernel = 1;
		}
		attr->type = type;
		attr->config = id;

		return 0;
	}

	for (i = 0; i < ARRAY_SIZE(event_symbols); i++) {
		if (!strncmp(str, event_symbols[i].symbol,
			     strlen(event_symbols[i].symbol))) {

			attr->type = event_symbols[i].type;
			attr->config = event_symbols[i].config;

			return 0;
		}
	}

	return parse_generic_hw_symbols(str, attr);
}

int parse_events(const struct option *opt, const char *str, int unset)
{
	struct perf_counter_attr attr;
	int ret;

	memset(&attr, 0, sizeof(attr));
again:
	if (nr_counters == MAX_COUNTERS)
		return -1;

	ret = parse_event_symbols(str, &attr);
	if (ret < 0)
		return ret;

	attrs[nr_counters] = attr;
	nr_counters++;

	str = strstr(str, ",");
	if (str) {
		str++;
		goto again;
	}

	return 0;
}

static const char * const event_type_descriptors[] = {
	"",
	"Hardware event",
	"Software event",
	"Tracepoint event",
	"Hardware cache event",
};

/*
 * Print the help text for the event symbols:
 */
void print_events(void)
{
	struct event_symbol *syms = event_symbols;
	unsigned int i, type, prev_type = -1;

	fprintf(stderr, "\n");
	fprintf(stderr, "List of pre-defined events (to be used in -e):\n");

	for (i = 0; i < ARRAY_SIZE(event_symbols); i++, syms++) {
		type = syms->type + 1;
		if (type > ARRAY_SIZE(event_type_descriptors))
			type = 0;

		if (type != prev_type)
			fprintf(stderr, "\n");

		fprintf(stderr, "  %-30s [%s]\n", syms->symbol,
			event_type_descriptors[type]);

		prev_type = type;
	}

	fprintf(stderr, "\n");
	fprintf(stderr, "  %-30s [raw hardware event descriptor]\n",
		"rNNN");
	fprintf(stderr, "\n");

	exit(129);
}
'>1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613
/*
 * net/sched/sch_htb.c	Hierarchical token bucket, feed tree version
 *
 *		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:	Martin Devera, <devik@cdi.cz>
 *
 * Credits (in time order) for older HTB versions:
 *              Stef Coene <stef.coene@docum.org>
 *			HTB support at LARTC mailing list
 *		Ondrej Kraus, <krauso@barr.cz>
 *			found missing INIT_QDISC(htb)
 *		Vladimir Smelhaus, Aamer Akhter, Bert Hubert
 *			helped a lot to locate nasty class stall bug
 *		Andi Kleen, Jamal Hadi, Bert Hubert
 *			code review and helpful comments on shaping
 *		Tomasz Wrona, <tw@eter.tym.pl>
 *			created test case so that I was able to fix nasty bug
 *		Wilfried Weissmann
 *			spotted bug in dequeue code and helped with fix
 *		Jiri Fojtasek
 *			fixed requeue routine
 *		and many others. thanks.
 */
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/list.h>
#include <linux/compiler.h>
#include <linux/rbtree.h>
#include <linux/workqueue.h>
#include <linux/slab.h>
#include <net/netlink.h>
#include <net/sch_generic.h>
#include <net/pkt_sched.h>

/* HTB algorithm.
    Author: devik@cdi.cz
    ========================================================================
    HTB is like TBF with multiple classes. It is also similar to CBQ because
    it allows to assign priority to each class in hierarchy.
    In fact it is another implementation of Floyd's formal sharing.

    Levels:
    Each class is assigned level. Leaf has ALWAYS level 0 and root
    classes have level TC_HTB_MAXDEPTH-1. Interior nodes has level
    one less than their parent.
*/

static int htb_hysteresis __read_mostly = 0; /* whether to use mode hysteresis for speedup */
#define HTB_VER 0x30011		/* major must be matched with number suplied by TC as version */

#if HTB_VER >> 16 != TC_HTB_PROTOVER
#error "Mismatched sch_htb.c and pkt_sch.h"
#endif

/* Module parameter and sysfs export */
module_param    (htb_hysteresis, int, 0640);
MODULE_PARM_DESC(htb_hysteresis, "Hysteresis mode, less CPU load, less accurate");

static int htb_rate_est = 0; /* htb classes have a default rate estimator */
module_param(htb_rate_est, int, 0640);
MODULE_PARM_DESC(htb_rate_est, "setup a default rate estimator (4sec 16sec) for htb classes");

/* used internaly to keep status of single class */
enum htb_cmode {
	HTB_CANT_SEND,		/* class can't send and can't borrow */
	HTB_MAY_BORROW,		/* class can't send but may borrow */
	HTB_CAN_SEND		/* class can send */
};

struct htb_prio {
	union {
		struct rb_root	row;
		struct rb_root	feed;
	};
	struct rb_node	*ptr;
	/* When class changes from state 1->2 and disconnects from
	 * parent's feed then we lost ptr value and start from the
	 * first child again. Here we store classid of the
	 * last valid ptr (used when ptr is NULL).
	 */
	u32		last_ptr_id;
};

/* interior & leaf nodes; props specific to leaves are marked L:
 * To reduce false sharing, place mostly read fields at beginning,
 * and mostly written ones at the end.
 */
struct htb_class {
	struct Qdisc_class_common common;
	struct psched_ratecfg	rate;
	struct psched_ratecfg	ceil;
	s64			buffer, cbuffer;/* token bucket depth/rate */
	s64			mbuffer;	/* max wait time */
	u32			prio;		/* these two are used only by leaves... */
	int			quantum;	/* but stored for parent-to-leaf return */

	struct tcf_proto	*filter_list;	/* class attached filters */
	int			filter_cnt;
	int			refcnt;		/* usage count of this class */

	int			level;		/* our level (see above) */
	unsigned int		children;
	struct htb_class	*parent;	/* parent class */

	struct gnet_stats_rate_est64 rate_est;

	/*
	 * Written often fields
	 */
	struct gnet_stats_basic_packed bstats;
	struct gnet_stats_queue	qstats;
	struct tc_htb_xstats	xstats;	/* our special stats */

	/* token bucket parameters */
	s64			tokens, ctokens;/* current number of tokens */
	s64			t_c;		/* checkpoint time */

	union {
		struct htb_class_leaf {
			struct list_head drop_list;
			int		deficit[TC_HTB_MAXDEPTH];
			struct Qdisc	*q;
		} leaf;
		struct htb_class_inner {
			struct htb_prio clprio[TC_HTB_NUMPRIO];
		} inner;
	} un;
	s64			pq_key;

	int			prio_activity;	/* for which prios are we active */
	enum htb_cmode		cmode;		/* current mode of the class */
	struct rb_node		pq_node;	/* node for event queue */
	struct rb_node		node[TC_HTB_NUMPRIO];	/* node for self or feed tree */
};

struct htb_level {
	struct rb_root	wait_pq;
	struct htb_prio hprio[TC_HTB_NUMPRIO];
};

struct htb_sched {
	struct Qdisc_class_hash clhash;
	int			defcls;		/* class where unclassified flows go to */
	int			rate2quantum;	/* quant = rate / rate2quantum */

	/* filters for qdisc itself */
	struct tcf_proto	*filter_list;

#define HTB_WARN_TOOMANYEVENTS	0x1
	unsigned int		warned;	/* only one warning */
	int			direct_qlen;
	struct work_struct	work;

	/* non shaped skbs; let them go directly thru */
	struct sk_buff_head	direct_queue;
	long			direct_pkts;

	struct qdisc_watchdog	watchdog;

	s64			now;	/* cached dequeue time */
	struct list_head	drops[TC_HTB_NUMPRIO];/* active leaves (for drops) */

	/* time of nearest event per level (row) */
	s64			near_ev_cache[TC_HTB_MAXDEPTH];

	int			row_mask[TC_HTB_MAXDEPTH];

	struct htb_level	hlevel[TC_HTB_MAXDEPTH];
};

/* find class in global hash table using given handle */
static inline struct htb_class *htb_find(u32 handle, struct Qdisc *sch)
{
	struct htb_sched *q = qdisc_priv(sch);
	struct Qdisc_class_common *clc;

	clc = qdisc_class_find(&q->clhash, handle);
	if (clc == NULL)
		return NULL;
	return container_of(clc, struct htb_class, common);
}

/**
 * htb_classify - classify a packet into class
 *
 * It returns NULL if the packet should be dropped or -1 if the packet
 * should be passed directly thru. In all other cases leaf class is returned.
 * We allow direct class selection by classid in priority. The we examine
 * filters in qdisc and in inner nodes (if higher filter points to the inner
 * node). If we end up with classid MAJOR:0 we enqueue the skb into special
 * internal fifo (direct). These packets then go directly thru. If we still
 * have no valid leaf we try to use MAJOR:default leaf. It still unsuccessful
 * then finish and return direct queue.
 */
#define HTB_DIRECT ((struct htb_class *)-1L)

static struct htb_class *htb_classify(struct sk_buff *skb, struct Qdisc *sch,
				      int *qerr)
{
	struct htb_sched *q = qdisc_priv(sch);
	struct htb_class *cl;
	struct tcf_result res;
	struct tcf_proto *tcf;
	int result;

	/* allow to select class by setting skb->priority to valid classid;
	 * note that nfmark can be used too by attaching filter fw with no
	 * rules in it
	 */
	if (skb->priority == sch->handle)
		return HTB_DIRECT;	/* X:0 (direct flow) selected */
	cl = htb_find(skb->priority, sch);
	if (cl && cl->level == 0)
		return cl;

	*qerr = NET_XMIT_SUCCESS | __NET_XMIT_BYPASS;
	tcf = q->filter_list;
	while (tcf && (result = tc_classify(skb, tcf, &res)) >= 0) {
#ifdef CONFIG_NET_CLS_ACT
		switch (result) {
		case TC_ACT_QUEUED:
		case TC_ACT_STOLEN:
			*qerr = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN;
		case TC_ACT_SHOT:
			return NULL;
		}
#endif
		cl = (void *)res.class;
		if (!cl) {
			if (res.classid == sch->handle)
				return HTB_DIRECT;	/* X:0 (direct flow) */
			cl = htb_find(res.classid, sch);
			if (!cl)
				break;	/* filter selected invalid classid */
		}
		if (!cl->level)
			return cl;	/* we hit leaf; return it */

		/* we have got inner class; apply inner filter chain */
		tcf = cl->filter_list;
	}
	/* classification failed; try to use default class */
	cl = htb_find(TC_H_MAKE(TC_H_MAJ(sch->handle), q->defcls), sch);
	if (!cl || cl->level)
		return HTB_DIRECT;	/* bad default .. this is safe bet */
	return cl;
}

/**
 * htb_add_to_id_tree - adds class to the round robin list
 *
 * Routine adds class to the list (actually tree) sorted by classid.
 * Make sure that class is not already on such list for given prio.
 */
static void htb_add_to_id_tree(struct rb_root *root,
			       struct htb_class *cl, int prio)
{
	struct rb_node **p = &root->rb_node, *parent = NULL;

	while (*p) {
		struct htb_class *c;
		parent = *p;
		c = rb_entry(parent, struct htb_class, node[prio]);

		if (cl->common.classid > c->common.classid)
			p = &parent->rb_right;
		else
			p = &parent->rb_left;
	}
	rb_link_node(&cl->node[prio], parent, p);
	rb_insert_color(&cl->node[prio], root);
}

/**
 * htb_add_to_wait_tree - adds class to the event queue with delay
 *
 * The class is added to priority event queue to indicate that class will
 * change its mode in cl->pq_key microseconds. Make sure that class is not
 * already in the queue.
 */
static void htb_add_to_wait_tree(struct htb_sched *q,
				 struct htb_class *cl, s64 delay)
{
	struct rb_node **p = &q->hlevel[cl->level].wait_pq.rb_node, *parent = NULL;

	cl->pq_key = q->now + delay;
	if (cl->pq_key == q->now)
		cl->pq_key++;

	/* update the nearest event cache */
	if (q->near_ev_cache[cl->level] > cl->pq_key)
		q->near_ev_cache[cl->level] = cl->pq_key;

	while (*p) {
		struct htb_class *c;
		parent = *p;
		c = rb_entry(parent, struct htb_class, pq_node);
		if (cl->pq_key >= c->pq_key)
			p = &parent->rb_right;
		else
			p = &parent->rb_left;
	}
	rb_link_node(&cl->pq_node, parent, p);
	rb_insert_color(&cl->pq_node, &q->hlevel[cl->level].wait_pq);
}

/**
 * htb_next_rb_node - finds next node in binary tree
 *
 * When we are past last key we return NULL.
 * Average complexity is 2 steps per call.
 */
static inline void htb_next_rb_node(struct rb_node **n)
{
	*n = rb_next(*n);
}

/**
 * htb_add_class_to_row - add class to its row
 *
 * The class is added to row at priorities marked in mask.
 * It does nothing if mask == 0.
 */
static inline void htb_add_class_to_row(struct htb_sched *q,
					struct htb_class *cl, int mask)
{
	q->row_mask[cl->level] |= mask;
	while (mask) {
		int prio = ffz(~mask);
		mask &= ~(1 << prio);
		htb_add_to_id_tree(&q->hlevel[cl->level].hprio[prio].row, cl, prio);
	}
}

/* If this triggers, it is a bug in this code, but it need not be fatal */
static void htb_safe_rb_erase(struct rb_node *rb, struct rb_root *root)
{
	if (RB_EMPTY_NODE(rb)) {
		WARN_ON(1);
	} else {
		rb_erase(rb, root);
		RB_CLEAR_NODE(rb);
	}
}


/**
 * htb_remove_class_from_row - removes class from its row
 *
 * The class is removed from row at priorities marked in mask.
 * It does nothing if mask == 0.
 */
static inline void htb_remove_class_from_row(struct htb_sched *q,
						 struct htb_class *cl, int mask)
{
	int m = 0;
	struct htb_level *hlevel = &q->hlevel[cl->level];

	while (mask) {
		int prio = ffz(~mask);
		struct htb_prio *hprio = &hlevel->hprio[prio];

		mask &= ~(1 << prio);
		if (hprio->ptr == cl->node + prio)
			htb_next_rb_node(&hprio->ptr);

		htb_safe_rb_erase(cl->node + prio, &hprio->row);
		if (!hprio->row.rb_node)
			m |= 1 << prio;
	}
	q->row_mask[cl->level] &= ~m;
}

/**
 * htb_activate_prios - creates active classe's feed chain
 *
 * The class is connected to ancestors and/or appropriate rows
 * for priorities it is participating on. cl->cmode must be new
 * (activated) mode. It does nothing if cl->prio_activity == 0.
 */
static void htb_activate_prios(struct htb_sched *q, struct htb_class *cl)
{
	struct htb_class *p = cl->parent;
	long m, mask = cl->prio_activity;

	while (cl->cmode == HTB_MAY_BORROW && p && mask) {
		m = mask;
		while (m) {
			int prio = ffz(~m);
			m &= ~(1 << prio);

			if (p->un.inner.clprio[prio].feed.rb_node)
				/* parent already has its feed in use so that
				 * reset bit in mask as parent is already ok
				 */
				mask &= ~(1 << prio);

			htb_add_to_id_tree(&p->un.inner.clprio[prio].feed, cl, prio);
		}
		p->prio_activity |= mask;
		cl = p;
		p = cl->parent;

	}
	if (cl->cmode == HTB_CAN_SEND && mask)
		htb_add_class_to_row(q, cl, mask);
}

/**
 * htb_deactivate_prios - remove class from feed chain
 *
 * cl->cmode must represent old mode (before deactivation). It does
 * nothing if cl->prio_activity == 0. Class is removed from all feed
 * chains and rows.
 */
static void htb_deactivate_prios(struct htb_sched *q, struct htb_class *cl)
{
	struct htb_class *p = cl->parent;
	long m, mask = cl->prio_activity;

	while (cl->cmode == HTB_MAY_BORROW && p && mask) {
		m = mask;
		mask = 0;
		while (m) {
			int prio = ffz(~m);
			m &= ~(1 << prio);

			if (p->un.inner.clprio[prio].ptr == cl->node + prio) {
				/* we are removing child which is pointed to from
				 * parent feed - forget the pointer but remember
				 * classid
				 */
				p->un.inner.clprio[prio].last_ptr_id = cl->common.classid;
				p->un.inner.clprio[prio].ptr = NULL;
			}

			htb_safe_rb_erase(cl->node + prio,
					  &p->un.inner.clprio[prio].feed);

			if (!p->un.inner.clprio[prio].feed.rb_node)
				mask |= 1 << prio;
		}

		p->prio_activity &= ~mask;
		cl = p;
		p = cl->parent;

	}
	if (cl->cmode == HTB_CAN_SEND && mask)
		htb_remove_class_from_row(q, cl, mask);
}

static inline s64 htb_lowater(const struct htb_class *cl)
{
	if (htb_hysteresis)
		return cl->cmode != HTB_CANT_SEND ? -cl->cbuffer : 0;
	else
		return 0;
}
static inline s64 htb_hiwater(const struct htb_class *cl)
{
	if (htb_hysteresis)
		return cl->cmode == HTB_CAN_SEND ? -cl->buffer : 0;
	else
		return 0;
}


/**
 * htb_class_mode - computes and returns current class mode
 *
 * It computes cl's mode at time cl->t_c+diff and returns it. If mode
 * is not HTB_CAN_SEND then cl->pq_key is updated to time difference
 * from now to time when cl will change its state.
 * Also it is worth to note that class mode doesn't change simply
 * at cl->{c,}tokens == 0 but there can rather be hysteresis of
 * 0 .. -cl->{c,}buffer range. It is meant to limit number of
 * mode transitions per time unit. The speed gain is about 1/6.
 */
static inline enum htb_cmode
htb_class_mode(struct htb_class *cl, s64 *diff)
{
	s64 toks;

	if ((toks = (cl->ctokens + *diff)) < htb_lowater(cl)) {
		*diff = -toks;
		return HTB_CANT_SEND;
	}

	if ((toks = (cl->tokens + *diff)) >= htb_hiwater(cl))
		return HTB_CAN_SEND;

	*diff = -toks;
	return HTB_MAY_BORROW;
}

/**
 * htb_change_class_mode - changes classe's mode
 *
 * This should be the only way how to change classe's mode under normal
 * cirsumstances. Routine will update feed lists linkage, change mode
 * and add class to the wait event queue if appropriate. New mode should
 * be different from old one and cl->pq_key has to be valid if changing
 * to mode other than HTB_CAN_SEND (see htb_add_to_wait_tree).
 */
static void
htb_change_class_mode(struct htb_sched *q, struct htb_class *cl, s64 *diff)
{
	enum htb_cmode new_mode = htb_class_mode(cl, diff);

	if (new_mode == cl->cmode)
		return;

	if (cl->prio_activity) {	/* not necessary: speed optimization */
		if (cl->cmode != HTB_CANT_SEND)
			htb_deactivate_prios(q, cl);
		cl->cmode = new_mode;
		if (new_mode != HTB_CANT_SEND)
			htb_activate_prios(q, cl);
	} else
		cl->cmode = new_mode;
}

/**
 * htb_activate - inserts leaf cl into appropriate active feeds
 *
 * Routine learns (new) priority of leaf and activates feed chain
 * for the prio. It can be called on already active leaf safely.
 * It also adds leaf into droplist.
 */
static inline void htb_activate(struct htb_sched *q, struct htb_class *cl)
{
	WARN_ON(cl->level || !cl->un.leaf.q || !cl->un.leaf.q->q.qlen);

	if (!cl->prio_activity) {
		cl->prio_activity = 1 << cl->prio;
		htb_activate_prios(q, cl);
		list_add_tail(&cl->un.leaf.drop_list,
			      q->drops + cl->prio);
	}
}

/**
 * htb_deactivate - remove leaf cl from active feeds
 *
 * Make sure that leaf is active. In the other words it can't be called
 * with non-active leaf. It also removes class from the drop list.
 */
static inline void htb_deactivate(struct htb_sched *q, struct htb_class *cl)
{
	WARN_ON(!cl->prio_activity);

	htb_deactivate_prios(q, cl);
	cl->prio_activity = 0;
	list_del_init(&cl->un.leaf.drop_list);
}

static int htb_enqueue(struct sk_buff *skb, struct Qdisc *sch)
{
	int uninitialized_var(ret);
	struct htb_sched *q = qdisc_priv(sch);
	struct htb_class *cl = htb_classify(skb, sch, &ret);

	if (cl == HTB_DIRECT) {
		/* enqueue to helper queue */
		if (q->direct_queue.qlen < q->direct_qlen) {
			__skb_queue_tail(&q->direct_queue, skb);
			q->direct_pkts++;
		} else {
			return qdisc_drop(skb, sch);
		}
#ifdef CONFIG_NET_CLS_ACT
	} else if (!cl) {
		if (ret & __NET_XMIT_BYPASS)
			sch->qstats.drops++;
		kfree_skb(skb);
		return ret;
#endif
	} else if ((ret = qdisc_enqueue(skb, cl->un.leaf.q)) != NET_XMIT_SUCCESS) {
		if (net_xmit_drop_count(ret)) {
			sch->qstats.drops++;
			cl->qstats.drops++;
		}
		return ret;
	} else {
		htb_activate(q, cl);
	}

	sch->q.qlen++;
	return NET_XMIT_SUCCESS;
}

static inline void htb_accnt_tokens(struct htb_class *cl, int bytes, s64 diff)
{
	s64 toks = diff + cl->tokens;

	if (toks > cl->buffer)
		toks = cl->buffer;
	toks -= (s64) psched_l2t_ns(&cl->rate, bytes);
	if (toks <= -cl->mbuffer)
		toks = 1 - cl->mbuffer;

	cl->tokens = toks;
}

static inline void htb_accnt_ctokens(struct htb_class *cl, int bytes, s64 diff)
{
	s64 toks = diff + cl->ctokens;

	if (toks > cl->cbuffer)
		toks = cl->cbuffer;
	toks -= (s64) psched_l2t_ns(&cl->ceil, bytes);
	if (toks <= -cl->mbuffer)
		toks = 1 - cl->mbuffer;

	cl->ctokens = toks;
}

/**
 * htb_charge_class - charges amount "bytes" to leaf and ancestors
 *
 * Routine assumes that packet "bytes" long was dequeued from leaf cl
 * borrowing from "level". It accounts bytes to ceil leaky bucket for
 * leaf and all ancestors and to rate bucket for ancestors at levels
 * "level" and higher. It also handles possible change of mode resulting
 * from the update. Note that mode can also increase here (MAY_BORROW to
 * CAN_SEND) because we can use more precise clock that event queue here.
 * In such case we remove class from event queue first.
 */
static void htb_charge_class(struct htb_sched *q, struct htb_class *cl,
			     int level, struct sk_buff *skb)
{
	int bytes = qdisc_pkt_len(skb);
	enum htb_cmode old_mode;
	s64 diff;

	while (cl) {
		diff = min_t(s64, q->now - cl->t_c, cl->mbuffer);
		if (cl->level >= level) {
			if (cl->level == level)
				cl->xstats.lends++;
			htb_accnt_tokens(cl, bytes, diff);
		} else {
			cl->xstats.borrows++;
			cl->tokens += diff;	/* we moved t_c; update tokens */
		}
		htb_accnt_ctokens(cl, bytes, diff);
		cl->t_c = q->now;

		old_mode = cl->cmode;
		diff = 0;
		htb_change_class_mode(q, cl, &diff);
		if (old_mode != cl->cmode) {
			if (old_mode != HTB_CAN_SEND)
				htb_safe_rb_erase(&cl->pq_node, &q->hlevel[cl->level].wait_pq);
			if (cl->cmode != HTB_CAN_SEND)
				htb_add_to_wait_tree(q, cl, diff);
		}

		/* update basic stats except for leaves which are already updated */
		if (cl->level)
			bstats_update(&cl->bstats, skb);

		cl = cl->parent;
	}
}

/**
 * htb_do_events - make mode changes to classes at the level
 *
 * Scans event queue for pending events and applies them. Returns time of
 * next pending event (0 for no event in pq, q->now for too many events).
 * Note: Applied are events whose have cl->pq_key <= q->now.
 */
static s64 htb_do_events(struct htb_sched *q, const int level,
			 unsigned long start)
{
	/* don't run for longer than 2 jiffies; 2 is used instead of
	 * 1 to simplify things when jiffy is going to be incremented
	 * too soon
	 */
	unsigned long stop_at = start + 2;
	struct rb_root *wait_pq = &q->hlevel[level].wait_pq;

	while (time_before(jiffies, stop_at)) {
		struct htb_class *cl;
		s64 diff;
		struct rb_node *p = rb_first(wait_pq);

		if (!p)
			return 0;

		cl = rb_entry(p, struct htb_class, pq_node);
		if (cl->pq_key > q->now)
			return cl->pq_key;

		htb_safe_rb_erase(p, wait_pq);
		diff = min_t(s64, q->now - cl->t_c, cl->mbuffer);
		htb_change_class_mode(q, cl, &diff);
		if (cl->cmode != HTB_CAN_SEND)
			htb_add_to_wait_tree(q, cl, diff);
	}

	/* too much load - let's continue after a break for scheduling */
	if (!(q->warned & HTB_WARN_TOOMANYEVENTS)) {
		pr_warning("htb: too many events!\n");
		q->warned |= HTB_WARN_TOOMANYEVENTS;
	}

	return q->now;
}

/* Returns class->node+prio from id-tree where classe's id is >= id. NULL
 * is no such one exists.
 */
static struct rb_node *htb_id_find_next_upper(int prio, struct rb_node *n,
					      u32 id)
{
	struct rb_node *r = NULL;
	while (n) {
		struct htb_class *cl =
		    rb_entry(n, struct htb_class, node[prio]);

		if (id > cl->common.classid) {
			n = n->rb_right;
		} else if (id < cl->common.classid) {
			r = n;
			n = n->rb_left;
		} else {
			return n;
		}
	}
	return r;
}

/**
 * htb_lookup_leaf - returns next leaf class in DRR order
 *
 * Find leaf where current feed pointers points to.
 */
static struct htb_class *htb_lookup_leaf(struct htb_prio *hprio, const int prio)
{
	int i;
	struct {
		struct rb_node *root;
		struct rb_node **pptr;
		u32 *pid;
	} stk[TC_HTB_MAXDEPTH], *sp = stk;

	BUG_ON(!hprio->row.rb_node);
	sp->root = hprio->row.rb_node;
	sp->pptr = &hprio->ptr;
	sp->pid = &hprio->last_ptr_id;

	for (i = 0; i < 65535; i++) {
		if (!*sp->pptr && *sp->pid) {
			/* ptr was invalidated but id is valid - try to recover
			 * the original or next ptr
			 */
			*sp->pptr =
			    htb_id_find_next_upper(prio, sp->root, *sp->pid);
		}
		*sp->pid = 0;	/* ptr is valid now so that remove this hint as it
				 * can become out of date quickly
				 */
		if (!*sp->pptr) {	/* we are at right end; rewind & go up */
			*sp->pptr = sp->root;
			while ((*sp->pptr)->rb_left)
				*sp->pptr = (*sp->pptr)->rb_left;
			if (sp > stk) {
				sp--;
				if (!*sp->pptr) {
					WARN_ON(1);
					return NULL;
				}
				htb_next_rb_node(sp->pptr);
			}
		} else {
			struct htb_class *cl;
			struct htb_prio *clp;

			cl = rb_entry(*sp->pptr, struct htb_class, node[prio]);
			if (!cl->level)
				return cl;
			clp = &cl->un.inner.clprio[prio];
			(++sp)->root = clp->feed.rb_node;
			sp->pptr = &clp->ptr;
			sp->pid = &clp->last_ptr_id;
		}
	}
	WARN_ON(1);
	return NULL;
}

/* dequeues packet at given priority and level; call only if
 * you are sure that there is active class at prio/level
 */
static struct sk_buff *htb_dequeue_tree(struct htb_sched *q, const int prio,
					const int level)
{
	struct sk_buff *skb = NULL;
	struct htb_class *cl, *start;
	struct htb_level *hlevel = &q->hlevel[level];
	struct htb_prio *hprio = &hlevel->hprio[prio];

	/* look initial class up in the row */
	start = cl = htb_lookup_leaf(hprio, prio);

	do {
next:
		if (unlikely(!cl))
			return NULL;

		/* class can be empty - it is unlikely but can be true if leaf
		 * qdisc drops packets in enqueue routine or if someone used
		 * graft operation on the leaf since last dequeue;
		 * simply deactivate and skip such class
		 */
		if (unlikely(cl->un.leaf.q->q.qlen == 0)) {
			struct htb_class *next;
			htb_deactivate(q, cl);

			/* row/level might become empty */
			if ((q->row_mask[level] & (1 << prio)) == 0)
				return NULL;

			next = htb_lookup_leaf(hprio, prio);

			if (cl == start)	/* fix start if we just deleted it */
				start = next;
			cl = next;
			goto next;
		}

		skb = cl->un.leaf.q->dequeue(cl->un.leaf.q);
		if (likely(skb != NULL))
			break;

		qdisc_warn_nonwc("htb", cl->un.leaf.q);
		htb_next_rb_node(level ? &cl->parent->un.inner.clprio[prio].ptr:
					 &q->hlevel[0].hprio[prio].ptr);
		cl = htb_lookup_leaf(hprio, prio);

	} while (cl != start);

	if (likely(skb != NULL)) {
		bstats_update(&cl->bstats, skb);
		cl->un.leaf.deficit[level] -= qdisc_pkt_len(skb);
		if (cl->un.leaf.deficit[level] < 0) {
			cl->un.leaf.deficit[level] += cl->quantum;
			htb_next_rb_node(level ? &cl->parent->un.inner.clprio[prio].ptr :
						 &q->hlevel[0].hprio[prio].ptr);
		}
		/* this used to be after charge_class but this constelation
		 * gives us slightly better performance
		 */
		if (!cl->un.leaf.q->q.qlen)
			htb_deactivate(q, cl);
		htb_charge_class(q, cl, level, skb);
	}
	return skb;
}

static struct sk_buff *htb_dequeue(struct Qdisc *sch)
{
	struct sk_buff *skb;
	struct htb_sched *q = qdisc_priv(sch);
	int level;
	s64 next_event;
	unsigned long start_at;

	/* try to dequeue direct packets as high prio (!) to minimize cpu work */
	skb = __skb_dequeue(&q->direct_queue);
	if (skb != NULL) {
ok:
		qdisc_bstats_update(sch, skb);
		qdisc_unthrottled(sch);
		sch->q.qlen--;
		return skb;
	}

	if (!sch->q.qlen)
		goto fin;
	q->now = ktime_to_ns(ktime_get());
	start_at = jiffies;

	next_event = q->now + 5LLU * NSEC_PER_SEC;

	for (level = 0; level < TC_HTB_MAXDEPTH; level++) {
		/* common case optimization - skip event handler quickly */
		int m;
		s64 event = q->near_ev_cache[level];

		if (q->now >= event) {
			event = htb_do_events(q, level, start_at);
			if (!event)
				event = q->now + NSEC_PER_SEC;
			q->near_ev_cache[level] = event;
		}

		if (next_event > event)
			next_event = event;

		m = ~q->row_mask[level];
		while (m != (int)(-1)) {
			int prio = ffz(m);

			m |= 1 << prio;
			skb = htb_dequeue_tree(q, prio, level);
			if (likely(skb != NULL))
				goto ok;
		}
	}
	sch->qstats.overlimits++;
	if (likely(next_event > q->now)) {
		if (!test_bit(__QDISC_STATE_DEACTIVATED,
			      &qdisc_root_sleeping(q->watchdog.qdisc)->state)) {
			ktime_t time = ns_to_ktime(next_event);
			qdisc_throttled(q->watchdog.qdisc);
			hrtimer_start(&q->watchdog.timer, time,
				      HRTIMER_MODE_ABS);
		}
	} else {
		schedule_work(&q->work);
	}
fin:
	return skb;
}

/* try to drop from each class (by prio) until one succeed */
static unsigned int htb_drop(struct Qdisc *sch)
{
	struct htb_sched *q = qdisc_priv(sch);
	int prio;

	for (prio = TC_HTB_NUMPRIO - 1; prio >= 0; prio--) {
		struct list_head *p;
		list_for_each(p, q->drops + prio) {
			struct htb_class *cl = list_entry(p, struct htb_class,
							  un.leaf.drop_list);
			unsigned int len;
			if (cl->un.leaf.q->ops->drop &&
			    (len = cl->un.leaf.q->ops->drop(cl->un.leaf.q))) {
				sch->q.qlen--;
				if (!cl->un.leaf.q->q.qlen)
					htb_deactivate(q, cl);
				return len;
			}
		}
	}
	return 0;
}

/* reset all classes */
/* always caled under BH & queue lock */
static void htb_reset(struct Qdisc *sch)
{
	struct htb_sched *q = qdisc_priv(sch);
	struct htb_class *cl;
	unsigned int i;

	for (i = 0; i < q->clhash.hashsize; i++) {
		hlist_for_each_entry(cl, &q->clhash.hash[i], common.hnode) {
			if (cl->level)
				memset(&cl->un.inner, 0, sizeof(cl->un.inner));
			else {
				if (cl->un.leaf.q)
					qdisc_reset(cl->un.leaf.q);
				INIT_LIST_HEAD(&cl->un.leaf.drop_list);
			}
			cl->prio_activity = 0;
			cl->cmode = HTB_CAN_SEND;

		}
	}
	qdisc_watchdog_cancel(&q->watchdog);
	__skb_queue_purge(&q->direct_queue);
	sch->q.qlen = 0;
	memset(q->hlevel, 0, sizeof(q->hlevel));
	memset(q->row_mask, 0, sizeof(q->row_mask));
	for (i = 0; i < TC_HTB_NUMPRIO; i++)
		INIT_LIST_HEAD(q->drops + i);
}

static const struct nla_policy htb_policy[TCA_HTB_MAX + 1] = {
	[TCA_HTB_PARMS]	= { .len = sizeof(struct tc_htb_opt) },
	[TCA_HTB_INIT]	= { .len = sizeof(struct tc_htb_glob) },
	[TCA_HTB_CTAB]	= { .type = NLA_BINARY, .len = TC_RTAB_SIZE },
	[TCA_HTB_RTAB]	= { .type = NLA_BINARY, .len = TC_RTAB_SIZE },
	[TCA_HTB_DIRECT_QLEN] = { .type = NLA_U32 },
};

static void htb_work_func(struct work_struct *work)
{
	struct htb_sched *q = container_of(work, struct htb_sched, work);
	struct Qdisc *sch = q->watchdog.qdisc;

	__netif_schedule(qdisc_root(sch));
}

static int htb_init(struct Qdisc *sch, struct nlattr *opt)
{
	struct htb_sched *q = qdisc_priv(sch);
	struct nlattr *tb[TCA_HTB_MAX + 1];
	struct tc_htb_glob *gopt;
	int err;
	int i;

	if (!opt)
		return -EINVAL;

	err = nla_parse_nested(tb, TCA_HTB_MAX, opt, htb_policy);
	if (err < 0)
		return err;

	if (!tb[TCA_HTB_INIT])
		return -EINVAL;

	gopt = nla_data(tb[TCA_HTB_INIT]);
	if (gopt->version != HTB_VER >> 16)
		return -EINVAL;

	err = qdisc_class_hash_init(&q->clhash);
	if (err < 0)
		return err;
	for (i = 0; i < TC_HTB_NUMPRIO; i++)
		INIT_LIST_HEAD(q->drops + i);

	qdisc_watchdog_init(&q->watchdog, sch);
	INIT_WORK(&q->work, htb_work_func);
	skb_queue_head_init(&q->direct_queue);

	if (tb[TCA_HTB_DIRECT_QLEN])
		q->direct_qlen = nla_get_u32(tb[TCA_HTB_DIRECT_QLEN]);
	else {
		q->direct_qlen = qdisc_dev(sch)->tx_queue_len;
		if (q->direct_qlen < 2)	/* some devices have zero tx_queue_len */
			q->direct_qlen = 2;
	}