trace_events.h 16.8 KB
Newer Older
1
/* SPDX-License-Identifier: GPL-2.0 */
2

3 4
#ifndef _LINUX_TRACE_EVENT_H
#define _LINUX_TRACE_EVENT_H
5 6

#include <linux/ring_buffer.h>
7
#include <linux/trace_seq.h>
8
#include <linux/percpu.h>
9
#include <linux/hardirq.h>
10
#include <linux/perf_event.h>
11
#include <linux/tracepoint.h>
12 13

struct trace_array;
14
struct trace_buffer;
15
struct tracer;
16
struct dentry;
17
struct bpf_prog;
18

19 20 21
const char *trace_print_flags_seq(struct trace_seq *p, const char *delim,
				  unsigned long flags,
				  const struct trace_print_flags *flag_array);
22

23 24
const char *trace_print_symbols_seq(struct trace_seq *p, unsigned long val,
				    const struct trace_print_flags *symbol_array);
25

26
#if BITS_PER_LONG == 32
R
Ross Zwisler 已提交
27 28 29 30
const char *trace_print_flags_seq_u64(struct trace_seq *p, const char *delim,
		      unsigned long long flags,
		      const struct trace_print_flags_u64 *flag_array);

31 32 33
const char *trace_print_symbols_seq_u64(struct trace_seq *p,
					unsigned long long val,
					const struct trace_print_flags_u64
34 35 36
								 *symbol_array);
#endif

37 38
const char *trace_print_bitmask_seq(struct trace_seq *p, void *bitmask_ptr,
				    unsigned int bitmask_size);
39

40
const char *trace_print_hex_seq(struct trace_seq *p,
41
				const unsigned char *buf, int len,
42
				bool concatenate);
K
Kei Tokunaga 已提交
43

44
const char *trace_print_array_seq(struct trace_seq *p,
45
				   const void *buf, int count,
D
Dave Martin 已提交
46 47
				   size_t el_size);

48 49 50
struct trace_iterator;
struct trace_event;

51 52
int trace_raw_output_prep(struct trace_iterator *iter,
			  struct trace_event *event);
53

54 55 56 57 58 59 60
/*
 * The trace entry - the most basic unit of tracing. This is what
 * is printed in the end as a single line in the trace output, such as:
 *
 *     bash-15816 [01]   235.197585: idle_cpu <- irq_enter
 */
struct trace_entry {
61
	unsigned short		type;
62 63 64 65 66
	unsigned char		flags;
	unsigned char		preempt_count;
	int			pid;
};

67
#define TRACE_EVENT_TYPE_MAX						\
68 69
	((1 << (sizeof(((struct trace_entry *)0)->type) * 8)) - 1)

70 71 72 73 74 75 76
/*
 * Trace iterator - used by printout routines who present trace
 * results to users and which routines might sleep, etc:
 */
struct trace_iterator {
	struct trace_array	*tr;
	struct tracer		*trace;
77
	struct trace_buffer	*trace_buffer;
78 79 80
	void			*private;
	int			cpu_file;
	struct mutex		mutex;
81
	struct ring_buffer_iter	**buffer_iter;
82
	unsigned long		iter_flags;
83

84 85 86
	/* trace_seq for __print_flags() and __print_symbolic() etc. */
	struct trace_seq	tmp_seq;

87 88 89 90 91
	cpumask_var_t		started;

	/* it's true when current open file is snapshot */
	bool			snapshot;

92 93 94
	/* The below is zeroed out in pipe_read */
	struct trace_seq	seq;
	struct trace_entry	*ent;
95
	unsigned long		lost_events;
96
	int			leftover;
97
	int			ent_size;
98 99 100 101 102 103
	int			cpu;
	u64			ts;

	loff_t			pos;
	long			idx;

104
	/* All new field here will be zeroed out in pipe_read */
105 106
};

107 108 109 110 111 112
enum trace_iter_flags {
	TRACE_FILE_LAT_FMT	= 1,
	TRACE_FILE_ANNOTATE	= 2,
	TRACE_FILE_TIME_IN_NS	= 4,
};

113 114

typedef enum print_line_t (*trace_print_func)(struct trace_iterator *iter,
115 116 117
				      int flags, struct trace_event *event);

struct trace_event_functions {
118 119 120 121 122 123
	trace_print_func	trace;
	trace_print_func	raw;
	trace_print_func	hex;
	trace_print_func	binary;
};

124 125 126 127 128 129 130
struct trace_event {
	struct hlist_node		node;
	struct list_head		list;
	int				type;
	struct trace_event_functions	*funcs;
};

131 132
extern int register_trace_event(struct trace_event *event);
extern int unregister_trace_event(struct trace_event *event);
133 134 135 136 137 138 139 140 141

/* Return values for print_line callback */
enum print_line_t {
	TRACE_TYPE_PARTIAL_LINE	= 0,	/* Retry after flushing the seq */
	TRACE_TYPE_HANDLED	= 1,
	TRACE_TYPE_UNHANDLED	= 2,	/* Relay to other output functions */
	TRACE_TYPE_NO_CONSUME	= 3	/* Handled but ask to not consume */
};

142
enum print_line_t trace_handle_return(struct trace_seq *s);
143

144 145 146
void tracing_generic_entry_update(struct trace_entry *entry,
				  unsigned long flags,
				  int pc);
147
struct trace_event_file;
148 149 150

struct ring_buffer_event *
trace_event_buffer_lock_reserve(struct ring_buffer **current_buffer,
151
				struct trace_event_file *trace_file,
152 153
				int type, unsigned long len,
				unsigned long flags, int pc);
154

155 156 157 158 159 160 161 162 163
#define TRACE_RECORD_CMDLINE	BIT(0)
#define TRACE_RECORD_TGID	BIT(1)

void tracing_record_taskinfo(struct task_struct *task, int flags);
void tracing_record_taskinfo_sched_switch(struct task_struct *prev,
					  struct task_struct *next, int flags);

void tracing_record_cmdline(struct task_struct *task);
void tracing_record_tgid(struct task_struct *task);
164

165
int trace_output_call(struct trace_iterator *iter, char *name, char *fmt, ...);
166

167 168
struct event_filter;

169 170 171
enum trace_reg {
	TRACE_REG_REGISTER,
	TRACE_REG_UNREGISTER,
172
#ifdef CONFIG_PERF_EVENTS
173 174
	TRACE_REG_PERF_REGISTER,
	TRACE_REG_PERF_UNREGISTER,
175 176
	TRACE_REG_PERF_OPEN,
	TRACE_REG_PERF_CLOSE,
177 178 179 180 181
	/*
	 * These (ADD/DEL) use a 'boolean' return value, where 1 (true) means a
	 * custom action was taken and the default action is not to be
	 * performed.
	 */
182 183
	TRACE_REG_PERF_ADD,
	TRACE_REG_PERF_DEL,
184
#endif
185 186
};

187
struct trace_event_call;
188

189
struct trace_event_class {
190
	const char		*system;
191 192 193 194
	void			*probe;
#ifdef CONFIG_PERF_EVENTS
	void			*perf_probe;
#endif
195
	int			(*reg)(struct trace_event_call *event,
196
				       enum trace_reg type, void *data);
197 198
	int			(*define_fields)(struct trace_event_call *);
	struct list_head	*(*get_fields)(struct trace_event_call *);
199
	struct list_head	fields;
200
	int			(*raw_init)(struct trace_event_call *);
201 202
};

203
extern int trace_event_reg(struct trace_event_call *event,
204
			    enum trace_reg type, void *data);
205

206
struct trace_event_buffer {
207 208
	struct ring_buffer		*buffer;
	struct ring_buffer_event	*event;
209
	struct trace_event_file		*trace_file;
210 211 212 213 214
	void				*entry;
	unsigned long			flags;
	int				pc;
};

215
void *trace_event_buffer_reserve(struct trace_event_buffer *fbuffer,
216
				  struct trace_event_file *trace_file,
217 218
				  unsigned long len);

219
void trace_event_buffer_commit(struct trace_event_buffer *fbuffer);
220

221 222
enum {
	TRACE_EVENT_FL_FILTERED_BIT,
223
	TRACE_EVENT_FL_CAP_ANY_BIT,
L
Li Zefan 已提交
224
	TRACE_EVENT_FL_NO_SET_FILTER_BIT,
225
	TRACE_EVENT_FL_IGNORE_ENABLE_BIT,
226
	TRACE_EVENT_FL_TRACEPOINT_BIT,
A
Alexei Starovoitov 已提交
227
	TRACE_EVENT_FL_KPROBE_BIT,
228
	TRACE_EVENT_FL_UPROBE_BIT,
229 230
};

231 232 233 234 235
/*
 * Event flags:
 *  FILTERED	  - The event has a filter attached
 *  CAP_ANY	  - Any user can enable for perf
 *  NO_SET_FILTER - Set when filter has error and is to be ignored
236
 *  IGNORE_ENABLE - For trace internal events, do not enable with debugfs file
237
 *  TRACEPOINT    - Event is a tracepoint
A
Alexei Starovoitov 已提交
238
 *  KPROBE        - Event is a kprobe
239
 *  UPROBE        - Event is a uprobe
240
 */
241
enum {
242
	TRACE_EVENT_FL_FILTERED		= (1 << TRACE_EVENT_FL_FILTERED_BIT),
243
	TRACE_EVENT_FL_CAP_ANY		= (1 << TRACE_EVENT_FL_CAP_ANY_BIT),
L
Li Zefan 已提交
244
	TRACE_EVENT_FL_NO_SET_FILTER	= (1 << TRACE_EVENT_FL_NO_SET_FILTER_BIT),
245
	TRACE_EVENT_FL_IGNORE_ENABLE	= (1 << TRACE_EVENT_FL_IGNORE_ENABLE_BIT),
246
	TRACE_EVENT_FL_TRACEPOINT	= (1 << TRACE_EVENT_FL_TRACEPOINT_BIT),
A
Alexei Starovoitov 已提交
247
	TRACE_EVENT_FL_KPROBE		= (1 << TRACE_EVENT_FL_KPROBE_BIT),
248
	TRACE_EVENT_FL_UPROBE		= (1 << TRACE_EVENT_FL_UPROBE_BIT),
249 250
};

251 252
#define TRACE_EVENT_FL_UKPROBE (TRACE_EVENT_FL_KPROBE | TRACE_EVENT_FL_UPROBE)

253
struct trace_event_call {
254
	struct list_head	list;
255
	struct trace_event_class *class;
256 257 258 259 260
	union {
		char			*name;
		/* Set TRACE_EVENT_FL_TRACEPOINT flag when using "tp" */
		struct tracepoint	*tp;
	};
261
	struct trace_event	event;
262
	char			*print_fmt;
263
	struct event_filter	*filter;
264
	void			*mod;
265
	void			*data;
266 267 268 269
	/*
	 *   bit 0:		filter_active
	 *   bit 1:		allow trace by non root (cap any)
	 *   bit 2:		failed to apply filter
270
	 *   bit 3:		trace internal event (do not enable)
271
	 *   bit 4:		Event was enabled by module
272
	 *   bit 5:		use call filter rather than file filter
273
	 *   bit 6:		Event is a tracepoint
274
	 */
275 276 277 278 279
	int			flags; /* static flags of different events */

#ifdef CONFIG_PERF_EVENTS
	int				perf_refcount;
	struct hlist_head __percpu	*perf_events;
280
	struct bpf_prog_array __rcu	*prog_array;
281

282
	int	(*perf_perm)(struct trace_event_call *,
283
			     struct perf_event *);
284 285 286
#endif
};

287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
#ifdef CONFIG_PERF_EVENTS
static inline bool bpf_prog_array_valid(struct trace_event_call *call)
{
	/*
	 * This inline function checks whether call->prog_array
	 * is valid or not. The function is called in various places,
	 * outside rcu_read_lock/unlock, as a heuristic to speed up execution.
	 *
	 * If this function returns true, and later call->prog_array
	 * becomes false inside rcu_read_lock/unlock region,
	 * we bail out then. If this function return false,
	 * there is a risk that we might miss a few events if the checking
	 * were delayed until inside rcu_read_lock/unlock region and
	 * call->prog_array happened to become non-NULL then.
	 *
	 * Here, READ_ONCE() is used instead of rcu_access_pointer().
	 * rcu_access_pointer() requires the actual definition of
	 * "struct bpf_prog_array" while READ_ONCE() only needs
	 * a declaration of the same type.
	 */
	return !!READ_ONCE(call->prog_array);
}
#endif

311
static inline const char *
312
trace_event_name(struct trace_event_call *call)
313 314 315 316 317 318 319
{
	if (call->flags & TRACE_EVENT_FL_TRACEPOINT)
		return call->tp ? call->tp->name : NULL;
	else
		return call->name;
}

320
struct trace_array;
321
struct trace_subsystem_dir;
322 323

enum {
324 325
	EVENT_FILE_FL_ENABLED_BIT,
	EVENT_FILE_FL_RECORDED_CMD_BIT,
326
	EVENT_FILE_FL_RECORDED_TGID_BIT,
327 328 329 330 331 332
	EVENT_FILE_FL_FILTERED_BIT,
	EVENT_FILE_FL_NO_SET_FILTER_BIT,
	EVENT_FILE_FL_SOFT_MODE_BIT,
	EVENT_FILE_FL_SOFT_DISABLED_BIT,
	EVENT_FILE_FL_TRIGGER_MODE_BIT,
	EVENT_FILE_FL_TRIGGER_COND_BIT,
333
	EVENT_FILE_FL_PID_FILTER_BIT,
334
	EVENT_FILE_FL_WAS_ENABLED_BIT,
335 336 337
};

/*
338
 * Event file flags:
339
 *  ENABLED	  - The event is enabled
340
 *  RECORDED_CMD  - The comms should be recorded at sched_switch
341
 *  RECORDED_TGID - The tgids should be recorded at sched_switch
342 343
 *  FILTERED	  - The event has a filter attached
 *  NO_SET_FILTER - Set when filter has error and is to be ignored
344 345 346
 *  SOFT_MODE     - The event is enabled/disabled by SOFT_DISABLED
 *  SOFT_DISABLED - When set, do not trace the event (even though its
 *                   tracepoint may be enabled)
347
 *  TRIGGER_MODE  - When set, invoke the triggers associated with the event
348
 *  TRIGGER_COND  - When set, one or more triggers has an associated filter
349
 *  PID_FILTER    - When set, the event is filtered based on pid
350
 *  WAS_ENABLED   - Set when enabled to know to clear trace on module removal
351 352
 */
enum {
353 354
	EVENT_FILE_FL_ENABLED		= (1 << EVENT_FILE_FL_ENABLED_BIT),
	EVENT_FILE_FL_RECORDED_CMD	= (1 << EVENT_FILE_FL_RECORDED_CMD_BIT),
355
	EVENT_FILE_FL_RECORDED_TGID	= (1 << EVENT_FILE_FL_RECORDED_TGID_BIT),
356 357 358 359 360 361
	EVENT_FILE_FL_FILTERED		= (1 << EVENT_FILE_FL_FILTERED_BIT),
	EVENT_FILE_FL_NO_SET_FILTER	= (1 << EVENT_FILE_FL_NO_SET_FILTER_BIT),
	EVENT_FILE_FL_SOFT_MODE		= (1 << EVENT_FILE_FL_SOFT_MODE_BIT),
	EVENT_FILE_FL_SOFT_DISABLED	= (1 << EVENT_FILE_FL_SOFT_DISABLED_BIT),
	EVENT_FILE_FL_TRIGGER_MODE	= (1 << EVENT_FILE_FL_TRIGGER_MODE_BIT),
	EVENT_FILE_FL_TRIGGER_COND	= (1 << EVENT_FILE_FL_TRIGGER_COND_BIT),
362
	EVENT_FILE_FL_PID_FILTER	= (1 << EVENT_FILE_FL_PID_FILTER_BIT),
363
	EVENT_FILE_FL_WAS_ENABLED	= (1 << EVENT_FILE_FL_WAS_ENABLED_BIT),
364 365
};

366
struct trace_event_file {
367
	struct list_head		list;
368
	struct trace_event_call		*event_call;
369
	struct event_filter __rcu	*filter;
370 371
	struct dentry			*dir;
	struct trace_array		*tr;
372
	struct trace_subsystem_dir	*system;
373
	struct list_head		triggers;
374

375 376
	/*
	 * 32 bit flags:
377 378
	 *   bit 0:		enabled
	 *   bit 1:		enabled cmd record
379 380
	 *   bit 2:		enable/disable with the soft disable bit
	 *   bit 3:		soft disabled
381
	 *   bit 4:		trigger enabled
382
	 *
383 384 385
	 * Note: The bits must be set atomically to prevent races
	 * from other writers. Reads of flags do not need to be in
	 * sync as they occur in critical sections. But the way flags
386
	 * is currently used, these changes do not affect the code
387 388
	 * except that when a change is made, it may have a slight
	 * delay in propagating the changes to other CPUs due to
389
	 * caching and such. Which is mostly OK ;-)
390
	 */
391
	unsigned long		flags;
392
	atomic_t		sm_ref;	/* soft-mode reference counter */
393
	atomic_t		tm_ref;	/* trigger-mode reference counter */
394 395
};

396 397 398
#define __TRACE_EVENT_FLAGS(name, value)				\
	static int __init trace_init_flags_##name(void)			\
	{								\
399
		event_##name.flags |= value;				\
400 401 402 403
		return 0;						\
	}								\
	early_initcall(trace_init_flags_##name);

404
#define __TRACE_EVENT_PERF_PERM(name, expr...)				\
405
	static int perf_perm_##name(struct trace_event_call *tp_event, \
406 407 408 409 410 411 412 413 414 415 416
				    struct perf_event *p_event)		\
	{								\
		return ({ expr; });					\
	}								\
	static int __init trace_init_perf_perm_##name(void)		\
	{								\
		event_##name.perf_perm = &perf_perm_##name;		\
		return 0;						\
	}								\
	early_initcall(trace_init_perf_perm_##name);

417
#define PERF_MAX_TRACE_SIZE	2048
418

419
#define MAX_FILTER_STR_VAL	256	/* Should handle KSYM_SYMBOL_LEN */
420

421 422
enum event_trigger_type {
	ETT_NONE		= (0),
423
	ETT_TRACE_ONOFF		= (1 << 0),
424
	ETT_SNAPSHOT		= (1 << 1),
425
	ETT_STACKTRACE		= (1 << 2),
426
	ETT_EVENT_ENABLE	= (1 << 3),
427
	ETT_EVENT_HIST		= (1 << 4),
428
	ETT_HIST_ENABLE		= (1 << 5),
429 430
};

L
Li Zefan 已提交
431
extern int filter_match_preds(struct event_filter *filter, void *rec);
432

433
extern enum event_trigger_type event_triggers_call(struct trace_event_file *file,
434
						   void *rec);
435
extern void event_triggers_post_call(struct trace_event_file *file,
436 437
				     enum event_trigger_type tt,
				     void *rec);
438

439 440
bool trace_event_ignore_this_pid(struct trace_event_file *trace_file);

441
/**
442
 * trace_trigger_soft_disabled - do triggers and test if soft disabled
443 444 445 446 447 448 449 450
 * @file: The file pointer of the event to test
 *
 * If any triggers without filters are attached to this event, they
 * will be called here. If the event is soft disabled and has no
 * triggers that require testing the fields, it will return true,
 * otherwise false.
 */
static inline bool
451
trace_trigger_soft_disabled(struct trace_event_file *file)
452 453 454
{
	unsigned long eflags = file->flags;

455 456
	if (!(eflags & EVENT_FILE_FL_TRIGGER_COND)) {
		if (eflags & EVENT_FILE_FL_TRIGGER_MODE)
457
			event_triggers_call(file, NULL);
458
		if (eflags & EVENT_FILE_FL_SOFT_DISABLED)
459
			return true;
460 461
		if (eflags & EVENT_FILE_FL_PID_FILTER)
			return trace_event_ignore_this_pid(file);
462 463 464 465
	}
	return false;
}

466
#ifdef CONFIG_BPF_EVENTS
467 468 469
unsigned int trace_call_bpf(struct trace_event_call *call, void *ctx);
int perf_event_attach_bpf_prog(struct perf_event *event, struct bpf_prog *prog);
void perf_event_detach_bpf_prog(struct perf_event *event);
470
#else
471
static inline unsigned int trace_call_bpf(struct trace_event_call *call, void *ctx)
472 473 474
{
	return 1;
}
475 476 477 478 479 480 481 482 483

static inline int
perf_event_attach_bpf_prog(struct perf_event *event, struct bpf_prog *prog)
{
	return -EOPNOTSUPP;
}

static inline void perf_event_detach_bpf_prog(struct perf_event *event) { }

484 485
#endif

486 487 488 489
enum {
	FILTER_OTHER = 0,
	FILTER_STATIC_STRING,
	FILTER_DYN_STRING,
490
	FILTER_PTR_STRING,
491
	FILTER_TRACE_FN,
492 493
	FILTER_COMM,
	FILTER_CPU,
494 495
};

496 497
extern int trace_event_raw_init(struct trace_event_call *call);
extern int trace_define_field(struct trace_event_call *call, const char *type,
498 499
			      const char *name, int offset, int size,
			      int is_signed, int filter_type);
500 501
extern int trace_add_event_call(struct trace_event_call *call);
extern int trace_remove_event_call(struct trace_event_call *call);
502
extern int trace_event_get_offsets(struct trace_event_call *call);
503

504
#define is_signed_type(type)	(((type)(-1)) < (type)1)
505

506 507
int trace_set_clr_event(const char *system, const char *event, int set);

508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526
/*
 * The double __builtin_constant_p is because gcc will give us an error
 * if we try to allocate the static variable to fmt if it is not a
 * constant. Even with the outer if statement optimizing out.
 */
#define event_trace_printk(ip, fmt, args...)				\
do {									\
	__trace_printk_check_format(fmt, ##args);			\
	tracing_record_cmdline(current);				\
	if (__builtin_constant_p(fmt)) {				\
		static const char *trace_printk_fmt			\
		  __attribute__((section("__trace_printk_fmt"))) =	\
			__builtin_constant_p(fmt) ? fmt : NULL;		\
									\
		__trace_bprintk(ip, trace_printk_fmt, ##args);		\
	} else								\
		__trace_printk(ip, fmt, ##args);			\
} while (0)

527
#ifdef CONFIG_PERF_EVENTS
L
Li Zefan 已提交
528
struct perf_event;
529 530 531

DECLARE_PER_CPU(struct pt_regs, perf_trace_regs);

532 533
extern int  perf_trace_init(struct perf_event *event);
extern void perf_trace_destroy(struct perf_event *event);
P
Peter Zijlstra 已提交
534 535
extern int  perf_trace_add(struct perf_event *event, int flags);
extern void perf_trace_del(struct perf_event *event, int flags);
536 537 538 539
#ifdef CONFIG_KPROBE_EVENTS
extern int  perf_kprobe_init(struct perf_event *event, bool is_retprobe);
extern void perf_kprobe_destroy(struct perf_event *event);
#endif
540
extern int  ftrace_profile_set_filter(struct perf_event *event, int event_id,
L
Li Zefan 已提交
541 542
				     char *filter_str);
extern void ftrace_profile_free_filter(struct perf_event *event);
543 544
void perf_trace_buf_update(void *record, u16 type);
void *perf_trace_buf_alloc(int size, struct pt_regs **regs, int *rctxp);
545

546 547 548 549 550
void perf_trace_run_bpf_submit(void *raw_data, int size, int rctx,
			       struct trace_event_call *call, u64 count,
			       struct pt_regs *regs, struct hlist_head *head,
			       struct task_struct *task);

551
static inline void
552
perf_trace_buf_submit(void *raw_data, int size, int rctx, u16 type,
553
		       u64 count, struct pt_regs *regs, void *head,
554
		       struct task_struct *task)
555
{
556
	perf_tp_event(type, count, raw_data, size, regs, head, rctx, task);
557
}
558

L
Li Zefan 已提交
559 560
#endif

561
#endif /* _LINUX_TRACE_EVENT_H */