tracepoint.c 13.3 KB
Newer Older
1
/*
2
 * Copyright (C) 2008-2014 Mathieu Desnoyers
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
 *
 * 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.
 *
 * This program is distributed in the hope that 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., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 */
#include <linux/module.h>
#include <linux/mutex.h>
#include <linux/types.h>
#include <linux/jhash.h>
#include <linux/list.h>
#include <linux/rcupdate.h>
#include <linux/tracepoint.h>
#include <linux/err.h>
#include <linux/slab.h>
27
#include <linux/sched.h>
28
#include <linux/static_key.h>
29

30 31
extern struct tracepoint * const __start___tracepoints_ptrs[];
extern struct tracepoint * const __stop___tracepoints_ptrs[];
32 33 34 35

/* Set to 1 to enable tracepoint debug output */
static const int tracepoint_debug;

36
#ifdef CONFIG_MODULES
37
/*
38
 * Tracepoint module list mutex protects the local module list.
39
 */
40
static DEFINE_MUTEX(tracepoint_module_list_mutex);
41

42
/* Local list of struct tp_module */
43 44 45
static LIST_HEAD(tracepoint_module_list);
#endif /* CONFIG_MODULES */

46
/*
47 48
 * tracepoints_mutex protects the builtin and module tracepoints.
 * tracepoints_mutex nests inside tracepoint_module_list_mutex.
49
 */
50
static DEFINE_MUTEX(tracepoints_mutex);
51 52 53

/*
 * Note about RCU :
54
 * It is used to delay the free of multiple probes array until a quiescent
55 56
 * state is reached.
 */
57
struct tp_probes {
58
	struct rcu_head rcu;
59
	struct tracepoint_func probes[0];
60
};
61

62
static inline void *allocate_probes(int count)
63
{
64
	struct tp_probes *p  = kmalloc(count * sizeof(struct tracepoint_func)
65 66
			+ sizeof(struct tp_probes), GFP_KERNEL);
	return p == NULL ? NULL : p->probes;
67 68
}

69
static void rcu_free_old_probes(struct rcu_head *head)
70
{
71
	kfree(container_of(head, struct tp_probes, rcu));
72 73
}

74
static inline void release_probes(struct tracepoint_func *old)
75 76 77 78
{
	if (old) {
		struct tp_probes *tp_probes = container_of(old,
			struct tp_probes, probes[0]);
79
		call_rcu_sched(&tp_probes->rcu, rcu_free_old_probes);
80
	}
81 82
}

83
static void debug_print_probes(struct tracepoint_func *funcs)
84 85 86
{
	int i;

87
	if (!tracepoint_debug || !funcs)
88 89
		return;

90 91
	for (i = 0; funcs[i].func; i++)
		printk(KERN_DEBUG "Probe %d : %p\n", i, funcs[i].func);
92 93
}

94 95
static struct tracepoint_func *func_add(struct tracepoint_func **funcs,
		struct tracepoint_func *tp_func)
96 97
{
	int nr_probes = 0;
98
	struct tracepoint_func *old, *new;
99

100
	if (WARN_ON(!tp_func->func))
101
		return ERR_PTR(-EINVAL);
102

103 104
	debug_print_probes(*funcs);
	old = *funcs;
105 106
	if (old) {
		/* (N -> N+1), (N != 0, 1) probes */
107
		for (nr_probes = 0; old[nr_probes].func; nr_probes++)
108 109
			if (old[nr_probes].func == tp_func->func &&
			    old[nr_probes].data == tp_func->data)
110 111 112
				return ERR_PTR(-EEXIST);
	}
	/* + 2 : one for new probe, one for NULL func */
113
	new = allocate_probes(nr_probes + 2);
114 115 116
	if (new == NULL)
		return ERR_PTR(-ENOMEM);
	if (old)
117
		memcpy(new, old, nr_probes * sizeof(struct tracepoint_func));
118
	new[nr_probes] = *tp_func;
119
	new[nr_probes + 1].func = NULL;
120 121
	*funcs = new;
	debug_print_probes(*funcs);
122 123 124
	return old;
}

125 126
static void *func_remove(struct tracepoint_func **funcs,
		struct tracepoint_func *tp_func)
127 128
{
	int nr_probes = 0, nr_del = 0, i;
129
	struct tracepoint_func *old, *new;
130

131
	old = *funcs;
132

133
	if (!old)
134
		return ERR_PTR(-ENOENT);
135

136
	debug_print_probes(*funcs);
137
	/* (N -> M), (N > 1, M >= 0) probes */
138
	if (tp_func->func) {
139
		for (nr_probes = 0; old[nr_probes].func; nr_probes++) {
140 141
			if (old[nr_probes].func == tp_func->func &&
			     old[nr_probes].data == tp_func->data)
142 143
				nr_del++;
		}
144 145
	}

146 147 148 149
	/*
	 * If probe is NULL, then nr_probes = nr_del = 0, and then the
	 * entire entry will be removed.
	 */
150 151
	if (nr_probes - nr_del == 0) {
		/* N -> 0, (N > 1) */
152 153
		*funcs = NULL;
		debug_print_probes(*funcs);
154 155 156 157 158
		return old;
	} else {
		int j = 0;
		/* N -> M, (N > 1, M > 0) */
		/* + 1 for NULL */
159
		new = allocate_probes(nr_probes - nr_del + 1);
160 161
		if (new == NULL)
			return ERR_PTR(-ENOMEM);
162
		for (i = 0; old[i].func; i++)
163 164
			if (old[i].func != tp_func->func
					|| old[i].data != tp_func->data)
165
				new[j++] = old[i];
166
		new[nr_probes - nr_del].func = NULL;
167
		*funcs = new;
168
	}
169
	debug_print_probes(*funcs);
170 171 172 173
	return old;
}

/*
174
 * Add the probe function to a tracepoint.
175
 */
176 177
static int tracepoint_add_func(struct tracepoint *tp,
		struct tracepoint_func *func)
178
{
179
	struct tracepoint_func *old, *tp_funcs;
180

181 182
	if (tp->regfunc && !static_key_enabled(&tp->key))
		tp->regfunc();
183

184 185
	tp_funcs = rcu_dereference_protected(tp->funcs,
			lockdep_is_held(&tracepoints_mutex));
186 187 188 189 190
	old = func_add(&tp_funcs, func);
	if (IS_ERR(old)) {
		WARN_ON_ONCE(1);
		return PTR_ERR(old);
	}
191

192 193 194 195 196 197 198
	/*
	 * rcu_assign_pointer has a smp_wmb() which makes sure that the new
	 * probe callbacks array is consistent before setting a pointer to it.
	 * This array is referenced by __DO_TRACE from
	 * include/linux/tracepoints.h. A matching smp_read_barrier_depends()
	 * is used.
	 */
199 200 201
	rcu_assign_pointer(tp->funcs, tp_funcs);
	if (!static_key_enabled(&tp->key))
		static_key_slow_inc(&tp->key);
202
	release_probes(old);
203
	return 0;
204 205 206
}

/*
207
 * Remove a probe function from a tracepoint.
208 209 210 211
 * Note: only waiting an RCU period after setting elem->call to the empty
 * function insures that the original callback is not used anymore. This insured
 * by preempt_disable around the call site.
 */
212 213
static int tracepoint_remove_func(struct tracepoint *tp,
		struct tracepoint_func *func)
214
{
215
	struct tracepoint_func *old, *tp_funcs;
216

217 218
	tp_funcs = rcu_dereference_protected(tp->funcs,
			lockdep_is_held(&tracepoints_mutex));
219 220 221 222
	old = func_remove(&tp_funcs, func);
	if (IS_ERR(old)) {
		WARN_ON_ONCE(1);
		return PTR_ERR(old);
223
	}
224

225 226 227 228
	if (!tp_funcs) {
		/* Removed last function */
		if (tp->unregfunc && static_key_enabled(&tp->key))
			tp->unregfunc();
229

230 231
		if (static_key_enabled(&tp->key))
			static_key_slow_dec(&tp->key);
232
	}
233
	rcu_assign_pointer(tp->funcs, tp_funcs);
234
	release_probes(old);
235
	return 0;
236 237
}

238 239
/**
 * tracepoint_probe_register -  Connect a probe to a tracepoint
240
 * @tp: tracepoint
241
 * @probe: probe handler
242
 * @data: tracepoint data
243
 *
244 245 246 247 248
 * Returns 0 if ok, error value on error.
 * Note: if @tp is within a module, the caller is responsible for
 * unregistering the probe before the module is gone. This can be
 * performed either with a tracepoint module going notifier, or from
 * within module exit functions.
249
 */
250
int tracepoint_probe_register(struct tracepoint *tp, void *probe, void *data)
251
{
252 253
	struct tracepoint_func tp_func;
	int ret;
254 255

	mutex_lock(&tracepoints_mutex);
256 257 258
	tp_func.func = probe;
	tp_func.data = data;
	ret = tracepoint_add_func(tp, &tp_func);
259
	mutex_unlock(&tracepoints_mutex);
260
	return ret;
261 262 263 264 265
}
EXPORT_SYMBOL_GPL(tracepoint_probe_register);

/**
 * tracepoint_probe_unregister -  Disconnect a probe from a tracepoint
266
 * @tp: tracepoint
267
 * @probe: probe function pointer
268
 * @data: tracepoint data
269
 *
270
 * Returns 0 if ok, error value on error.
271
 */
272
int tracepoint_probe_unregister(struct tracepoint *tp, void *probe, void *data)
273
{
274 275
	struct tracepoint_func tp_func;
	int ret;
276 277

	mutex_lock(&tracepoints_mutex);
278 279 280
	tp_func.func = probe;
	tp_func.data = data;
	ret = tracepoint_remove_func(tp, &tp_func);
281
	mutex_unlock(&tracepoints_mutex);
282
	return ret;
283 284 285
}
EXPORT_SYMBOL_GPL(tracepoint_probe_unregister);

286
#ifdef CONFIG_MODULES
287 288
bool trace_module_has_bad_taint(struct module *mod)
{
289 290
	return mod->taints & ~((1 << TAINT_OOT_MODULE) | (1 << TAINT_CRAP) |
			       (1 << TAINT_UNSIGNED_MODULE));
291 292
}

293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360
static BLOCKING_NOTIFIER_HEAD(tracepoint_notify_list);

/**
 * register_tracepoint_notifier - register tracepoint coming/going notifier
 * @nb: notifier block
 *
 * Notifiers registered with this function are called on module
 * coming/going with the tracepoint_module_list_mutex held.
 * The notifier block callback should expect a "struct tp_module" data
 * pointer.
 */
int register_tracepoint_module_notifier(struct notifier_block *nb)
{
	struct tp_module *tp_mod;
	int ret;

	mutex_lock(&tracepoint_module_list_mutex);
	ret = blocking_notifier_chain_register(&tracepoint_notify_list, nb);
	if (ret)
		goto end;
	list_for_each_entry(tp_mod, &tracepoint_module_list, list)
		(void) nb->notifier_call(nb, MODULE_STATE_COMING, tp_mod);
end:
	mutex_unlock(&tracepoint_module_list_mutex);
	return ret;
}
EXPORT_SYMBOL_GPL(register_tracepoint_module_notifier);

/**
 * unregister_tracepoint_notifier - unregister tracepoint coming/going notifier
 * @nb: notifier block
 *
 * The notifier block callback should expect a "struct tp_module" data
 * pointer.
 */
int unregister_tracepoint_module_notifier(struct notifier_block *nb)
{
	struct tp_module *tp_mod;
	int ret;

	mutex_lock(&tracepoint_module_list_mutex);
	ret = blocking_notifier_chain_unregister(&tracepoint_notify_list, nb);
	if (ret)
		goto end;
	list_for_each_entry(tp_mod, &tracepoint_module_list, list)
		(void) nb->notifier_call(nb, MODULE_STATE_GOING, tp_mod);
end:
	mutex_unlock(&tracepoint_module_list_mutex);
	return ret;

}
EXPORT_SYMBOL_GPL(unregister_tracepoint_module_notifier);

/*
 * Ensure the tracer unregistered the module's probes before the module
 * teardown is performed. Prevents leaks of probe and data pointers.
 */
static void tp_module_going_check_quiescent(struct tracepoint * const *begin,
		struct tracepoint * const *end)
{
	struct tracepoint * const *iter;

	if (!begin)
		return;
	for (iter = begin; iter < end; iter++)
		WARN_ON_ONCE((*iter)->funcs);
}

361 362
static int tracepoint_module_coming(struct module *mod)
{
363
	struct tp_module *tp_mod;
364 365
	int ret = 0;

366 367 368
	if (!mod->num_tracepoints)
		return 0;

369
	/*
370 371
	 * We skip modules that taint the kernel, especially those with different
	 * module headers (for forced load), to make sure we don't cause a crash.
372
	 * Staging, out-of-tree, and unsigned GPL modules are fine.
373
	 */
374
	if (trace_module_has_bad_taint(mod))
375
		return 0;
376
	mutex_lock(&tracepoint_module_list_mutex);
377 378 379 380 381
	tp_mod = kmalloc(sizeof(struct tp_module), GFP_KERNEL);
	if (!tp_mod) {
		ret = -ENOMEM;
		goto end;
	}
382
	tp_mod->mod = mod;
383
	list_add_tail(&tp_mod->list, &tracepoint_module_list);
384 385
	blocking_notifier_call_chain(&tracepoint_notify_list,
			MODULE_STATE_COMING, tp_mod);
386
end:
387
	mutex_unlock(&tracepoint_module_list_mutex);
388 389 390
	return ret;
}

391
static void tracepoint_module_going(struct module *mod)
392
{
393
	struct tp_module *tp_mod;
394

395
	if (!mod->num_tracepoints)
396
		return;
397

398 399
	mutex_lock(&tracepoint_module_list_mutex);
	list_for_each_entry(tp_mod, &tracepoint_module_list, list) {
400
		if (tp_mod->mod == mod) {
401 402 403 404 405 406 407 408 409 410
			blocking_notifier_call_chain(&tracepoint_notify_list,
					MODULE_STATE_GOING, tp_mod);
			list_del(&tp_mod->list);
			kfree(tp_mod);
			/*
			 * Called the going notifier before checking for
			 * quiescence.
			 */
			tp_module_going_check_quiescent(mod->tracepoints_ptrs,
				mod->tracepoints_ptrs + mod->num_tracepoints);
411 412 413 414 415 416 417 418 419
			break;
		}
	}
	/*
	 * In the case of modules that were tainted at "coming", we'll simply
	 * walk through the list without finding it. We cannot use the "tainted"
	 * flag on "going", in case a module taints the kernel only after being
	 * loaded.
	 */
420
	mutex_unlock(&tracepoint_module_list_mutex);
421
}
422

423 424
static int tracepoint_module_notify(struct notifier_block *self,
		unsigned long val, void *data)
425 426
{
	struct module *mod = data;
427
	int ret = 0;
428 429 430

	switch (val) {
	case MODULE_STATE_COMING:
431 432 433 434
		ret = tracepoint_module_coming(mod);
		break;
	case MODULE_STATE_LIVE:
		break;
435
	case MODULE_STATE_GOING:
436 437 438
		tracepoint_module_going(mod);
		break;
	case MODULE_STATE_UNFORMED:
439 440
		break;
	}
441
	return ret;
442 443
}

444
static struct notifier_block tracepoint_module_nb = {
445 446 447 448
	.notifier_call = tracepoint_module_notify,
	.priority = 0,
};

449
static __init int init_tracepoints(void)
450
{
451 452 453
	int ret;

	ret = register_module_notifier(&tracepoint_module_nb);
454
	if (ret)
455
		pr_warning("Failed to register tracepoint module enter notifier\n");
456

457
	return ret;
458 459
}
__initcall(init_tracepoints);
460
#endif /* CONFIG_MODULES */
461

462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
static void for_each_tracepoint_range(struct tracepoint * const *begin,
		struct tracepoint * const *end,
		void (*fct)(struct tracepoint *tp, void *priv),
		void *priv)
{
	struct tracepoint * const *iter;

	if (!begin)
		return;
	for (iter = begin; iter < end; iter++)
		fct(*iter, priv);
}

/**
 * for_each_kernel_tracepoint - iteration on all kernel tracepoints
 * @fct: callback
 * @priv: private data
 */
void for_each_kernel_tracepoint(void (*fct)(struct tracepoint *tp, void *priv),
		void *priv)
{
	for_each_tracepoint_range(__start___tracepoints_ptrs,
		__stop___tracepoints_ptrs, fct, priv);
}
EXPORT_SYMBOL_GPL(for_each_kernel_tracepoint);

488
#ifdef CONFIG_HAVE_SYSCALL_TRACEPOINTS
489

490
/* NB: reg/unreg are called while guarded with the tracepoints_mutex */
491 492 493 494
static int sys_tracepoint_refcount;

void syscall_regfunc(void)
{
495
	struct task_struct *p, *t;
496 497

	if (!sys_tracepoint_refcount) {
498 499
		read_lock(&tasklist_lock);
		for_each_process_thread(p, t) {
500
			set_tsk_thread_flag(t, TIF_SYSCALL_TRACEPOINT);
501 502
		}
		read_unlock(&tasklist_lock);
503 504 505 506 507 508
	}
	sys_tracepoint_refcount++;
}

void syscall_unregfunc(void)
{
509
	struct task_struct *p, *t;
510 511 512

	sys_tracepoint_refcount--;
	if (!sys_tracepoint_refcount) {
513 514
		read_lock(&tasklist_lock);
		for_each_process_thread(p, t) {
515
			clear_tsk_thread_flag(t, TIF_SYSCALL_TRACEPOINT);
516 517
		}
		read_unlock(&tasklist_lock);
518 519
	}
}
520
#endif