kprobes.c 55.9 KB
Newer Older
L
Linus Torvalds 已提交
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
/*
 *  Kernel Probes (KProbes)
 *  kernel/kprobes.c
 *
 * 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.
 *
 * Copyright (C) IBM Corporation, 2002, 2004
 *
 * 2002-Oct	Created by Vamsi Krishna S <vamsi_krishna@in.ibm.com> Kernel
 *		Probes initial implementation (includes suggestions from
 *		Rusty Russell).
 * 2004-Aug	Updated by Prasanna S Panchamukhi <prasanna@in.ibm.com> with
 *		hlists and exceptions notifier as suggested by Andi Kleen.
 * 2004-July	Suparna Bhattacharya <suparna@in.ibm.com> added jumper probes
 *		interface to access function arguments.
 * 2004-Sep	Prasanna S Panchamukhi <prasanna@in.ibm.com> Changed Kprobes
 *		exceptions notifier to be first on the priority list.
30 31 32
 * 2005-May	Hien Nguyen <hien@us.ibm.com>, Jim Keniston
 *		<jkenisto@us.ibm.com> and Prasanna S Panchamukhi
 *		<prasanna@in.ibm.com> added function-return probes.
L
Linus Torvalds 已提交
33 34 35 36
 */
#include <linux/kprobes.h>
#include <linux/hash.h>
#include <linux/init.h>
T
Tim Schmielau 已提交
37
#include <linux/slab.h>
R
Randy Dunlap 已提交
38
#include <linux/stddef.h>
L
Linus Torvalds 已提交
39
#include <linux/module.h>
40
#include <linux/moduleloader.h>
41
#include <linux/kallsyms.h>
42
#include <linux/freezer.h>
43 44
#include <linux/seq_file.h>
#include <linux/debugfs.h>
45
#include <linux/sysctl.h>
46
#include <linux/kdebug.h>
47
#include <linux/memory.h>
48
#include <linux/ftrace.h>
49
#include <linux/cpu.h>
50
#include <linux/jump_label.h>
51

52
#include <asm-generic/sections.h>
L
Linus Torvalds 已提交
53 54
#include <asm/cacheflush.h>
#include <asm/errno.h>
55
#include <asm/uaccess.h>
L
Linus Torvalds 已提交
56 57 58 59

#define KPROBE_HASH_BITS 6
#define KPROBE_TABLE_SIZE (1 << KPROBE_HASH_BITS)

60 61 62 63 64 65 66 67 68 69

/*
 * Some oddball architectures like 64bit powerpc have function descriptors
 * so this must be overridable.
 */
#ifndef kprobe_lookup_name
#define kprobe_lookup_name(name, addr) \
	addr = ((kprobe_opcode_t *)(kallsyms_lookup_name(name)))
#endif

70
static int kprobes_initialized;
L
Linus Torvalds 已提交
71
static struct hlist_head kprobe_table[KPROBE_TABLE_SIZE];
72
static struct hlist_head kretprobe_inst_table[KPROBE_TABLE_SIZE];
L
Linus Torvalds 已提交
73

74
/* NOTE: change this value only with kprobe_mutex held */
75
static bool kprobes_all_disarmed;
76

77 78
/* This protects kprobe_table and optimizing_list */
static DEFINE_MUTEX(kprobe_mutex);
79
static DEFINE_PER_CPU(struct kprobe *, kprobe_instance) = NULL;
80
static struct {
81
	spinlock_t lock ____cacheline_aligned_in_smp;
82 83 84 85 86 87
} kretprobe_table_locks[KPROBE_TABLE_SIZE];

static spinlock_t *kretprobe_table_lock_ptr(unsigned long hash)
{
	return &(kretprobe_table_locks[hash].lock);
}
L
Linus Torvalds 已提交
88

89 90 91 92 93 94 95
/*
 * Normally, functions that we'd want to prohibit kprobes in, are marked
 * __kprobes. But, there are cases where such functions already belong to
 * a different section (__sched for preempt_schedule)
 *
 * For such cases, we now have a blacklist
 */
96
static struct kprobe_blackpoint kprobe_blacklist[] = {
97
	{"preempt_schedule",},
98
	{"native_get_debugreg",},
99 100
	{"irq_entries_start",},
	{"common_interrupt",},
101
	{"mcount",},	/* mcount can be called from everywhere */
102 103 104
	{NULL}    /* Terminator */
};

105
#ifdef __ARCH_WANT_KPROBES_INSN_SLOT
106 107 108 109 110 111 112
/*
 * kprobe->ainsn.insn points to the copy of the instruction to be
 * single-stepped. x86_64, POWER4 and above have no-exec support and
 * stepping on the instruction on a vmalloced/kmalloced/data page
 * is a recipe for disaster
 */
struct kprobe_insn_page {
113
	struct list_head list;
114 115
	kprobe_opcode_t *insns;		/* Page of instruction slots */
	int nused;
116
	int ngarbage;
117
	char slot_used[];
118 119
};

120 121 122 123 124 125 126 127 128 129 130 131 132 133 134
#define KPROBE_INSN_PAGE_SIZE(slots)			\
	(offsetof(struct kprobe_insn_page, slot_used) +	\
	 (sizeof(char) * (slots)))

struct kprobe_insn_cache {
	struct list_head pages;	/* list of kprobe_insn_page */
	size_t insn_size;	/* size of instruction slot */
	int nr_garbage;
};

static int slots_per_page(struct kprobe_insn_cache *c)
{
	return PAGE_SIZE/(c->insn_size * sizeof(kprobe_opcode_t));
}

135 136 137 138 139 140
enum kprobe_slot_state {
	SLOT_CLEAN = 0,
	SLOT_DIRTY = 1,
	SLOT_USED = 2,
};

141 142 143 144 145 146 147
static DEFINE_MUTEX(kprobe_insn_mutex);	/* Protects kprobe_insn_slots */
static struct kprobe_insn_cache kprobe_insn_slots = {
	.pages = LIST_HEAD_INIT(kprobe_insn_slots.pages),
	.insn_size = MAX_INSN_SIZE,
	.nr_garbage = 0,
};
static int __kprobes collect_garbage_slots(struct kprobe_insn_cache *c);
148

149
/**
150
 * __get_insn_slot() - Find a slot on an executable page for an instruction.
151 152
 * We allocate an executable page if there's no room on existing ones.
 */
153
static kprobe_opcode_t __kprobes *__get_insn_slot(struct kprobe_insn_cache *c)
154 155 156
{
	struct kprobe_insn_page *kip;

157
 retry:
158 159
	list_for_each_entry(kip, &c->pages, list) {
		if (kip->nused < slots_per_page(c)) {
160
			int i;
161
			for (i = 0; i < slots_per_page(c); i++) {
162 163
				if (kip->slot_used[i] == SLOT_CLEAN) {
					kip->slot_used[i] = SLOT_USED;
164
					kip->nused++;
165
					return kip->insns + (i * c->insn_size);
166 167
				}
			}
168 169 170
			/* kip->nused is broken. Fix it. */
			kip->nused = slots_per_page(c);
			WARN_ON(1);
171 172 173
		}
	}

174
	/* If there are any garbage slots, collect it and try again. */
175
	if (c->nr_garbage && collect_garbage_slots(c) == 0)
176
		goto retry;
177 178 179

	/* All out of space.  Need to allocate a new page. */
	kip = kmalloc(KPROBE_INSN_PAGE_SIZE(slots_per_page(c)), GFP_KERNEL);
180
	if (!kip)
181 182 183 184 185 186 187 188 189 190 191 192
		return NULL;

	/*
	 * Use module_alloc so this page is within +/- 2GB of where the
	 * kernel image and loaded module images reside. This is required
	 * so x86_64 can correctly handle the %rip-relative fixups.
	 */
	kip->insns = module_alloc(PAGE_SIZE);
	if (!kip->insns) {
		kfree(kip);
		return NULL;
	}
193
	INIT_LIST_HEAD(&kip->list);
194
	memset(kip->slot_used, SLOT_CLEAN, slots_per_page(c));
195
	kip->slot_used[0] = SLOT_USED;
196
	kip->nused = 1;
197
	kip->ngarbage = 0;
198
	list_add(&kip->list, &c->pages);
199 200 201
	return kip->insns;
}

202

203 204
kprobe_opcode_t __kprobes *get_insn_slot(void)
{
205 206
	kprobe_opcode_t *ret = NULL;

207
	mutex_lock(&kprobe_insn_mutex);
208
	ret = __get_insn_slot(&kprobe_insn_slots);
209
	mutex_unlock(&kprobe_insn_mutex);
210

211 212 213
	return ret;
}

214 215 216
/* Return 1 if all garbages are collected, otherwise 0. */
static int __kprobes collect_one_slot(struct kprobe_insn_page *kip, int idx)
{
217
	kip->slot_used[idx] = SLOT_CLEAN;
218 219 220 221 222 223 224 225
	kip->nused--;
	if (kip->nused == 0) {
		/*
		 * Page is no longer in use.  Free it unless
		 * it's the last one.  We keep the last one
		 * so as not to have to set it up again the
		 * next time somebody inserts a probe.
		 */
226
		if (!list_is_singular(&kip->list)) {
227
			list_del(&kip->list);
228 229 230 231 232 233 234 235
			module_free(NULL, kip->insns);
			kfree(kip);
		}
		return 1;
	}
	return 0;
}

236
static int __kprobes collect_garbage_slots(struct kprobe_insn_cache *c)
237
{
238
	struct kprobe_insn_page *kip, *next;
239

240 241
	/* Ensure no-one is interrupted on the garbages */
	synchronize_sched();
242

243
	list_for_each_entry_safe(kip, next, &c->pages, list) {
244 245 246 247
		int i;
		if (kip->ngarbage == 0)
			continue;
		kip->ngarbage = 0;	/* we will collect all garbages */
248
		for (i = 0; i < slots_per_page(c); i++) {
249
			if (kip->slot_used[i] == SLOT_DIRTY &&
250 251 252 253
			    collect_one_slot(kip, i))
				break;
		}
	}
254
	c->nr_garbage = 0;
255 256 257
	return 0;
}

258 259
static void __kprobes __free_insn_slot(struct kprobe_insn_cache *c,
				       kprobe_opcode_t *slot, int dirty)
260 261 262
{
	struct kprobe_insn_page *kip;

263
	list_for_each_entry(kip, &c->pages, list) {
264 265
		long idx = ((long)slot - (long)kip->insns) /
				(c->insn_size * sizeof(kprobe_opcode_t));
266 267
		if (idx >= 0 && idx < slots_per_page(c)) {
			WARN_ON(kip->slot_used[idx] != SLOT_USED);
268
			if (dirty) {
269
				kip->slot_used[idx] = SLOT_DIRTY;
270
				kip->ngarbage++;
271 272
				if (++c->nr_garbage > slots_per_page(c))
					collect_garbage_slots(c);
273
			} else
274 275
				collect_one_slot(kip, idx);
			return;
276 277
		}
	}
278 279 280
	/* Could not free this slot. */
	WARN_ON(1);
}
281

282 283 284 285
void __kprobes free_insn_slot(kprobe_opcode_t * slot, int dirty)
{
	mutex_lock(&kprobe_insn_mutex);
	__free_insn_slot(&kprobe_insn_slots, slot, dirty);
286
	mutex_unlock(&kprobe_insn_mutex);
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
#ifdef CONFIG_OPTPROBES
/* For optimized_kprobe buffer */
static DEFINE_MUTEX(kprobe_optinsn_mutex); /* Protects kprobe_optinsn_slots */
static struct kprobe_insn_cache kprobe_optinsn_slots = {
	.pages = LIST_HEAD_INIT(kprobe_optinsn_slots.pages),
	/* .insn_size is initialized later */
	.nr_garbage = 0,
};
/* Get a slot for optimized_kprobe buffer */
kprobe_opcode_t __kprobes *get_optinsn_slot(void)
{
	kprobe_opcode_t *ret = NULL;

	mutex_lock(&kprobe_optinsn_mutex);
	ret = __get_insn_slot(&kprobe_optinsn_slots);
	mutex_unlock(&kprobe_optinsn_mutex);

	return ret;
}

void __kprobes free_optinsn_slot(kprobe_opcode_t * slot, int dirty)
{
	mutex_lock(&kprobe_optinsn_mutex);
	__free_insn_slot(&kprobe_optinsn_slots, slot, dirty);
	mutex_unlock(&kprobe_optinsn_mutex);
}
#endif
315
#endif
316

317 318 319
/* We have preemption disabled.. so it is safe to use __ versions */
static inline void set_kprobe_instance(struct kprobe *kp)
{
C
Christoph Lameter 已提交
320
	__this_cpu_write(kprobe_instance, kp);
321 322 323 324
}

static inline void reset_kprobe_instance(void)
{
C
Christoph Lameter 已提交
325
	__this_cpu_write(kprobe_instance, NULL);
326 327
}

328 329
/*
 * This routine is called either:
330
 * 	- under the kprobe_mutex - during kprobe_[un]register()
331
 * 				OR
332
 * 	- with preemption disabled - from arch/xxx/kernel/kprobes.c
333
 */
334
struct kprobe __kprobes *get_kprobe(void *addr)
L
Linus Torvalds 已提交
335 336 337
{
	struct hlist_head *head;
	struct hlist_node *node;
338
	struct kprobe *p;
L
Linus Torvalds 已提交
339 340

	head = &kprobe_table[hash_ptr(addr, KPROBE_HASH_BITS)];
341
	hlist_for_each_entry_rcu(p, node, head, hlist) {
L
Linus Torvalds 已提交
342 343 344
		if (p->addr == addr)
			return p;
	}
345

L
Linus Torvalds 已提交
346 347 348
	return NULL;
}

349 350 351 352 353 354 355 356
static int __kprobes aggr_pre_handler(struct kprobe *p, struct pt_regs *regs);

/* Return true if the kprobe is an aggregator */
static inline int kprobe_aggrprobe(struct kprobe *p)
{
	return p->pre_handler == aggr_pre_handler;
}

357 358 359 360 361 362 363
/* Return true(!0) if the kprobe is unused */
static inline int kprobe_unused(struct kprobe *p)
{
	return kprobe_aggrprobe(p) && kprobe_disabled(p) &&
	       list_empty(&p->list);
}

364 365 366
/*
 * Keep all fields in the kprobe consistent
 */
367
static inline void copy_kprobe(struct kprobe *ap, struct kprobe *p)
368
{
369 370
	memcpy(&p->opcode, &ap->opcode, sizeof(kprobe_opcode_t));
	memcpy(&p->ainsn, &ap->ainsn, sizeof(struct arch_specific_insn));
371 372 373
}

#ifdef CONFIG_OPTPROBES
374 375 376
/* NOTE: change this value only with kprobe_mutex held */
static bool kprobes_allow_optimization;

377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
/*
 * Call all pre_handler on the list, but ignores its return value.
 * This must be called from arch-dep optimized caller.
 */
void __kprobes opt_pre_handler(struct kprobe *p, struct pt_regs *regs)
{
	struct kprobe *kp;

	list_for_each_entry_rcu(kp, &p->list, list) {
		if (kp->pre_handler && likely(!kprobe_disabled(kp))) {
			set_kprobe_instance(kp);
			kp->pre_handler(kp, regs);
		}
		reset_kprobe_instance();
	}
}

394 395 396 397 398 399 400 401 402 403 404
/* Free optimized instructions and optimized_kprobe */
static __kprobes void free_aggr_kprobe(struct kprobe *p)
{
	struct optimized_kprobe *op;

	op = container_of(p, struct optimized_kprobe, kp);
	arch_remove_optimized_kprobe(op);
	arch_remove_kprobe(p);
	kfree(op);
}

405 406 407 408 409 410 411 412 413 414 415 416 417
/* Return true(!0) if the kprobe is ready for optimization. */
static inline int kprobe_optready(struct kprobe *p)
{
	struct optimized_kprobe *op;

	if (kprobe_aggrprobe(p)) {
		op = container_of(p, struct optimized_kprobe, kp);
		return arch_prepared_optinsn(&op->optinsn);
	}

	return 0;
}

418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
/* Return true(!0) if the kprobe is disarmed. Note: p must be on hash list */
static inline int kprobe_disarmed(struct kprobe *p)
{
	struct optimized_kprobe *op;

	/* If kprobe is not aggr/opt probe, just return kprobe is disabled */
	if (!kprobe_aggrprobe(p))
		return kprobe_disabled(p);

	op = container_of(p, struct optimized_kprobe, kp);

	return kprobe_disabled(p) && list_empty(&op->list);
}

/* Return true(!0) if the probe is queued on (un)optimizing lists */
static int __kprobes kprobe_queued(struct kprobe *p)
{
	struct optimized_kprobe *op;

	if (kprobe_aggrprobe(p)) {
		op = container_of(p, struct optimized_kprobe, kp);
		if (!list_empty(&op->list))
			return 1;
	}
	return 0;
}

445 446 447 448
/*
 * Return an optimized kprobe whose optimizing code replaces
 * instructions including addr (exclude breakpoint).
 */
N
Namhyung Kim 已提交
449
static struct kprobe *__kprobes get_optimized_kprobe(unsigned long addr)
450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
{
	int i;
	struct kprobe *p = NULL;
	struct optimized_kprobe *op;

	/* Don't check i == 0, since that is a breakpoint case. */
	for (i = 1; !p && i < MAX_OPTIMIZED_LENGTH; i++)
		p = get_kprobe((void *)(addr - i));

	if (p && kprobe_optready(p)) {
		op = container_of(p, struct optimized_kprobe, kp);
		if (arch_within_optimized_kprobe(op, addr))
			return p;
	}

	return NULL;
}

/* Optimization staging list, protected by kprobe_mutex */
static LIST_HEAD(optimizing_list);
470
static LIST_HEAD(unoptimizing_list);
471 472 473

static void kprobe_optimizer(struct work_struct *work);
static DECLARE_DELAYED_WORK(optimizing_work, kprobe_optimizer);
474
static DECLARE_COMPLETION(optimizer_comp);
475 476
#define OPTIMIZE_DELAY 5

477 478 479 480 481
/*
 * Optimize (replace a breakpoint with a jump) kprobes listed on
 * optimizing_list.
 */
static __kprobes void do_optimize_kprobes(void)
482
{
483 484 485 486 487
	/* Optimization never be done when disarmed */
	if (kprobes_all_disarmed || !kprobes_allow_optimization ||
	    list_empty(&optimizing_list))
		return;

488 489 490 491 492 493 494 495 496 497 498 499
	/*
	 * The optimization/unoptimization refers online_cpus via
	 * stop_machine() and cpu-hotplug modifies online_cpus.
	 * And same time, text_mutex will be held in cpu-hotplug and here.
	 * This combination can cause a deadlock (cpu-hotplug try to lock
	 * text_mutex but stop_machine can not be done because online_cpus
	 * has been changed)
	 * To avoid this deadlock, we need to call get_online_cpus()
	 * for preventing cpu-hotplug outside of text_mutex locking.
	 */
	get_online_cpus();
	mutex_lock(&text_mutex);
500
	arch_optimize_kprobes(&optimizing_list);
501 502
	mutex_unlock(&text_mutex);
	put_online_cpus();
503 504
}

505 506 507 508 509 510 511 512 513 514 515 516 517 518 519
/*
 * Unoptimize (replace a jump with a breakpoint and remove the breakpoint
 * if need) kprobes listed on unoptimizing_list.
 */
static __kprobes void do_unoptimize_kprobes(struct list_head *free_list)
{
	struct optimized_kprobe *op, *tmp;

	/* Unoptimization must be done anytime */
	if (list_empty(&unoptimizing_list))
		return;

	/* Ditto to do_optimize_kprobes */
	get_online_cpus();
	mutex_lock(&text_mutex);
520 521 522
	arch_unoptimize_kprobes(&unoptimizing_list, free_list);
	/* Loop free_list for disarming */
	list_for_each_entry_safe(op, tmp, free_list, list) {
523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558
		/* Disarm probes if marked disabled */
		if (kprobe_disabled(&op->kp))
			arch_disarm_kprobe(&op->kp);
		if (kprobe_unused(&op->kp)) {
			/*
			 * Remove unused probes from hash list. After waiting
			 * for synchronization, these probes are reclaimed.
			 * (reclaiming is done by do_free_cleaned_kprobes.)
			 */
			hlist_del_rcu(&op->kp.hlist);
		} else
			list_del_init(&op->list);
	}
	mutex_unlock(&text_mutex);
	put_online_cpus();
}

/* Reclaim all kprobes on the free_list */
static __kprobes void do_free_cleaned_kprobes(struct list_head *free_list)
{
	struct optimized_kprobe *op, *tmp;

	list_for_each_entry_safe(op, tmp, free_list, list) {
		BUG_ON(!kprobe_unused(&op->kp));
		list_del_init(&op->list);
		free_aggr_kprobe(&op->kp);
	}
}

/* Start optimizer after OPTIMIZE_DELAY passed */
static __kprobes void kick_kprobe_optimizer(void)
{
	if (!delayed_work_pending(&optimizing_work))
		schedule_delayed_work(&optimizing_work, OPTIMIZE_DELAY);
}

559 560 561
/* Kprobe jump optimizer */
static __kprobes void kprobe_optimizer(struct work_struct *work)
{
562 563
	LIST_HEAD(free_list);

564 565 566 567 568
	/* Lock modules while optimizing kprobes */
	mutex_lock(&module_mutex);
	mutex_lock(&kprobe_mutex);

	/*
569 570 571 572 573 574 575
	 * Step 1: Unoptimize kprobes and collect cleaned (unused and disarmed)
	 * kprobes before waiting for quiesence period.
	 */
	do_unoptimize_kprobes(&free_list);

	/*
	 * Step 2: Wait for quiesence period to ensure all running interrupts
576 577 578 579 580 581 582
	 * are done. Because optprobe may modify multiple instructions
	 * there is a chance that Nth instruction is interrupted. In that
	 * case, running interrupt can return to 2nd-Nth byte of jump
	 * instruction. This wait is for avoiding it.
	 */
	synchronize_sched();

583
	/* Step 3: Optimize kprobes after quiesence period */
584
	do_optimize_kprobes();
585 586 587 588

	/* Step 4: Free cleaned kprobes after quiesence period */
	do_free_cleaned_kprobes(&free_list);

589 590
	mutex_unlock(&kprobe_mutex);
	mutex_unlock(&module_mutex);
591

592
	/* Step 5: Kick optimizer again if needed */
593
	if (!list_empty(&optimizing_list) || !list_empty(&unoptimizing_list))
594 595 596 597
		kick_kprobe_optimizer();
	else
		/* Wake up all waiters */
		complete_all(&optimizer_comp);
598 599 600 601 602 603 604
}

/* Wait for completing optimization and unoptimization */
static __kprobes void wait_for_kprobe_optimizer(void)
{
	if (delayed_work_pending(&optimizing_work))
		wait_for_completion(&optimizer_comp);
605 606 607 608 609 610 611 612
}

/* Optimize kprobe if p is ready to be optimized */
static __kprobes void optimize_kprobe(struct kprobe *p)
{
	struct optimized_kprobe *op;

	/* Check if the kprobe is disabled or not ready for optimization. */
613
	if (!kprobe_optready(p) || !kprobes_allow_optimization ||
614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630
	    (kprobe_disabled(p) || kprobes_all_disarmed))
		return;

	/* Both of break_handler and post_handler are not supported. */
	if (p->break_handler || p->post_handler)
		return;

	op = container_of(p, struct optimized_kprobe, kp);

	/* Check there is no other kprobes at the optimized instructions */
	if (arch_check_optimized_kprobe(op) < 0)
		return;

	/* Check if it is already optimized. */
	if (op->kp.flags & KPROBE_FLAG_OPTIMIZED)
		return;
	op->kp.flags |= KPROBE_FLAG_OPTIMIZED;
631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648

	if (!list_empty(&op->list))
		/* This is under unoptimizing. Just dequeue the probe */
		list_del_init(&op->list);
	else {
		list_add(&op->list, &optimizing_list);
		kick_kprobe_optimizer();
	}
}

/* Short cut to direct unoptimizing */
static __kprobes void force_unoptimize_kprobe(struct optimized_kprobe *op)
{
	get_online_cpus();
	arch_unoptimize_kprobe(op);
	put_online_cpus();
	if (kprobe_disabled(&op->kp))
		arch_disarm_kprobe(&op->kp);
649 650 651
}

/* Unoptimize a kprobe if p is optimized */
652
static __kprobes void unoptimize_kprobe(struct kprobe *p, bool force)
653 654 655
{
	struct optimized_kprobe *op;

656 657 658 659 660 661 662 663 664 665 666 667
	if (!kprobe_aggrprobe(p) || kprobe_disarmed(p))
		return; /* This is not an optprobe nor optimized */

	op = container_of(p, struct optimized_kprobe, kp);
	if (!kprobe_optimized(p)) {
		/* Unoptimized or unoptimizing case */
		if (force && !list_empty(&op->list)) {
			/*
			 * Only if this is unoptimizing kprobe and forced,
			 * forcibly unoptimize it. (No need to unoptimize
			 * unoptimized kprobe again :)
			 */
668
			list_del_init(&op->list);
669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686
			force_unoptimize_kprobe(op);
		}
		return;
	}

	op->kp.flags &= ~KPROBE_FLAG_OPTIMIZED;
	if (!list_empty(&op->list)) {
		/* Dequeue from the optimization queue */
		list_del_init(&op->list);
		return;
	}
	/* Optimized kprobe case */
	if (force)
		/* Forcibly update the code: this is a special case */
		force_unoptimize_kprobe(op);
	else {
		list_add(&op->list, &unoptimizing_list);
		kick_kprobe_optimizer();
687 688 689
	}
}

M
Masami Hiramatsu 已提交
690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710
/* Cancel unoptimizing for reusing */
static void reuse_unused_kprobe(struct kprobe *ap)
{
	struct optimized_kprobe *op;

	BUG_ON(!kprobe_unused(ap));
	/*
	 * Unused kprobe MUST be on the way of delayed unoptimizing (means
	 * there is still a relative jump) and disabled.
	 */
	op = container_of(ap, struct optimized_kprobe, kp);
	if (unlikely(list_empty(&op->list)))
		printk(KERN_WARNING "Warning: found a stray unused "
			"aggrprobe@%p\n", ap->addr);
	/* Enable the probe again */
	ap->flags &= ~KPROBE_FLAG_DISABLED;
	/* Optimize it again (remove from op->list) */
	BUG_ON(!kprobe_optready(ap));
	optimize_kprobe(ap);
}

711 712 713 714 715 716
/* Remove optimized instructions */
static void __kprobes kill_optimized_kprobe(struct kprobe *p)
{
	struct optimized_kprobe *op;

	op = container_of(p, struct optimized_kprobe, kp);
717 718
	if (!list_empty(&op->list))
		/* Dequeue from the (un)optimization queue */
719
		list_del_init(&op->list);
720 721 722

	op->kp.flags &= ~KPROBE_FLAG_OPTIMIZED;
	/* Don't touch the code, because it is already freed. */
723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768
	arch_remove_optimized_kprobe(op);
}

/* Try to prepare optimized instructions */
static __kprobes void prepare_optimized_kprobe(struct kprobe *p)
{
	struct optimized_kprobe *op;

	op = container_of(p, struct optimized_kprobe, kp);
	arch_prepare_optimized_kprobe(op);
}

/* Allocate new optimized_kprobe and try to prepare optimized instructions */
static __kprobes struct kprobe *alloc_aggr_kprobe(struct kprobe *p)
{
	struct optimized_kprobe *op;

	op = kzalloc(sizeof(struct optimized_kprobe), GFP_KERNEL);
	if (!op)
		return NULL;

	INIT_LIST_HEAD(&op->list);
	op->kp.addr = p->addr;
	arch_prepare_optimized_kprobe(op);

	return &op->kp;
}

static void __kprobes init_aggr_kprobe(struct kprobe *ap, struct kprobe *p);

/*
 * Prepare an optimized_kprobe and optimize it
 * NOTE: p must be a normal registered kprobe
 */
static __kprobes void try_to_optimize_kprobe(struct kprobe *p)
{
	struct kprobe *ap;
	struct optimized_kprobe *op;

	ap = alloc_aggr_kprobe(p);
	if (!ap)
		return;

	op = container_of(ap, struct optimized_kprobe, kp);
	if (!arch_prepared_optinsn(&op->optinsn)) {
		/* If failed to setup optimizing, fallback to kprobe */
769 770
		arch_remove_optimized_kprobe(op);
		kfree(op);
771 772 773 774 775 776 777
		return;
	}

	init_aggr_kprobe(ap, p);
	optimize_kprobe(ap);
}

778
#ifdef CONFIG_SYSCTL
779
/* This should be called with kprobe_mutex locked */
780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800
static void __kprobes optimize_all_kprobes(void)
{
	struct hlist_head *head;
	struct hlist_node *node;
	struct kprobe *p;
	unsigned int i;

	/* If optimization is already allowed, just return */
	if (kprobes_allow_optimization)
		return;

	kprobes_allow_optimization = true;
	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
		head = &kprobe_table[i];
		hlist_for_each_entry_rcu(p, node, head, hlist)
			if (!kprobe_disabled(p))
				optimize_kprobe(p);
	}
	printk(KERN_INFO "Kprobes globally optimized\n");
}

801
/* This should be called with kprobe_mutex locked */
802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817
static void __kprobes unoptimize_all_kprobes(void)
{
	struct hlist_head *head;
	struct hlist_node *node;
	struct kprobe *p;
	unsigned int i;

	/* If optimization is already prohibited, just return */
	if (!kprobes_allow_optimization)
		return;

	kprobes_allow_optimization = false;
	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
		head = &kprobe_table[i];
		hlist_for_each_entry_rcu(p, node, head, hlist) {
			if (!kprobe_disabled(p))
818
				unoptimize_kprobe(p, false);
819 820
		}
	}
821 822 823
	/* Wait for unoptimizing completion */
	wait_for_kprobe_optimizer();
	printk(KERN_INFO "Kprobes globally unoptimized\n");
824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846
}

int sysctl_kprobes_optimization;
int proc_kprobes_optimization_handler(struct ctl_table *table, int write,
				      void __user *buffer, size_t *length,
				      loff_t *ppos)
{
	int ret;

	mutex_lock(&kprobe_mutex);
	sysctl_kprobes_optimization = kprobes_allow_optimization ? 1 : 0;
	ret = proc_dointvec_minmax(table, write, buffer, length, ppos);

	if (sysctl_kprobes_optimization)
		optimize_all_kprobes();
	else
		unoptimize_all_kprobes();
	mutex_unlock(&kprobe_mutex);

	return ret;
}
#endif /* CONFIG_SYSCTL */

847
/* Put a breakpoint for a probe. Must be called with text_mutex locked */
848 849
static void __kprobes __arm_kprobe(struct kprobe *p)
{
850
	struct kprobe *_p;
851 852

	/* Check collision with other optimized kprobes */
853 854
	_p = get_optimized_kprobe((unsigned long)p->addr);
	if (unlikely(_p))
855 856
		/* Fallback to unoptimized kprobe */
		unoptimize_kprobe(_p, true);
857 858 859 860 861

	arch_arm_kprobe(p);
	optimize_kprobe(p);	/* Try to optimize (add kprobe to a list) */
}

862 863
/* Remove the breakpoint of a probe. Must be called with text_mutex locked */
static void __kprobes __disarm_kprobe(struct kprobe *p, bool reopt)
864
{
865
	struct kprobe *_p;
866

867
	unoptimize_kprobe(p, false);	/* Try to unoptimize */
868

869 870 871 872 873 874 875 876
	if (!kprobe_queued(p)) {
		arch_disarm_kprobe(p);
		/* If another kprobe was blocked, optimize it. */
		_p = get_optimized_kprobe((unsigned long)p->addr);
		if (unlikely(_p) && reopt)
			optimize_kprobe(_p);
	}
	/* TODO: reoptimize others after unoptimized this probe */
877 878 879 880 881
}

#else /* !CONFIG_OPTPROBES */

#define optimize_kprobe(p)			do {} while (0)
882
#define unoptimize_kprobe(p, f)			do {} while (0)
883 884 885 886
#define kill_optimized_kprobe(p)		do {} while (0)
#define prepare_optimized_kprobe(p)		do {} while (0)
#define try_to_optimize_kprobe(p)		do {} while (0)
#define __arm_kprobe(p)				arch_arm_kprobe(p)
887 888 889
#define __disarm_kprobe(p, o)			arch_disarm_kprobe(p)
#define kprobe_disarmed(p)			kprobe_disabled(p)
#define wait_for_kprobe_optimizer()		do {} while (0)
890

M
Masami Hiramatsu 已提交
891 892 893 894 895 896 897
/* There should be no unused kprobes can be reused without optimization */
static void reuse_unused_kprobe(struct kprobe *ap)
{
	printk(KERN_ERR "Error: There should be no unused kprobe here.\n");
	BUG_ON(kprobe_unused(ap));
}

898 899
static __kprobes void free_aggr_kprobe(struct kprobe *p)
{
900
	arch_remove_kprobe(p);
901 902 903 904 905 906 907 908 909
	kfree(p);
}

static __kprobes struct kprobe *alloc_aggr_kprobe(struct kprobe *p)
{
	return kzalloc(sizeof(struct kprobe), GFP_KERNEL);
}
#endif /* CONFIG_OPTPROBES */

910 911 912
/* Arm a kprobe with text_mutex */
static void __kprobes arm_kprobe(struct kprobe *kp)
{
913 914 915 916 917
	/*
	 * Here, since __arm_kprobe() doesn't use stop_machine(),
	 * this doesn't cause deadlock on text_mutex. So, we don't
	 * need get_online_cpus().
	 */
918
	mutex_lock(&text_mutex);
919
	__arm_kprobe(kp);
920 921 922 923 924 925
	mutex_unlock(&text_mutex);
}

/* Disarm a kprobe with text_mutex */
static void __kprobes disarm_kprobe(struct kprobe *kp)
{
926
	/* Ditto */
927
	mutex_lock(&text_mutex);
928
	__disarm_kprobe(kp, true);
929 930 931
	mutex_unlock(&text_mutex);
}

932 933 934 935
/*
 * Aggregate handlers for multiple kprobes support - these handlers
 * take care of invoking the individual kprobe handlers on p->list
 */
936
static int __kprobes aggr_pre_handler(struct kprobe *p, struct pt_regs *regs)
937 938 939
{
	struct kprobe *kp;

940
	list_for_each_entry_rcu(kp, &p->list, list) {
941
		if (kp->pre_handler && likely(!kprobe_disabled(kp))) {
942
			set_kprobe_instance(kp);
943 944
			if (kp->pre_handler(kp, regs))
				return 1;
945
		}
946
		reset_kprobe_instance();
947 948 949 950
	}
	return 0;
}

951 952
static void __kprobes aggr_post_handler(struct kprobe *p, struct pt_regs *regs,
					unsigned long flags)
953 954 955
{
	struct kprobe *kp;

956
	list_for_each_entry_rcu(kp, &p->list, list) {
957
		if (kp->post_handler && likely(!kprobe_disabled(kp))) {
958
			set_kprobe_instance(kp);
959
			kp->post_handler(kp, regs, flags);
960
			reset_kprobe_instance();
961 962 963 964
		}
	}
}

965 966
static int __kprobes aggr_fault_handler(struct kprobe *p, struct pt_regs *regs,
					int trapnr)
967
{
C
Christoph Lameter 已提交
968
	struct kprobe *cur = __this_cpu_read(kprobe_instance);
969

970 971 972 973
	/*
	 * if we faulted "during" the execution of a user specified
	 * probe handler, invoke just that probe's fault handler
	 */
974 975
	if (cur && cur->fault_handler) {
		if (cur->fault_handler(cur, regs, trapnr))
976 977 978 979 980
			return 1;
	}
	return 0;
}

981
static int __kprobes aggr_break_handler(struct kprobe *p, struct pt_regs *regs)
982
{
C
Christoph Lameter 已提交
983
	struct kprobe *cur = __this_cpu_read(kprobe_instance);
984 985 986 987 988
	int ret = 0;

	if (cur && cur->break_handler) {
		if (cur->break_handler(cur, regs))
			ret = 1;
989
	}
990 991
	reset_kprobe_instance();
	return ret;
992 993
}

994 995 996 997
/* Walks the list and increments nmissed count for multiprobe case */
void __kprobes kprobes_inc_nmissed_count(struct kprobe *p)
{
	struct kprobe *kp;
998
	if (!kprobe_aggrprobe(p)) {
999 1000 1001 1002 1003 1004 1005 1006
		p->nmissed++;
	} else {
		list_for_each_entry_rcu(kp, &p->list, list)
			kp->nmissed++;
	}
	return;
}

1007 1008
void __kprobes recycle_rp_inst(struct kretprobe_instance *ri,
				struct hlist_head *head)
1009
{
1010 1011
	struct kretprobe *rp = ri->rp;

1012 1013
	/* remove rp inst off the rprobe_inst_table */
	hlist_del(&ri->hlist);
1014 1015 1016 1017 1018
	INIT_HLIST_NODE(&ri->hlist);
	if (likely(rp)) {
		spin_lock(&rp->lock);
		hlist_add_head(&ri->hlist, &rp->free_instances);
		spin_unlock(&rp->lock);
1019 1020
	} else
		/* Unregistering */
1021
		hlist_add_head(&ri->hlist, head);
1022 1023
}

1024
void __kprobes kretprobe_hash_lock(struct task_struct *tsk,
1025
			 struct hlist_head **head, unsigned long *flags)
1026
__acquires(hlist_lock)
1027 1028 1029 1030 1031 1032 1033 1034 1035
{
	unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS);
	spinlock_t *hlist_lock;

	*head = &kretprobe_inst_table[hash];
	hlist_lock = kretprobe_table_lock_ptr(hash);
	spin_lock_irqsave(hlist_lock, *flags);
}

1036 1037
static void __kprobes kretprobe_table_lock(unsigned long hash,
	unsigned long *flags)
1038
__acquires(hlist_lock)
1039
{
1040 1041 1042 1043
	spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash);
	spin_lock_irqsave(hlist_lock, *flags);
}

1044 1045
void __kprobes kretprobe_hash_unlock(struct task_struct *tsk,
	unsigned long *flags)
1046
__releases(hlist_lock)
1047 1048 1049 1050 1051 1052 1053 1054
{
	unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS);
	spinlock_t *hlist_lock;

	hlist_lock = kretprobe_table_lock_ptr(hash);
	spin_unlock_irqrestore(hlist_lock, *flags);
}

N
Namhyung Kim 已提交
1055 1056
static void __kprobes kretprobe_table_unlock(unsigned long hash,
       unsigned long *flags)
1057
__releases(hlist_lock)
1058 1059 1060
{
	spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash);
	spin_unlock_irqrestore(hlist_lock, *flags);
1061 1062 1063
}

/*
1064 1065 1066 1067
 * This function is called from finish_task_switch when task tk becomes dead,
 * so that we can recycle any function-return probe instances associated
 * with this task. These left over instances represent probed functions
 * that have been called but will never return.
1068
 */
1069
void __kprobes kprobe_flush_task(struct task_struct *tk)
1070
{
B
bibo,mao 已提交
1071
	struct kretprobe_instance *ri;
1072
	struct hlist_head *head, empty_rp;
1073
	struct hlist_node *node, *tmp;
1074
	unsigned long hash, flags = 0;
1075

1076 1077 1078 1079 1080 1081 1082
	if (unlikely(!kprobes_initialized))
		/* Early boot.  kretprobe_table_locks not yet initialized. */
		return;

	hash = hash_ptr(tk, KPROBE_HASH_BITS);
	head = &kretprobe_inst_table[hash];
	kretprobe_table_lock(hash, &flags);
B
bibo,mao 已提交
1083 1084
	hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
		if (ri->task == tk)
1085
			recycle_rp_inst(ri, &empty_rp);
B
bibo,mao 已提交
1086
	}
1087 1088
	kretprobe_table_unlock(hash, &flags);
	INIT_HLIST_HEAD(&empty_rp);
1089 1090 1091 1092
	hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) {
		hlist_del(&ri->hlist);
		kfree(ri);
	}
1093 1094 1095 1096 1097
}

static inline void free_rp_inst(struct kretprobe *rp)
{
	struct kretprobe_instance *ri;
1098 1099
	struct hlist_node *pos, *next;

1100 1101
	hlist_for_each_entry_safe(ri, pos, next, &rp->free_instances, hlist) {
		hlist_del(&ri->hlist);
1102 1103 1104 1105
		kfree(ri);
	}
}

1106 1107
static void __kprobes cleanup_rp_inst(struct kretprobe *rp)
{
1108
	unsigned long flags, hash;
1109 1110
	struct kretprobe_instance *ri;
	struct hlist_node *pos, *next;
1111 1112
	struct hlist_head *head;

1113
	/* No race here */
1114 1115 1116 1117 1118 1119 1120 1121
	for (hash = 0; hash < KPROBE_TABLE_SIZE; hash++) {
		kretprobe_table_lock(hash, &flags);
		head = &kretprobe_inst_table[hash];
		hlist_for_each_entry_safe(ri, pos, next, head, hlist) {
			if (ri->rp == rp)
				ri->rp = NULL;
		}
		kretprobe_table_unlock(hash, &flags);
1122 1123 1124 1125
	}
	free_rp_inst(rp);
}

1126
/*
1127
* Add the new probe to ap->list. Fail if this is the
1128 1129
* second jprobe at the address - two jprobes can't coexist
*/
1130
static int __kprobes add_new_kprobe(struct kprobe *ap, struct kprobe *p)
1131
{
1132
	BUG_ON(kprobe_gone(ap) || kprobe_gone(p));
1133 1134

	if (p->break_handler || p->post_handler)
1135
		unoptimize_kprobe(ap, true);	/* Fall back to normal kprobe */
1136

1137
	if (p->break_handler) {
1138
		if (ap->break_handler)
1139
			return -EEXIST;
1140 1141
		list_add_tail_rcu(&p->list, &ap->list);
		ap->break_handler = aggr_break_handler;
1142
	} else
1143 1144 1145
		list_add_rcu(&p->list, &ap->list);
	if (p->post_handler && !ap->post_handler)
		ap->post_handler = aggr_post_handler;
1146 1147 1148 1149 1150

	if (kprobe_disabled(ap) && !kprobe_disabled(p)) {
		ap->flags &= ~KPROBE_FLAG_DISABLED;
		if (!kprobes_all_disarmed)
			/* Arm the breakpoint again. */
1151
			__arm_kprobe(ap);
1152
	}
1153 1154 1155
	return 0;
}

1156 1157 1158 1159
/*
 * Fill in the required fields of the "manager kprobe". Replace the
 * earlier kprobe in the hlist with the manager kprobe
 */
1160
static void __kprobes init_aggr_kprobe(struct kprobe *ap, struct kprobe *p)
1161
{
1162
	/* Copy p's insn slot to ap */
1163
	copy_kprobe(p, ap);
1164
	flush_insn_slot(ap);
1165
	ap->addr = p->addr;
1166
	ap->flags = p->flags & ~KPROBE_FLAG_OPTIMIZED;
1167 1168
	ap->pre_handler = aggr_pre_handler;
	ap->fault_handler = aggr_fault_handler;
1169 1170
	/* We don't care the kprobe which has gone. */
	if (p->post_handler && !kprobe_gone(p))
1171
		ap->post_handler = aggr_post_handler;
1172
	if (p->break_handler && !kprobe_gone(p))
1173
		ap->break_handler = aggr_break_handler;
1174 1175

	INIT_LIST_HEAD(&ap->list);
1176
	INIT_HLIST_NODE(&ap->hlist);
1177

1178
	list_add_rcu(&p->list, &ap->list);
1179
	hlist_replace_rcu(&p->hlist, &ap->hlist);
1180 1181 1182 1183 1184 1185
}

/*
 * This is the second or subsequent kprobe at the address - handle
 * the intricacies
 */
1186
static int __kprobes register_aggr_kprobe(struct kprobe *orig_p,
1187
					  struct kprobe *p)
1188 1189
{
	int ret = 0;
1190
	struct kprobe *ap = orig_p;
1191

1192 1193 1194
	if (!kprobe_aggrprobe(orig_p)) {
		/* If orig_p is not an aggr_kprobe, create new aggr_kprobe. */
		ap = alloc_aggr_kprobe(orig_p);
1195 1196
		if (!ap)
			return -ENOMEM;
1197
		init_aggr_kprobe(ap, orig_p);
1198
	} else if (kprobe_unused(ap))
M
Masami Hiramatsu 已提交
1199 1200
		/* This probe is going to die. Rescue it */
		reuse_unused_kprobe(ap);
1201 1202

	if (kprobe_gone(ap)) {
1203 1204 1205 1206 1207 1208
		/*
		 * Attempting to insert new probe at the same location that
		 * had a probe in the module vaddr area which already
		 * freed. So, the instruction slot has already been
		 * released. We need a new slot for the new probe.
		 */
1209
		ret = arch_prepare_kprobe(ap);
1210
		if (ret)
1211 1212 1213 1214 1215
			/*
			 * Even if fail to allocate new slot, don't need to
			 * free aggr_probe. It will be used next time, or
			 * freed by unregister_kprobe.
			 */
1216
			return ret;
1217

1218 1219 1220
		/* Prepare optimized instructions if possible. */
		prepare_optimized_kprobe(ap);

1221
		/*
1222 1223
		 * Clear gone flag to prevent allocating new slot again, and
		 * set disabled flag because it is not armed yet.
1224
		 */
1225 1226
		ap->flags = (ap->flags & ~KPROBE_FLAG_GONE)
			    | KPROBE_FLAG_DISABLED;
1227
	}
1228

1229
	/* Copy ap's insn slot to p */
1230 1231
	copy_kprobe(ap, p);
	return add_new_kprobe(ap, p);
1232 1233
}

1234 1235
static int __kprobes in_kprobes_functions(unsigned long addr)
{
1236 1237
	struct kprobe_blackpoint *kb;

1238 1239
	if (addr >= (unsigned long)__kprobes_text_start &&
	    addr < (unsigned long)__kprobes_text_end)
1240
		return -EINVAL;
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251
	/*
	 * If there exists a kprobe_blacklist, verify and
	 * fail any probe registration in the prohibited area
	 */
	for (kb = kprobe_blacklist; kb->name != NULL; kb++) {
		if (kb->start_addr) {
			if (addr >= kb->start_addr &&
			    addr < (kb->start_addr + kb->range))
				return -EINVAL;
		}
	}
1252 1253 1254
	return 0;
}

1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272
/*
 * If we have a symbol_name argument, look it up and add the offset field
 * to it. This way, we can specify a relative address to a symbol.
 */
static kprobe_opcode_t __kprobes *kprobe_addr(struct kprobe *p)
{
	kprobe_opcode_t *addr = p->addr;
	if (p->symbol_name) {
		if (addr)
			return NULL;
		kprobe_lookup_name(p->symbol_name, addr);
	}

	if (!addr)
		return NULL;
	return (kprobe_opcode_t *)(((char *)addr) + p->offset);
}

1273 1274 1275
/* Check passed kprobe is valid and return kprobe in kprobe_table. */
static struct kprobe * __kprobes __get_valid_kprobe(struct kprobe *p)
{
1276
	struct kprobe *ap, *list_p;
1277

1278 1279
	ap = get_kprobe(p->addr);
	if (unlikely(!ap))
1280 1281
		return NULL;

1282 1283
	if (p != ap) {
		list_for_each_entry_rcu(list_p, &ap->list, list)
1284 1285 1286 1287 1288 1289
			if (list_p == p)
			/* kprobe p is a valid probe */
				goto valid;
		return NULL;
	}
valid:
1290
	return ap;
1291 1292 1293 1294 1295 1296 1297 1298
}

/* Return error if the kprobe is being re-registered */
static inline int check_kprobe_rereg(struct kprobe *p)
{
	int ret = 0;

	mutex_lock(&kprobe_mutex);
1299
	if (__get_valid_kprobe(p))
1300 1301
		ret = -EINVAL;
	mutex_unlock(&kprobe_mutex);
1302

1303 1304 1305
	return ret;
}

1306
int __kprobes register_kprobe(struct kprobe *p)
L
Linus Torvalds 已提交
1307 1308
{
	int ret = 0;
1309
	struct kprobe *old_p;
1310
	struct module *probed_mod;
1311
	kprobe_opcode_t *addr;
1312

1313 1314
	addr = kprobe_addr(p);
	if (!addr)
1315
		return -EINVAL;
1316
	p->addr = addr;
1317

1318 1319 1320 1321
	ret = check_kprobe_rereg(p);
	if (ret)
		return ret;

1322
	jump_label_lock();
1323
	preempt_disable();
1324
	if (!kernel_text_address((unsigned long) p->addr) ||
1325
	    in_kprobes_functions((unsigned long) p->addr) ||
1326
	    ftrace_text_reserved(p->addr, p->addr) ||
1327 1328
	    jump_label_text_reserved(p->addr, p->addr))
		goto fail_with_jump_label;
1329

1330 1331 1332
	/* User can pass only KPROBE_FLAG_DISABLED to register_kprobe */
	p->flags &= KPROBE_FLAG_DISABLED;

1333 1334 1335
	/*
	 * Check if are we probing a module.
	 */
1336
	probed_mod = __module_text_address((unsigned long) p->addr);
1337 1338
	if (probed_mod) {
		/*
1339 1340
		 * We must hold a refcount of the probed module while updating
		 * its code to prohibit unexpected unloading.
1341
		 */
1342 1343 1344
		if (unlikely(!try_module_get(probed_mod)))
			goto fail_with_jump_label;

1345 1346 1347 1348 1349 1350 1351
		/*
		 * If the module freed .init.text, we couldn't insert
		 * kprobes in there.
		 */
		if (within_module_init((unsigned long)p->addr, probed_mod) &&
		    probed_mod->state != MODULE_STATE_COMING) {
			module_put(probed_mod);
1352
			goto fail_with_jump_label;
1353
		}
1354
	}
1355
	preempt_enable();
1356
	jump_label_unlock();
L
Linus Torvalds 已提交
1357

1358
	p->nmissed = 0;
1359
	INIT_LIST_HEAD(&p->list);
I
Ingo Molnar 已提交
1360
	mutex_lock(&kprobe_mutex);
1361

1362 1363
	jump_label_lock(); /* needed to call jump_label_text_reserved() */

1364 1365 1366
	get_online_cpus();	/* For avoiding text_mutex deadlock. */
	mutex_lock(&text_mutex);

1367 1368
	old_p = get_kprobe(p->addr);
	if (old_p) {
1369
		/* Since this may unoptimize old_p, locking text_mutex. */
1370
		ret = register_aggr_kprobe(old_p, p);
L
Linus Torvalds 已提交
1371 1372 1373
		goto out;
	}

1374 1375
	ret = arch_prepare_kprobe(p);
	if (ret)
1376
		goto out;
1377

1378
	INIT_HLIST_NODE(&p->hlist);
1379
	hlist_add_head_rcu(&p->hlist,
L
Linus Torvalds 已提交
1380 1381
		       &kprobe_table[hash_ptr(p->addr, KPROBE_HASH_BITS)]);

1382
	if (!kprobes_all_disarmed && !kprobe_disabled(p))
1383 1384 1385 1386
		__arm_kprobe(p);

	/* Try to optimize kprobe */
	try_to_optimize_kprobe(p);
1387

L
Linus Torvalds 已提交
1388
out:
1389 1390
	mutex_unlock(&text_mutex);
	put_online_cpus();
1391
	jump_label_unlock();
I
Ingo Molnar 已提交
1392
	mutex_unlock(&kprobe_mutex);
1393

1394
	if (probed_mod)
1395
		module_put(probed_mod);
1396

L
Linus Torvalds 已提交
1397
	return ret;
1398 1399 1400 1401 1402

fail_with_jump_label:
	preempt_enable();
	jump_label_unlock();
	return -EINVAL;
L
Linus Torvalds 已提交
1403
}
1404
EXPORT_SYMBOL_GPL(register_kprobe);
L
Linus Torvalds 已提交
1405

1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446
/* Check if all probes on the aggrprobe are disabled */
static int __kprobes aggr_kprobe_disabled(struct kprobe *ap)
{
	struct kprobe *kp;

	list_for_each_entry_rcu(kp, &ap->list, list)
		if (!kprobe_disabled(kp))
			/*
			 * There is an active probe on the list.
			 * We can't disable this ap.
			 */
			return 0;

	return 1;
}

/* Disable one kprobe: Make sure called under kprobe_mutex is locked */
static struct kprobe *__kprobes __disable_kprobe(struct kprobe *p)
{
	struct kprobe *orig_p;

	/* Get an original kprobe for return */
	orig_p = __get_valid_kprobe(p);
	if (unlikely(orig_p == NULL))
		return NULL;

	if (!kprobe_disabled(p)) {
		/* Disable probe if it is a child probe */
		if (p != orig_p)
			p->flags |= KPROBE_FLAG_DISABLED;

		/* Try to disarm and disable this/parent probe */
		if (p == orig_p || aggr_kprobe_disabled(orig_p)) {
			disarm_kprobe(orig_p);
			orig_p->flags |= KPROBE_FLAG_DISABLED;
		}
	}

	return orig_p;
}

1447 1448 1449 1450 1451
/*
 * Unregister a kprobe without a scheduler synchronization.
 */
static int __kprobes __unregister_kprobe_top(struct kprobe *p)
{
1452
	struct kprobe *ap, *list_p;
1453

1454 1455
	/* Disable kprobe. This will disarm it if needed. */
	ap = __disable_kprobe(p);
1456
	if (ap == NULL)
1457 1458
		return -EINVAL;

1459
	if (ap == p)
1460
		/*
1461 1462
		 * This probe is an independent(and non-optimized) kprobe
		 * (not an aggrprobe). Remove from the hash list.
1463
		 */
1464 1465 1466 1467 1468
		goto disarmed;

	/* Following process expects this probe is an aggrprobe */
	WARN_ON(!kprobe_aggrprobe(ap));

1469 1470 1471 1472 1473
	if (list_is_singular(&ap->list) && kprobe_disarmed(ap))
		/*
		 * !disarmed could be happen if the probe is under delayed
		 * unoptimizing.
		 */
1474 1475 1476
		goto disarmed;
	else {
		/* If disabling probe has special handlers, update aggrprobe */
1477
		if (p->break_handler && !kprobe_gone(p))
1478
			ap->break_handler = NULL;
1479
		if (p->post_handler && !kprobe_gone(p)) {
1480
			list_for_each_entry_rcu(list_p, &ap->list, list) {
1481 1482 1483
				if ((list_p != p) && (list_p->post_handler))
					goto noclean;
			}
1484
			ap->post_handler = NULL;
1485 1486
		}
noclean:
1487 1488 1489 1490
		/*
		 * Remove from the aggrprobe: this path will do nothing in
		 * __unregister_kprobe_bottom().
		 */
1491
		list_del_rcu(&p->list);
1492 1493 1494 1495 1496 1497
		if (!kprobe_disabled(ap) && !kprobes_all_disarmed)
			/*
			 * Try to optimize this probe again, because post
			 * handler may have been changed.
			 */
			optimize_kprobe(ap);
1498
	}
1499
	return 0;
1500 1501

disarmed:
1502
	BUG_ON(!kprobe_disarmed(ap));
1503 1504
	hlist_del_rcu(&ap->hlist);
	return 0;
1505
}
1506

1507 1508
static void __kprobes __unregister_kprobe_bottom(struct kprobe *p)
{
1509
	struct kprobe *ap;
1510

1511
	if (list_empty(&p->list))
1512
		/* This is an independent kprobe */
1513
		arch_remove_kprobe(p);
1514
	else if (list_is_singular(&p->list)) {
1515
		/* This is the last child of an aggrprobe */
1516
		ap = list_entry(p->list.next, struct kprobe, list);
1517
		list_del(&p->list);
1518
		free_aggr_kprobe(ap);
1519
	}
1520
	/* Otherwise, do nothing. */
1521 1522
}

1523
int __kprobes register_kprobes(struct kprobe **kps, int num)
1524 1525 1526 1527 1528 1529
{
	int i, ret = 0;

	if (num <= 0)
		return -EINVAL;
	for (i = 0; i < num; i++) {
1530
		ret = register_kprobe(kps[i]);
1531 1532 1533
		if (ret < 0) {
			if (i > 0)
				unregister_kprobes(kps, i);
1534
			break;
1535
		}
1536
	}
1537 1538
	return ret;
}
1539
EXPORT_SYMBOL_GPL(register_kprobes);
1540 1541 1542 1543 1544

void __kprobes unregister_kprobe(struct kprobe *p)
{
	unregister_kprobes(&p, 1);
}
1545
EXPORT_SYMBOL_GPL(unregister_kprobe);
1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562

void __kprobes unregister_kprobes(struct kprobe **kps, int num)
{
	int i;

	if (num <= 0)
		return;
	mutex_lock(&kprobe_mutex);
	for (i = 0; i < num; i++)
		if (__unregister_kprobe_top(kps[i]) < 0)
			kps[i]->addr = NULL;
	mutex_unlock(&kprobe_mutex);

	synchronize_sched();
	for (i = 0; i < num; i++)
		if (kps[i]->addr)
			__unregister_kprobe_bottom(kps[i]);
L
Linus Torvalds 已提交
1563
}
1564
EXPORT_SYMBOL_GPL(unregister_kprobes);
L
Linus Torvalds 已提交
1565 1566

static struct notifier_block kprobe_exceptions_nb = {
1567 1568 1569 1570
	.notifier_call = kprobe_exceptions_notify,
	.priority = 0x7fffffff /* we need to be notified first */
};

1571 1572 1573 1574
unsigned long __weak arch_deref_entry_point(void *entry)
{
	return (unsigned long)entry;
}
L
Linus Torvalds 已提交
1575

1576
int __kprobes register_jprobes(struct jprobe **jps, int num)
L
Linus Torvalds 已提交
1577
{
1578 1579
	struct jprobe *jp;
	int ret = 0, i;
1580

1581
	if (num <= 0)
1582
		return -EINVAL;
1583
	for (i = 0; i < num; i++) {
1584
		unsigned long addr, offset;
1585 1586 1587
		jp = jps[i];
		addr = arch_deref_entry_point(jp->entry);

1588 1589 1590 1591 1592 1593 1594 1595
		/* Verify probepoint is a function entry point */
		if (kallsyms_lookup_size_offset(addr, NULL, &offset) &&
		    offset == 0) {
			jp->kp.pre_handler = setjmp_pre_handler;
			jp->kp.break_handler = longjmp_break_handler;
			ret = register_kprobe(&jp->kp);
		} else
			ret = -EINVAL;
1596

1597 1598 1599
		if (ret < 0) {
			if (i > 0)
				unregister_jprobes(jps, i);
1600 1601 1602 1603 1604
			break;
		}
	}
	return ret;
}
1605
EXPORT_SYMBOL_GPL(register_jprobes);
1606

1607 1608
int __kprobes register_jprobe(struct jprobe *jp)
{
1609
	return register_jprobes(&jp, 1);
L
Linus Torvalds 已提交
1610
}
1611
EXPORT_SYMBOL_GPL(register_jprobe);
L
Linus Torvalds 已提交
1612

1613
void __kprobes unregister_jprobe(struct jprobe *jp)
L
Linus Torvalds 已提交
1614
{
1615 1616
	unregister_jprobes(&jp, 1);
}
1617
EXPORT_SYMBOL_GPL(unregister_jprobe);
1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635

void __kprobes unregister_jprobes(struct jprobe **jps, int num)
{
	int i;

	if (num <= 0)
		return;
	mutex_lock(&kprobe_mutex);
	for (i = 0; i < num; i++)
		if (__unregister_kprobe_top(&jps[i]->kp) < 0)
			jps[i]->kp.addr = NULL;
	mutex_unlock(&kprobe_mutex);

	synchronize_sched();
	for (i = 0; i < num; i++) {
		if (jps[i]->kp.addr)
			__unregister_kprobe_bottom(&jps[i]->kp);
	}
L
Linus Torvalds 已提交
1636
}
1637
EXPORT_SYMBOL_GPL(unregister_jprobes);
L
Linus Torvalds 已提交
1638

1639
#ifdef CONFIG_KRETPROBES
1640 1641 1642 1643 1644 1645 1646 1647
/*
 * This kprobe pre_handler is registered with every kretprobe. When probe
 * hits it will set up the return probe.
 */
static int __kprobes pre_handler_kretprobe(struct kprobe *p,
					   struct pt_regs *regs)
{
	struct kretprobe *rp = container_of(p, struct kretprobe, kp);
1648 1649
	unsigned long hash, flags = 0;
	struct kretprobe_instance *ri;
1650 1651

	/*TODO: consider to only swap the RA after the last pre_handler fired */
1652 1653
	hash = hash_ptr(current, KPROBE_HASH_BITS);
	spin_lock_irqsave(&rp->lock, flags);
1654 1655
	if (!hlist_empty(&rp->free_instances)) {
		ri = hlist_entry(rp->free_instances.first,
1656 1657 1658 1659
				struct kretprobe_instance, hlist);
		hlist_del(&ri->hlist);
		spin_unlock_irqrestore(&rp->lock, flags);

1660 1661
		ri->rp = rp;
		ri->task = current;
1662

1663
		if (rp->entry_handler && rp->entry_handler(ri, regs))
1664 1665
			return 0;

1666 1667 1668
		arch_prepare_kretprobe(ri, regs);

		/* XXX(hch): why is there no hlist_move_head? */
1669 1670 1671 1672 1673
		INIT_HLIST_NODE(&ri->hlist);
		kretprobe_table_lock(hash, &flags);
		hlist_add_head(&ri->hlist, &kretprobe_inst_table[hash]);
		kretprobe_table_unlock(hash, &flags);
	} else {
1674
		rp->nmissed++;
1675 1676
		spin_unlock_irqrestore(&rp->lock, flags);
	}
1677 1678 1679
	return 0;
}

1680
int __kprobes register_kretprobe(struct kretprobe *rp)
1681 1682 1683 1684
{
	int ret = 0;
	struct kretprobe_instance *inst;
	int i;
1685
	void *addr;
1686 1687

	if (kretprobe_blacklist_size) {
1688 1689 1690
		addr = kprobe_addr(&rp->kp);
		if (!addr)
			return -EINVAL;
1691 1692 1693 1694 1695 1696

		for (i = 0; kretprobe_blacklist[i].name != NULL; i++) {
			if (kretprobe_blacklist[i].addr == addr)
				return -EINVAL;
		}
	}
1697 1698

	rp->kp.pre_handler = pre_handler_kretprobe;
1699 1700 1701
	rp->kp.post_handler = NULL;
	rp->kp.fault_handler = NULL;
	rp->kp.break_handler = NULL;
1702 1703 1704 1705

	/* Pre-allocate memory for max kretprobe instances */
	if (rp->maxactive <= 0) {
#ifdef CONFIG_PREEMPT
1706
		rp->maxactive = max_t(unsigned int, 10, 2*num_possible_cpus());
1707
#else
1708
		rp->maxactive = num_possible_cpus();
1709 1710
#endif
	}
1711
	spin_lock_init(&rp->lock);
1712 1713
	INIT_HLIST_HEAD(&rp->free_instances);
	for (i = 0; i < rp->maxactive; i++) {
1714 1715
		inst = kmalloc(sizeof(struct kretprobe_instance) +
			       rp->data_size, GFP_KERNEL);
1716 1717 1718 1719
		if (inst == NULL) {
			free_rp_inst(rp);
			return -ENOMEM;
		}
1720 1721
		INIT_HLIST_NODE(&inst->hlist);
		hlist_add_head(&inst->hlist, &rp->free_instances);
1722 1723 1724 1725
	}

	rp->nmissed = 0;
	/* Establish function entry probe point */
1726
	ret = register_kprobe(&rp->kp);
1727
	if (ret != 0)
1728 1729 1730
		free_rp_inst(rp);
	return ret;
}
1731
EXPORT_SYMBOL_GPL(register_kretprobe);
1732

1733
int __kprobes register_kretprobes(struct kretprobe **rps, int num)
1734 1735 1736 1737 1738 1739
{
	int ret = 0, i;

	if (num <= 0)
		return -EINVAL;
	for (i = 0; i < num; i++) {
1740
		ret = register_kretprobe(rps[i]);
1741 1742 1743
		if (ret < 0) {
			if (i > 0)
				unregister_kretprobes(rps, i);
1744 1745 1746 1747 1748
			break;
		}
	}
	return ret;
}
1749
EXPORT_SYMBOL_GPL(register_kretprobes);
1750 1751 1752 1753 1754

void __kprobes unregister_kretprobe(struct kretprobe *rp)
{
	unregister_kretprobes(&rp, 1);
}
1755
EXPORT_SYMBOL_GPL(unregister_kretprobe);
1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776

void __kprobes unregister_kretprobes(struct kretprobe **rps, int num)
{
	int i;

	if (num <= 0)
		return;
	mutex_lock(&kprobe_mutex);
	for (i = 0; i < num; i++)
		if (__unregister_kprobe_top(&rps[i]->kp) < 0)
			rps[i]->kp.addr = NULL;
	mutex_unlock(&kprobe_mutex);

	synchronize_sched();
	for (i = 0; i < num; i++) {
		if (rps[i]->kp.addr) {
			__unregister_kprobe_bottom(&rps[i]->kp);
			cleanup_rp_inst(rps[i]);
		}
	}
}
1777
EXPORT_SYMBOL_GPL(unregister_kretprobes);
1778

1779
#else /* CONFIG_KRETPROBES */
1780
int __kprobes register_kretprobe(struct kretprobe *rp)
1781 1782 1783
{
	return -ENOSYS;
}
1784
EXPORT_SYMBOL_GPL(register_kretprobe);
1785

1786
int __kprobes register_kretprobes(struct kretprobe **rps, int num)
1787
{
1788
	return -ENOSYS;
1789
}
1790 1791
EXPORT_SYMBOL_GPL(register_kretprobes);

1792
void __kprobes unregister_kretprobe(struct kretprobe *rp)
1793
{
1794
}
1795
EXPORT_SYMBOL_GPL(unregister_kretprobe);
1796

1797 1798 1799
void __kprobes unregister_kretprobes(struct kretprobe **rps, int num)
{
}
1800
EXPORT_SYMBOL_GPL(unregister_kretprobes);
1801

1802 1803 1804 1805
static int __kprobes pre_handler_kretprobe(struct kprobe *p,
					   struct pt_regs *regs)
{
	return 0;
1806 1807
}

1808 1809
#endif /* CONFIG_KRETPROBES */

1810 1811 1812 1813
/* Set the kprobe gone and remove its instruction buffer. */
static void __kprobes kill_kprobe(struct kprobe *p)
{
	struct kprobe *kp;
1814

1815
	p->flags |= KPROBE_FLAG_GONE;
1816
	if (kprobe_aggrprobe(p)) {
1817 1818 1819 1820 1821 1822 1823 1824
		/*
		 * If this is an aggr_kprobe, we have to list all the
		 * chained probes and mark them GONE.
		 */
		list_for_each_entry_rcu(kp, &p->list, list)
			kp->flags |= KPROBE_FLAG_GONE;
		p->post_handler = NULL;
		p->break_handler = NULL;
1825
		kill_optimized_kprobe(p);
1826 1827 1828 1829 1830 1831 1832 1833
	}
	/*
	 * Here, we can remove insn_slot safely, because no thread calls
	 * the original probed function (which will be freed soon) any more.
	 */
	arch_remove_kprobe(p);
}

1834 1835 1836 1837 1838 1839 1840
/* Disable one kprobe */
int __kprobes disable_kprobe(struct kprobe *kp)
{
	int ret = 0;

	mutex_lock(&kprobe_mutex);

1841 1842
	/* Disable this kprobe */
	if (__disable_kprobe(kp) == NULL)
1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883
		ret = -EINVAL;

	mutex_unlock(&kprobe_mutex);
	return ret;
}
EXPORT_SYMBOL_GPL(disable_kprobe);

/* Enable one kprobe */
int __kprobes enable_kprobe(struct kprobe *kp)
{
	int ret = 0;
	struct kprobe *p;

	mutex_lock(&kprobe_mutex);

	/* Check whether specified probe is valid. */
	p = __get_valid_kprobe(kp);
	if (unlikely(p == NULL)) {
		ret = -EINVAL;
		goto out;
	}

	if (kprobe_gone(kp)) {
		/* This kprobe has gone, we couldn't enable it. */
		ret = -EINVAL;
		goto out;
	}

	if (p != kp)
		kp->flags &= ~KPROBE_FLAG_DISABLED;

	if (!kprobes_all_disarmed && kprobe_disabled(p)) {
		p->flags &= ~KPROBE_FLAG_DISABLED;
		arm_kprobe(p);
	}
out:
	mutex_unlock(&kprobe_mutex);
	return ret;
}
EXPORT_SYMBOL_GPL(enable_kprobe);

1884 1885 1886 1887 1888 1889 1890
void __kprobes dump_kprobe(struct kprobe *kp)
{
	printk(KERN_WARNING "Dumping kprobe:\n");
	printk(KERN_WARNING "Name: %s\nAddress: %p\nOffset: %x\n",
	       kp->symbol_name, kp->addr, kp->offset);
}

1891 1892 1893 1894 1895 1896 1897 1898 1899
/* Module notifier call back, checking kprobes on the module */
static int __kprobes kprobes_module_callback(struct notifier_block *nb,
					     unsigned long val, void *data)
{
	struct module *mod = data;
	struct hlist_head *head;
	struct hlist_node *node;
	struct kprobe *p;
	unsigned int i;
1900
	int checkcore = (val == MODULE_STATE_GOING);
1901

1902
	if (val != MODULE_STATE_GOING && val != MODULE_STATE_LIVE)
1903 1904 1905
		return NOTIFY_DONE;

	/*
1906 1907 1908 1909
	 * When MODULE_STATE_GOING was notified, both of module .text and
	 * .init.text sections would be freed. When MODULE_STATE_LIVE was
	 * notified, only .init.text section would be freed. We need to
	 * disable kprobes which have been inserted in the sections.
1910 1911 1912 1913 1914
	 */
	mutex_lock(&kprobe_mutex);
	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
		head = &kprobe_table[i];
		hlist_for_each_entry_rcu(p, node, head, hlist)
1915 1916 1917
			if (within_module_init((unsigned long)p->addr, mod) ||
			    (checkcore &&
			     within_module_core((unsigned long)p->addr, mod))) {
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934
				/*
				 * The vaddr this probe is installed will soon
				 * be vfreed buy not synced to disk. Hence,
				 * disarming the breakpoint isn't needed.
				 */
				kill_kprobe(p);
			}
	}
	mutex_unlock(&kprobe_mutex);
	return NOTIFY_DONE;
}

static struct notifier_block kprobe_module_nb = {
	.notifier_call = kprobes_module_callback,
	.priority = 0
};

L
Linus Torvalds 已提交
1935 1936 1937
static int __init init_kprobes(void)
{
	int i, err = 0;
1938 1939 1940 1941 1942
	unsigned long offset = 0, size = 0;
	char *modname, namebuf[128];
	const char *symbol_name;
	void *addr;
	struct kprobe_blackpoint *kb;
L
Linus Torvalds 已提交
1943 1944 1945

	/* FIXME allocate the probe table, currently defined statically */
	/* initialize all list heads */
1946
	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
L
Linus Torvalds 已提交
1947
		INIT_HLIST_HEAD(&kprobe_table[i]);
1948
		INIT_HLIST_HEAD(&kretprobe_inst_table[i]);
1949
		spin_lock_init(&(kretprobe_table_locks[i].lock));
1950
	}
L
Linus Torvalds 已提交
1951

1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973
	/*
	 * Lookup and populate the kprobe_blacklist.
	 *
	 * Unlike the kretprobe blacklist, we'll need to determine
	 * the range of addresses that belong to the said functions,
	 * since a kprobe need not necessarily be at the beginning
	 * of a function.
	 */
	for (kb = kprobe_blacklist; kb->name != NULL; kb++) {
		kprobe_lookup_name(kb->name, addr);
		if (!addr)
			continue;

		kb->start_addr = (unsigned long)addr;
		symbol_name = kallsyms_lookup(kb->start_addr,
				&size, &offset, &modname, namebuf);
		if (!symbol_name)
			kb->range = 0;
		else
			kb->range = size;
	}

1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984
	if (kretprobe_blacklist_size) {
		/* lookup the function address from its name */
		for (i = 0; kretprobe_blacklist[i].name != NULL; i++) {
			kprobe_lookup_name(kretprobe_blacklist[i].name,
					   kretprobe_blacklist[i].addr);
			if (!kretprobe_blacklist[i].addr)
				printk("kretprobe: lookup failed: %s\n",
				       kretprobe_blacklist[i].name);
		}
	}

1985 1986
#if defined(CONFIG_OPTPROBES)
#if defined(__ARCH_WANT_KPROBES_INSN_SLOT)
1987 1988 1989
	/* Init kprobe_optinsn_slots */
	kprobe_optinsn_slots.insn_size = MAX_OPTINSN_SIZE;
#endif
1990 1991 1992
	/* By default, kprobes can be optimized */
	kprobes_allow_optimization = true;
#endif
1993

1994 1995
	/* By default, kprobes are armed */
	kprobes_all_disarmed = false;
1996

1997
	err = arch_init_kprobes();
1998 1999
	if (!err)
		err = register_die_notifier(&kprobe_exceptions_nb);
2000 2001 2002
	if (!err)
		err = register_module_notifier(&kprobe_module_nb);

2003
	kprobes_initialized = (err == 0);
2004

2005 2006
	if (!err)
		init_test_probes();
L
Linus Torvalds 已提交
2007 2008 2009
	return err;
}

2010 2011
#ifdef CONFIG_DEBUG_FS
static void __kprobes report_probe(struct seq_file *pi, struct kprobe *p,
2012
		const char *sym, int offset, char *modname, struct kprobe *pp)
2013 2014 2015 2016 2017 2018 2019 2020 2021
{
	char *kprobe_type;

	if (p->pre_handler == pre_handler_kretprobe)
		kprobe_type = "r";
	else if (p->pre_handler == setjmp_pre_handler)
		kprobe_type = "j";
	else
		kprobe_type = "k";
2022

2023
	if (sym)
2024
		seq_printf(pi, "%p  %s  %s+0x%x  %s ",
2025
			p->addr, kprobe_type, sym, offset,
2026
			(modname ? modname : " "));
2027
	else
2028 2029 2030 2031 2032 2033 2034 2035 2036
		seq_printf(pi, "%p  %s  %p ",
			p->addr, kprobe_type, p->addr);

	if (!pp)
		pp = p;
	seq_printf(pi, "%s%s%s\n",
		(kprobe_gone(p) ? "[GONE]" : ""),
		((kprobe_disabled(p) && !kprobe_gone(p)) ?  "[DISABLED]" : ""),
		(kprobe_optimized(pp) ? "[OPTIMIZED]" : ""));
2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063
}

static void __kprobes *kprobe_seq_start(struct seq_file *f, loff_t *pos)
{
	return (*pos < KPROBE_TABLE_SIZE) ? pos : NULL;
}

static void __kprobes *kprobe_seq_next(struct seq_file *f, void *v, loff_t *pos)
{
	(*pos)++;
	if (*pos >= KPROBE_TABLE_SIZE)
		return NULL;
	return pos;
}

static void __kprobes kprobe_seq_stop(struct seq_file *f, void *v)
{
	/* Nothing to do */
}

static int __kprobes show_kprobe_addr(struct seq_file *pi, void *v)
{
	struct hlist_head *head;
	struct hlist_node *node;
	struct kprobe *p, *kp;
	const char *sym = NULL;
	unsigned int i = *(loff_t *) v;
A
Alexey Dobriyan 已提交
2064
	unsigned long offset = 0;
2065 2066 2067 2068 2069
	char *modname, namebuf[128];

	head = &kprobe_table[i];
	preempt_disable();
	hlist_for_each_entry_rcu(p, node, head, hlist) {
A
Alexey Dobriyan 已提交
2070
		sym = kallsyms_lookup((unsigned long)p->addr, NULL,
2071
					&offset, &modname, namebuf);
2072
		if (kprobe_aggrprobe(p)) {
2073
			list_for_each_entry_rcu(kp, &p->list, list)
2074
				report_probe(pi, kp, sym, offset, modname, p);
2075
		} else
2076
			report_probe(pi, p, sym, offset, modname, NULL);
2077 2078 2079 2080 2081
	}
	preempt_enable();
	return 0;
}

J
James Morris 已提交
2082
static const struct seq_operations kprobes_seq_ops = {
2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093
	.start = kprobe_seq_start,
	.next  = kprobe_seq_next,
	.stop  = kprobe_seq_stop,
	.show  = show_kprobe_addr
};

static int __kprobes kprobes_open(struct inode *inode, struct file *filp)
{
	return seq_open(filp, &kprobes_seq_ops);
}

2094
static const struct file_operations debugfs_kprobes_operations = {
2095 2096 2097 2098 2099 2100
	.open           = kprobes_open,
	.read           = seq_read,
	.llseek         = seq_lseek,
	.release        = seq_release,
};

2101
static void __kprobes arm_all_kprobes(void)
2102 2103 2104 2105 2106 2107 2108 2109
{
	struct hlist_head *head;
	struct hlist_node *node;
	struct kprobe *p;
	unsigned int i;

	mutex_lock(&kprobe_mutex);

2110 2111
	/* If kprobes are armed, just return */
	if (!kprobes_all_disarmed)
2112 2113
		goto already_enabled;

2114
	/* Arming kprobes doesn't optimize kprobe itself */
2115
	mutex_lock(&text_mutex);
2116 2117 2118
	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
		head = &kprobe_table[i];
		hlist_for_each_entry_rcu(p, node, head, hlist)
2119
			if (!kprobe_disabled(p))
2120
				__arm_kprobe(p);
2121
	}
2122
	mutex_unlock(&text_mutex);
2123

2124
	kprobes_all_disarmed = false;
2125 2126 2127 2128 2129 2130 2131
	printk(KERN_INFO "Kprobes globally enabled\n");

already_enabled:
	mutex_unlock(&kprobe_mutex);
	return;
}

2132
static void __kprobes disarm_all_kprobes(void)
2133 2134 2135 2136 2137 2138 2139 2140
{
	struct hlist_head *head;
	struct hlist_node *node;
	struct kprobe *p;
	unsigned int i;

	mutex_lock(&kprobe_mutex);

2141
	/* If kprobes are already disarmed, just return */
2142 2143 2144 2145
	if (kprobes_all_disarmed) {
		mutex_unlock(&kprobe_mutex);
		return;
	}
2146

2147
	kprobes_all_disarmed = true;
2148
	printk(KERN_INFO "Kprobes globally disabled\n");
2149

2150
	mutex_lock(&text_mutex);
2151 2152 2153
	for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
		head = &kprobe_table[i];
		hlist_for_each_entry_rcu(p, node, head, hlist) {
2154
			if (!arch_trampoline_kprobe(p) && !kprobe_disabled(p))
2155
				__disarm_kprobe(p, false);
2156 2157
		}
	}
2158
	mutex_unlock(&text_mutex);
2159 2160
	mutex_unlock(&kprobe_mutex);

2161 2162
	/* Wait for disarming all kprobes by optimizer */
	wait_for_kprobe_optimizer();
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174
}

/*
 * XXX: The debugfs bool file interface doesn't allow for callbacks
 * when the bool state is switched. We can reuse that facility when
 * available
 */
static ssize_t read_enabled_file_bool(struct file *file,
	       char __user *user_buf, size_t count, loff_t *ppos)
{
	char buf[3];

2175
	if (!kprobes_all_disarmed)
2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197
		buf[0] = '1';
	else
		buf[0] = '0';
	buf[1] = '\n';
	buf[2] = 0x00;
	return simple_read_from_buffer(user_buf, count, ppos, buf, 2);
}

static ssize_t write_enabled_file_bool(struct file *file,
	       const char __user *user_buf, size_t count, loff_t *ppos)
{
	char buf[32];
	int buf_size;

	buf_size = min(count, (sizeof(buf)-1));
	if (copy_from_user(buf, user_buf, buf_size))
		return -EFAULT;

	switch (buf[0]) {
	case 'y':
	case 'Y':
	case '1':
2198
		arm_all_kprobes();
2199 2200 2201 2202
		break;
	case 'n':
	case 'N':
	case '0':
2203
		disarm_all_kprobes();
2204 2205 2206 2207 2208 2209
		break;
	}

	return count;
}

2210
static const struct file_operations fops_kp = {
2211 2212
	.read =         read_enabled_file_bool,
	.write =        write_enabled_file_bool,
2213
	.llseek =	default_llseek,
2214 2215
};

2216 2217 2218
static int __kprobes debugfs_kprobe_init(void)
{
	struct dentry *dir, *file;
2219
	unsigned int value = 1;
2220 2221 2222 2223 2224

	dir = debugfs_create_dir("kprobes", NULL);
	if (!dir)
		return -ENOMEM;

R
Randy Dunlap 已提交
2225
	file = debugfs_create_file("list", 0444, dir, NULL,
2226 2227 2228 2229 2230 2231
				&debugfs_kprobes_operations);
	if (!file) {
		debugfs_remove(dir);
		return -ENOMEM;
	}

2232 2233 2234 2235 2236 2237 2238
	file = debugfs_create_file("enabled", 0600, dir,
					&value, &fops_kp);
	if (!file) {
		debugfs_remove(dir);
		return -ENOMEM;
	}

2239 2240 2241 2242 2243 2244 2245
	return 0;
}

late_initcall(debugfs_kprobe_init);
#endif /* CONFIG_DEBUG_FS */

module_init(init_kprobes);
L
Linus Torvalds 已提交
2246

2247
/* defined in arch/.../kernel/kprobes.c */
L
Linus Torvalds 已提交
2248
EXPORT_SYMBOL_GPL(jprobe_return);