kmemleak.c 54.5 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
/*
 * mm/kmemleak.c
 *
 * Copyright (C) 2008 ARM Limited
 * Written by Catalin Marinas <catalin.marinas@arm.com>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 * 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
 *
 *
 * For more information on the algorithm and kmemleak usage, please see
 * Documentation/kmemleak.txt.
 *
 * Notes on locking
 * ----------------
 *
 * The following locks and mutexes are used by kmemleak:
 *
 * - kmemleak_lock (rwlock): protects the object_list modifications and
 *   accesses to the object_tree_root. The object_list is the main list
 *   holding the metadata (struct kmemleak_object) for the allocated memory
32
 *   blocks. The object_tree_root is a red black tree used to look-up
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
 *   metadata based on a pointer to the corresponding memory block.  The
 *   kmemleak_object structures are added to the object_list and
 *   object_tree_root in the create_object() function called from the
 *   kmemleak_alloc() callback and removed in delete_object() called from the
 *   kmemleak_free() callback
 * - kmemleak_object.lock (spinlock): protects a kmemleak_object. Accesses to
 *   the metadata (e.g. count) are protected by this lock. Note that some
 *   members of this structure may be protected by other means (atomic or
 *   kmemleak_lock). This lock is also held when scanning the corresponding
 *   memory block to avoid the kernel freeing it via the kmemleak_free()
 *   callback. This is less heavyweight than holding a global lock like
 *   kmemleak_lock during scanning
 * - scan_mutex (mutex): ensures that only one thread may scan the memory for
 *   unreferenced objects at a time. The gray_list contains the objects which
 *   are already referenced or marked as false positives and need to be
 *   scanned. This list is only modified during a scanning episode when the
 *   scan_mutex is held. At the end of a scan, the gray_list is always empty.
 *   Note that the kmemleak_object.use_count is incremented when an object is
51 52 53 54
 *   added to the gray_list and therefore cannot be freed. This mutex also
 *   prevents multiple users of the "kmemleak" debugfs file together with
 *   modifications to the memory scanning parameters including the scan_thread
 *   pointer
55
 *
56
 * Locks and mutexes are acquired/nested in the following order:
57
 *
58 59 60 61
 *   scan_mutex [-> object->lock] -> kmemleak_lock -> other_object->lock (SINGLE_DEPTH_NESTING)
 *
 * No kmemleak_lock and object->lock nesting is allowed outside scan_mutex
 * regions.
62
 *
63 64 65 66 67 68 69 70
 * The kmemleak_object structures have a use_count incremented or decremented
 * using the get_object()/put_object() functions. When the use_count becomes
 * 0, this count can no longer be incremented and put_object() schedules the
 * kmemleak_object freeing via an RCU callback. All calls to the get_object()
 * function must be protected by rcu_read_lock() to avoid accessing a freed
 * structure.
 */

Joe Perches's avatar
Joe Perches committed
71 72
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

73 74 75 76 77 78
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/sched.h>
#include <linux/jiffies.h>
#include <linux/delay.h>
79
#include <linux/export.h>
80
#include <linux/kthread.h>
81
#include <linux/rbtree.h>
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
#include <linux/fs.h>
#include <linux/debugfs.h>
#include <linux/seq_file.h>
#include <linux/cpumask.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
#include <linux/rcupdate.h>
#include <linux/stacktrace.h>
#include <linux/cache.h>
#include <linux/percpu.h>
#include <linux/hardirq.h>
#include <linux/mmzone.h>
#include <linux/slab.h>
#include <linux/thread_info.h>
#include <linux/err.h>
#include <linux/uaccess.h>
#include <linux/string.h>
#include <linux/nodemask.h>
#include <linux/mm.h>
101
#include <linux/workqueue.h>
102
#include <linux/crc32.h>
103 104 105

#include <asm/sections.h>
#include <asm/processor.h>
106
#include <linux/atomic.h>
107

108
#include <linux/kasan.h>
109
#include <linux/kmemcheck.h>
110
#include <linux/kmemleak.h>
111
#include <linux/memory_hotplug.h>
112 113 114 115 116 117 118 119

/*
 * Kmemleak configuration and common defines.
 */
#define MAX_TRACE		16	/* stack trace length */
#define MSECS_MIN_AGE		5000	/* minimum object age for reporting */
#define SECS_FIRST_SCAN		60	/* delay before the first scan */
#define SECS_SCAN_WAIT		600	/* subsequent auto scanning delay */
120
#define MAX_SCAN_SIZE		4096	/* maximum size of a scanned block */
121 122 123

#define BYTES_PER_POINTER	sizeof(void *)

124
/* GFP bitmask for kmemleak internal allocations */
125 126
#define gfp_kmemleak_mask(gfp)	(((gfp) & (GFP_KERNEL | GFP_ATOMIC | \
					   __GFP_NOACCOUNT)) | \
127 128
				 __GFP_NORETRY | __GFP_NOMEMALLOC | \
				 __GFP_NOWARN)
129

130 131 132
/* scanning area inside a memory block */
struct kmemleak_scan_area {
	struct hlist_node node;
133 134
	unsigned long start;
	size_t size;
135 136
};

137 138 139
#define KMEMLEAK_GREY	0
#define KMEMLEAK_BLACK	-1

140 141 142 143
/*
 * Structure holding the metadata for each allocated memory block.
 * Modifications to such objects should be made while holding the
 * object->lock. Insertions or deletions from object_list, gray_list or
144
 * rb_node are already protected by the corresponding locks or mutex (see
145 146 147 148 149 150 151 152
 * the notes on locking above). These objects are reference-counted
 * (use_count) and freed using the RCU mechanism.
 */
struct kmemleak_object {
	spinlock_t lock;
	unsigned long flags;		/* object status flags */
	struct list_head object_list;
	struct list_head gray_list;
153
	struct rb_node rb_node;
154 155 156 157 158 159 160 161 162
	struct rcu_head rcu;		/* object_list lockless traversal */
	/* object usage count; object freed when use_count == 0 */
	atomic_t use_count;
	unsigned long pointer;
	size_t size;
	/* minimum number of a pointers found before it is considered leak */
	int min_count;
	/* the total number of pointers found pointing to this object */
	int count;
163 164
	/* checksum for detecting modified objects */
	u32 checksum;
165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
	/* memory ranges to be scanned inside an object (empty for all) */
	struct hlist_head area_list;
	unsigned long trace[MAX_TRACE];
	unsigned int trace_len;
	unsigned long jiffies;		/* creation timestamp */
	pid_t pid;			/* pid of the current task */
	char comm[TASK_COMM_LEN];	/* executable name */
};

/* flag representing the memory block allocation status */
#define OBJECT_ALLOCATED	(1 << 0)
/* flag set after the first reporting of an unreference object */
#define OBJECT_REPORTED		(1 << 1)
/* flag set to not scan the object */
#define OBJECT_NO_SCAN		(1 << 2)

181 182 183 184 185 186 187 188 189
/* number of bytes to print per line; must be 16 or 32 */
#define HEX_ROW_SIZE		16
/* number of bytes to print at a time (1, 2, 4, 8) */
#define HEX_GROUP_SIZE		1
/* include ASCII after the hex output */
#define HEX_ASCII		1
/* max number of lines to be printed */
#define HEX_MAX_LINES		2

190 191 192 193
/* the list of all allocated objects */
static LIST_HEAD(object_list);
/* the list of gray-colored objects (see color_gray comment below) */
static LIST_HEAD(gray_list);
194 195 196
/* search tree for object boundaries */
static struct rb_root object_tree_root = RB_ROOT;
/* rw_lock protecting the access to object_list and object_tree_root */
197 198 199 200 201 202 203
static DEFINE_RWLOCK(kmemleak_lock);

/* allocation caches for kmemleak internal data */
static struct kmem_cache *object_cache;
static struct kmem_cache *scan_area_cache;

/* set if tracing memory operations is enabled */
204
static int kmemleak_enabled;
205 206
/* same as above but only for the kmemleak_free() callback */
static int kmemleak_free_enabled;
207
/* set in the late_initcall if there were no errors */
208
static int kmemleak_initialized;
209
/* enables or disables early logging of the memory operations */
210
static int kmemleak_early_log = 1;
211
/* set if a kmemleak warning was issued */
212
static int kmemleak_warning;
213
/* set if a fatal kmemleak error has occurred */
214
static int kmemleak_error;
215 216 217 218 219 220

/* minimum and maximum address that may be valid pointers */
static unsigned long min_addr = ULONG_MAX;
static unsigned long max_addr;

static struct task_struct *scan_thread;
221
/* used to avoid reporting of recently allocated objects */
222
static unsigned long jiffies_min_age;
223
static unsigned long jiffies_last_scan;
224 225 226
/* delay between automatic memory scannings */
static signed long jiffies_scan_wait;
/* enables or disables the task stacks scanning */
227
static int kmemleak_stack_scan = 1;
228
/* protects the memory scanning, parameters and debug/kmemleak file access */
229
static DEFINE_MUTEX(scan_mutex);
230 231
/* setting kmemleak=on, will set this var, skipping the disable */
static int kmemleak_skip_disable;
232 233
/* If there are leaks that can be reported */
static bool kmemleak_found_leaks;
234 235

/*
236
 * Early object allocation/freeing logging. Kmemleak is initialized after the
237
 * kernel allocator. However, both the kernel allocator and kmemleak may
238
 * allocate memory blocks which need to be tracked. Kmemleak defines an
239 240 241 242 243 244 245
 * arbitrary buffer to hold the allocation/freeing information before it is
 * fully initialized.
 */

/* kmemleak operation type for early logging */
enum {
	KMEMLEAK_ALLOC,
246
	KMEMLEAK_ALLOC_PERCPU,
247
	KMEMLEAK_FREE,
248
	KMEMLEAK_FREE_PART,
249
	KMEMLEAK_FREE_PERCPU,
250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
	KMEMLEAK_NOT_LEAK,
	KMEMLEAK_IGNORE,
	KMEMLEAK_SCAN_AREA,
	KMEMLEAK_NO_SCAN
};

/*
 * Structure holding the information passed to kmemleak callbacks during the
 * early logging.
 */
struct early_log {
	int op_type;			/* kmemleak operation type */
	const void *ptr;		/* allocated/freed memory block */
	size_t size;			/* memory block size */
	int min_count;			/* minimum reference count */
265 266
	unsigned long trace[MAX_TRACE];	/* stack trace */
	unsigned int trace_len;		/* stack trace length */
267 268 269
};

/* early logging buffer and current position */
270 271 272
static struct early_log
	early_log[CONFIG_DEBUG_KMEMLEAK_EARLY_LOG_SIZE] __initdata;
static int crt_early_log __initdata;
273 274 275 276 277 278

static void kmemleak_disable(void);

/*
 * Print a warning and dump the stack trace.
 */
279 280 281
#define kmemleak_warn(x...)	do {		\
	pr_warning(x);				\
	dump_stack();				\
282
	kmemleak_warning = 1;			\
283 284 285
} while (0)

/*
286
 * Macro invoked when a serious kmemleak condition occurred and cannot be
287
 * recovered from. Kmemleak will be disabled and further allocation/freeing
288 289
 * tracing no longer available.
 */
290
#define kmemleak_stop(x...)	do {	\
291 292 293 294
	kmemleak_warn(x);		\
	kmemleak_disable();		\
} while (0)

295 296 297 298 299 300 301 302 303 304
/*
 * Printing of the objects hex dump to the seq file. The number of lines to be
 * printed is limited to HEX_MAX_LINES to prevent seq file spamming. The
 * actual number of printed bytes depends on HEX_ROW_SIZE. It must be called
 * with the object->lock held.
 */
static void hex_dump_object(struct seq_file *seq,
			    struct kmemleak_object *object)
{
	const u8 *ptr = (const u8 *)object->pointer;
305
	size_t len;
306 307

	/* limit the number of lines to HEX_MAX_LINES */
308
	len = min_t(size_t, object->size, HEX_MAX_LINES * HEX_ROW_SIZE);
309

310 311 312
	seq_printf(seq, "  hex dump (first %zu bytes):\n", len);
	seq_hex_dump(seq, "    ", DUMP_PREFIX_NONE, HEX_ROW_SIZE,
		     HEX_GROUP_SIZE, ptr, len, HEX_ASCII);
313 314
}

315 316 317 318 319 320 321 322 323 324
/*
 * Object colors, encoded with count and min_count:
 * - white - orphan object, not enough references to it (count < min_count)
 * - gray  - not orphan, not marked as false positive (min_count == 0) or
 *		sufficient references to it (count >= min_count)
 * - black - ignore, it doesn't contain references (e.g. text section)
 *		(min_count == -1). No function defined for this color.
 * Newly created objects don't have any color assigned (object->count == -1)
 * before the next memory scan when they become white.
 */
325
static bool color_white(const struct kmemleak_object *object)
326
{
327 328
	return object->count != KMEMLEAK_BLACK &&
		object->count < object->min_count;
329 330
}

331
static bool color_gray(const struct kmemleak_object *object)
332
{
333 334
	return object->min_count != KMEMLEAK_BLACK &&
		object->count >= object->min_count;
335 336 337 338 339 340 341
}

/*
 * Objects are considered unreferenced only if their color is white, they have
 * not be deleted and have a minimum age to avoid false positives caused by
 * pointers temporarily stored in CPU registers.
 */
342
static bool unreferenced_object(struct kmemleak_object *object)
343
{
344
	return (color_white(object) && object->flags & OBJECT_ALLOCATED) &&
345 346
		time_before_eq(object->jiffies + jiffies_min_age,
			       jiffies_last_scan);
347 348 349
}

/*
350 351
 * Printing of the unreferenced objects information to the seq file. The
 * print_unreferenced function must be called with the object->lock held.
352 353 354 355 356
 */
static void print_unreferenced(struct seq_file *seq,
			       struct kmemleak_object *object)
{
	int i;
357
	unsigned int msecs_age = jiffies_to_msecs(jiffies - object->jiffies);
358

359 360
	seq_printf(seq, "unreferenced object 0x%08lx (size %zu):\n",
		   object->pointer, object->size);
361 362 363
	seq_printf(seq, "  comm \"%s\", pid %d, jiffies %lu (age %d.%03ds)\n",
		   object->comm, object->pid, object->jiffies,
		   msecs_age / 1000, msecs_age % 1000);
364
	hex_dump_object(seq, object);
365
	seq_printf(seq, "  backtrace:\n");
366 367 368

	for (i = 0; i < object->trace_len; i++) {
		void *ptr = (void *)object->trace[i];
369
		seq_printf(seq, "    [<%p>] %pS\n", ptr, ptr);
370 371 372 373 374 375 376 377 378 379 380 381 382 383 384
	}
}

/*
 * Print the kmemleak_object information. This function is used mainly for
 * debugging special cases when kmemleak operations. It must be called with
 * the object->lock held.
 */
static void dump_object_info(struct kmemleak_object *object)
{
	struct stack_trace trace;

	trace.nr_entries = object->trace_len;
	trace.entries = object->trace;

Joe Perches's avatar
Joe Perches committed
385
	pr_notice("Object 0x%08lx (size %zu):\n",
386
		  object->pointer, object->size);
387 388 389 390
	pr_notice("  comm \"%s\", pid %d, jiffies %lu\n",
		  object->comm, object->pid, object->jiffies);
	pr_notice("  min_count = %d\n", object->min_count);
	pr_notice("  count = %d\n", object->count);
391
	pr_notice("  flags = 0x%lx\n", object->flags);
392
	pr_notice("  checksum = %u\n", object->checksum);
393 394 395 396 397
	pr_notice("  backtrace:\n");
	print_stack_trace(&trace, 4);
}

/*
398
 * Look-up a memory block metadata (kmemleak_object) in the object search
399 400 401 402 403 404
 * tree based on a pointer value. If alias is 0, only values pointing to the
 * beginning of the memory block are allowed. The kmemleak_lock must be held
 * when calling this function.
 */
static struct kmemleak_object *lookup_object(unsigned long ptr, int alias)
{
405 406 407 408 409 410 411 412 413 414 415 416
	struct rb_node *rb = object_tree_root.rb_node;

	while (rb) {
		struct kmemleak_object *object =
			rb_entry(rb, struct kmemleak_object, rb_node);
		if (ptr < object->pointer)
			rb = object->rb_node.rb_left;
		else if (object->pointer + object->size <= ptr)
			rb = object->rb_node.rb_right;
		else if (object->pointer == ptr || alias)
			return object;
		else {
417 418
			kmemleak_warn("Found object by alias at 0x%08lx\n",
				      ptr);
419
			dump_object_info(object);
420
			break;
421
		}
422 423
	}
	return NULL;
424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
}

/*
 * Increment the object use_count. Return 1 if successful or 0 otherwise. Note
 * that once an object's use_count reached 0, the RCU freeing was already
 * registered and the object should no longer be used. This function must be
 * called under the protection of rcu_read_lock().
 */
static int get_object(struct kmemleak_object *object)
{
	return atomic_inc_not_zero(&object->use_count);
}

/*
 * RCU callback to free a kmemleak_object.
 */
static void free_object_rcu(struct rcu_head *rcu)
{
442
	struct hlist_node *tmp;
443 444 445 446 447 448 449 450
	struct kmemleak_scan_area *area;
	struct kmemleak_object *object =
		container_of(rcu, struct kmemleak_object, rcu);

	/*
	 * Once use_count is 0 (guaranteed by put_object), there is no other
	 * code accessing this object, hence no need for locking.
	 */
451 452
	hlist_for_each_entry_safe(area, tmp, &object->area_list, node) {
		hlist_del(&area->node);
453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476
		kmem_cache_free(scan_area_cache, area);
	}
	kmem_cache_free(object_cache, object);
}

/*
 * Decrement the object use_count. Once the count is 0, free the object using
 * an RCU callback. Since put_object() may be called via the kmemleak_free() ->
 * delete_object() path, the delayed RCU freeing ensures that there is no
 * recursive call to the kernel allocator. Lock-less RCU object_list traversal
 * is also possible.
 */
static void put_object(struct kmemleak_object *object)
{
	if (!atomic_dec_and_test(&object->use_count))
		return;

	/* should only get here after delete_object was called */
	WARN_ON(object->flags & OBJECT_ALLOCATED);

	call_rcu(&object->rcu, free_object_rcu);
}

/*
477
 * Look up an object in the object search tree and increase its use_count.
478 479 480 481
 */
static struct kmemleak_object *find_and_get_object(unsigned long ptr, int alias)
{
	unsigned long flags;
482
	struct kmemleak_object *object;
483 484 485

	rcu_read_lock();
	read_lock_irqsave(&kmemleak_lock, flags);
486
	object = lookup_object(ptr, alias);
487 488 489 490 491 492 493 494 495 496
	read_unlock_irqrestore(&kmemleak_lock, flags);

	/* check whether the object is still available */
	if (object && !get_object(object))
		object = NULL;
	rcu_read_unlock();

	return object;
}

497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
/*
 * Look up an object in the object search tree and remove it from both
 * object_tree_root and object_list. The returned object's use_count should be
 * at least 1, as initially set by create_object().
 */
static struct kmemleak_object *find_and_remove_object(unsigned long ptr, int alias)
{
	unsigned long flags;
	struct kmemleak_object *object;

	write_lock_irqsave(&kmemleak_lock, flags);
	object = lookup_object(ptr, alias);
	if (object) {
		rb_erase(&object->rb_node, &object_tree_root);
		list_del_rcu(&object->object_list);
	}
	write_unlock_irqrestore(&kmemleak_lock, flags);

	return object;
}

518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533
/*
 * Save stack trace to the given array of MAX_TRACE size.
 */
static int __save_stack_trace(unsigned long *trace)
{
	struct stack_trace stack_trace;

	stack_trace.max_entries = MAX_TRACE;
	stack_trace.nr_entries = 0;
	stack_trace.entries = trace;
	stack_trace.skip = 2;
	save_stack_trace(&stack_trace);

	return stack_trace.nr_entries;
}

534 535 536 537
/*
 * Create the metadata (struct kmemleak_object) corresponding to an allocated
 * memory block and add it to the object_list and object_tree_root.
 */
538 539
static struct kmemleak_object *create_object(unsigned long ptr, size_t size,
					     int min_count, gfp_t gfp)
540 541
{
	unsigned long flags;
542 543
	struct kmemleak_object *object, *parent;
	struct rb_node **link, *rb_parent;
544

545
	object = kmem_cache_alloc(object_cache, gfp_kmemleak_mask(gfp));
546
	if (!object) {
547 548
		pr_warning("Cannot allocate a kmemleak_object structure\n");
		kmemleak_disable();
549
		return NULL;
550 551 552 553 554 555 556
	}

	INIT_LIST_HEAD(&object->object_list);
	INIT_LIST_HEAD(&object->gray_list);
	INIT_HLIST_HEAD(&object->area_list);
	spin_lock_init(&object->lock);
	atomic_set(&object->use_count, 1);
557
	object->flags = OBJECT_ALLOCATED;
558 559 560
	object->pointer = ptr;
	object->size = size;
	object->min_count = min_count;
561
	object->count = 0;			/* white color initially */
562
	object->jiffies = jiffies;
563
	object->checksum = 0;
564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583

	/* task information */
	if (in_irq()) {
		object->pid = 0;
		strncpy(object->comm, "hardirq", sizeof(object->comm));
	} else if (in_softirq()) {
		object->pid = 0;
		strncpy(object->comm, "softirq", sizeof(object->comm));
	} else {
		object->pid = current->pid;
		/*
		 * There is a small chance of a race with set_task_comm(),
		 * however using get_task_comm() here may cause locking
		 * dependency issues with current->alloc_lock. In the worst
		 * case, the command line is not correct.
		 */
		strncpy(object->comm, current->comm, sizeof(object->comm));
	}

	/* kernel backtrace */
584
	object->trace_len = __save_stack_trace(object->trace);
585 586

	write_lock_irqsave(&kmemleak_lock, flags);
587

588 589
	min_addr = min(min_addr, ptr);
	max_addr = max(max_addr, ptr + size);
590 591 592 593 594 595 596 597 598 599 600 601 602
	link = &object_tree_root.rb_node;
	rb_parent = NULL;
	while (*link) {
		rb_parent = *link;
		parent = rb_entry(rb_parent, struct kmemleak_object, rb_node);
		if (ptr + size <= parent->pointer)
			link = &parent->rb_node.rb_left;
		else if (parent->pointer + parent->size <= ptr)
			link = &parent->rb_node.rb_right;
		else {
			kmemleak_stop("Cannot insert 0x%lx into the object "
				      "search tree (overlaps existing)\n",
				      ptr);
603 604 605 606 607
			/*
			 * No need for parent->lock here since "parent" cannot
			 * be freed while the kmemleak_lock is held.
			 */
			dump_object_info(parent);
608
			kmem_cache_free(object_cache, object);
609
			object = NULL;
610 611
			goto out;
		}
612
	}
613 614 615
	rb_link_node(&object->rb_node, rb_parent, link);
	rb_insert_color(&object->rb_node, &object_tree_root);

616 617 618
	list_add_tail_rcu(&object->object_list, &object_list);
out:
	write_unlock_irqrestore(&kmemleak_lock, flags);
619
	return object;
620 621 622
}

/*
623
 * Mark the object as not allocated and schedule RCU freeing via put_object().
624
 */
625
static void __delete_object(struct kmemleak_object *object)
626 627 628 629
{
	unsigned long flags;

	WARN_ON(!(object->flags & OBJECT_ALLOCATED));
630
	WARN_ON(atomic_read(&object->use_count) < 1);
631 632 633 634 635 636 637 638 639 640 641

	/*
	 * Locking here also ensures that the corresponding memory block
	 * cannot be freed when it is being scanned.
	 */
	spin_lock_irqsave(&object->lock, flags);
	object->flags &= ~OBJECT_ALLOCATED;
	spin_unlock_irqrestore(&object->lock, flags);
	put_object(object);
}

642 643 644 645 646 647 648 649
/*
 * Look up the metadata (struct kmemleak_object) corresponding to ptr and
 * delete it.
 */
static void delete_object_full(unsigned long ptr)
{
	struct kmemleak_object *object;

650
	object = find_and_remove_object(ptr, 0);
651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670
	if (!object) {
#ifdef DEBUG
		kmemleak_warn("Freeing unknown object at 0x%08lx\n",
			      ptr);
#endif
		return;
	}
	__delete_object(object);
}

/*
 * Look up the metadata (struct kmemleak_object) corresponding to ptr and
 * delete it. If the memory block is partially freed, the function may create
 * additional metadata for the remaining parts of the block.
 */
static void delete_object_part(unsigned long ptr, size_t size)
{
	struct kmemleak_object *object;
	unsigned long start, end;

671
	object = find_and_remove_object(ptr, 1);
672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695
	if (!object) {
#ifdef DEBUG
		kmemleak_warn("Partially freeing unknown object at 0x%08lx "
			      "(size %zu)\n", ptr, size);
#endif
		return;
	}

	/*
	 * Create one or two objects that may result from the memory block
	 * split. Note that partial freeing is only done by free_bootmem() and
	 * this happens before kmemleak_init() is called. The path below is
	 * only executed during early log recording in kmemleak_init(), so
	 * GFP_KERNEL is enough.
	 */
	start = object->pointer;
	end = object->pointer + object->size;
	if (ptr > start)
		create_object(start, ptr - start, object->min_count,
			      GFP_KERNEL);
	if (ptr + size < end)
		create_object(ptr + size, end - ptr - size, object->min_count,
			      GFP_KERNEL);

696
	__delete_object(object);
697
}
698 699 700 701 702 703 704 705 706

static void __paint_it(struct kmemleak_object *object, int color)
{
	object->min_count = color;
	if (color == KMEMLEAK_BLACK)
		object->flags |= OBJECT_NO_SCAN;
}

static void paint_it(struct kmemleak_object *object, int color)
707 708
{
	unsigned long flags;
709 710 711 712 713 714 715 716

	spin_lock_irqsave(&object->lock, flags);
	__paint_it(object, color);
	spin_unlock_irqrestore(&object->lock, flags);
}

static void paint_ptr(unsigned long ptr, int color)
{
717 718 719 720
	struct kmemleak_object *object;

	object = find_and_get_object(ptr, 0);
	if (!object) {
721 722 723 724
		kmemleak_warn("Trying to color unknown object "
			      "at 0x%08lx as %s\n", ptr,
			      (color == KMEMLEAK_GREY) ? "Grey" :
			      (color == KMEMLEAK_BLACK) ? "Black" : "Unknown");
725 726
		return;
	}
727
	paint_it(object, color);
728 729 730
	put_object(object);
}

731
/*
732
 * Mark an object permanently as gray-colored so that it can no longer be
733 734 735 736 737 738 739
 * reported as a leak. This is used in general to mark a false positive.
 */
static void make_gray_object(unsigned long ptr)
{
	paint_ptr(ptr, KMEMLEAK_GREY);
}

740 741 742 743 744 745
/*
 * Mark the object as black-colored so that it is ignored from scans and
 * reporting.
 */
static void make_black_object(unsigned long ptr)
{
746
	paint_ptr(ptr, KMEMLEAK_BLACK);
747 748 749 750 751 752
}

/*
 * Add a scanning area to the object. If at least one such area is added,
 * kmemleak will only scan these ranges rather than the whole memory block.
 */
753
static void add_scan_area(unsigned long ptr, size_t size, gfp_t gfp)
754 755 756 757 758
{
	unsigned long flags;
	struct kmemleak_object *object;
	struct kmemleak_scan_area *area;

759
	object = find_and_get_object(ptr, 1);
760
	if (!object) {
Joe Perches's avatar
Joe Perches committed
761 762
		kmemleak_warn("Adding scan area to unknown object at 0x%08lx\n",
			      ptr);
763 764 765
		return;
	}

766
	area = kmem_cache_alloc(scan_area_cache, gfp_kmemleak_mask(gfp));
767
	if (!area) {
768
		pr_warning("Cannot allocate a scan area\n");
769 770 771 772
		goto out;
	}

	spin_lock_irqsave(&object->lock, flags);
773 774 775
	if (size == SIZE_MAX) {
		size = object->pointer + object->size - ptr;
	} else if (ptr + size > object->pointer + object->size) {
Joe Perches's avatar
Joe Perches committed
776
		kmemleak_warn("Scan area larger than object 0x%08lx\n", ptr);
777 778 779 780 781 782
		dump_object_info(object);
		kmem_cache_free(scan_area_cache, area);
		goto out_unlock;
	}

	INIT_HLIST_NODE(&area->node);
783 784
	area->start = ptr;
	area->size = size;
785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804

	hlist_add_head(&area->node, &object->area_list);
out_unlock:
	spin_unlock_irqrestore(&object->lock, flags);
out:
	put_object(object);
}

/*
 * Set the OBJECT_NO_SCAN flag for the object corresponding to the give
 * pointer. Such object will not be scanned by kmemleak but references to it
 * are searched.
 */
static void object_no_scan(unsigned long ptr)
{
	unsigned long flags;
	struct kmemleak_object *object;

	object = find_and_get_object(ptr, 0);
	if (!object) {
Joe Perches's avatar
Joe Perches committed
805
		kmemleak_warn("Not scanning unknown object at 0x%08lx\n", ptr);
806 807 808 809 810 811 812 813 814 815 816 817 818
		return;
	}

	spin_lock_irqsave(&object->lock, flags);
	object->flags |= OBJECT_NO_SCAN;
	spin_unlock_irqrestore(&object->lock, flags);
	put_object(object);
}

/*
 * Log an early kmemleak_* call to the early_log buffer. These calls will be
 * processed later once kmemleak is fully initialized.
 */
819
static void __init log_early(int op_type, const void *ptr, size_t size,
820
			     int min_count)
821 822 823 824
{
	unsigned long flags;
	struct early_log *log;

825
	if (kmemleak_error) {
826 827 828 829 830
		/* kmemleak stopped recording, just count the requests */
		crt_early_log++;
		return;
	}

831
	if (crt_early_log >= ARRAY_SIZE(early_log)) {
832
		crt_early_log++;
833
		kmemleak_disable();
834 835 836 837 838 839 840 841 842 843 844 845 846
		return;
	}

	/*
	 * There is no need for locking since the kernel is still in UP mode
	 * at this stage. Disabling the IRQs is enough.
	 */
	local_irq_save(flags);
	log = &early_log[crt_early_log];
	log->op_type = op_type;
	log->ptr = ptr;
	log->size = size;
	log->min_count = min_count;
847
	log->trace_len = __save_stack_trace(log->trace);
848 849 850 851
	crt_early_log++;
	local_irq_restore(flags);
}

852 853 854 855 856 857 858 859 860
/*
 * Log an early allocated block and populate the stack trace.
 */
static void early_alloc(struct early_log *log)
{
	struct kmemleak_object *object;
	unsigned long flags;
	int i;

861
	if (!kmemleak_enabled || !log->ptr || IS_ERR(log->ptr))
862 863 864 865 866 867 868
		return;

	/*
	 * RCU locking needed to ensure object is not freed via put_object().
	 */
	rcu_read_lock();
	object = create_object((unsigned long)log->ptr, log->size,
869
			       log->min_count, GFP_ATOMIC);
870 871
	if (!object)
		goto out;
872 873 874 875 876
	spin_lock_irqsave(&object->lock, flags);
	for (i = 0; i < log->trace_len; i++)
		object->trace[i] = log->trace[i];
	object->trace_len = log->trace_len;
	spin_unlock_irqrestore(&object->lock, flags);
877
out:
878 879 880
	rcu_read_unlock();
}

881 882 883 884 885 886 887 888 889 890 891 892 893 894
/*
 * Log an early allocated block and populate the stack trace.
 */
static void early_alloc_percpu(struct early_log *log)
{
	unsigned int cpu;
	const void __percpu *ptr = log->ptr;

	for_each_possible_cpu(cpu) {
		log->ptr = per_cpu_ptr(ptr, cpu);
		early_alloc(log);
	}
}

895 896 897 898 899 900 901 902 903 904 905 906 907
/**
 * kmemleak_alloc - register a newly allocated object
 * @ptr:	pointer to beginning of the object
 * @size:	size of the object
 * @min_count:	minimum number of references to this object. If during memory
 *		scanning a number of references less than @min_count is found,
 *		the object is reported as a memory leak. If @min_count is 0,
 *		the object is never reported as a leak. If @min_count is -1,
 *		the object is ignored (not scanned and not reported as a leak)
 * @gfp:	kmalloc() flags used for kmemleak internal memory allocations
 *
 * This function is called from the kernel allocators when a new object
 * (memory block) is allocated (kmem_cache_alloc, kmalloc, vmalloc etc.).
908
 */
909 910
void __ref kmemleak_alloc(const void *ptr, size_t size, int min_count,
			  gfp_t gfp)
911 912 913
{
	pr_debug("%s(0x%p, %zu, %d)\n", __func__, ptr, size, min_count);

914
	if (kmemleak_enabled && ptr && !IS_ERR(ptr))
915
		create_object((unsigned long)ptr, size, min_count, gfp);
916
	else if (kmemleak_early_log)
917
		log_early(KMEMLEAK_ALLOC, ptr, size, min_count);
918 919 920
}
EXPORT_SYMBOL_GPL(kmemleak_alloc);

921 922 923 924
/**
 * kmemleak_alloc_percpu - register a newly allocated __percpu object
 * @ptr:	__percpu pointer to beginning of the object
 * @size:	size of the object
925
 * @gfp:	flags used for kmemleak internal memory allocations
926 927
 *
 * This function is called from the kernel percpu allocator when a new object
928
 * (memory block) is allocated (alloc_percpu).
929
 */
930 931
void __ref kmemleak_alloc_percpu(const void __percpu *ptr, size_t size,
				 gfp_t gfp)
932 933 934 935 936 937 938 939 940
{
	unsigned int cpu;

	pr_debug("%s(0x%p, %zu)\n", __func__, ptr, size);

	/*
	 * Percpu allocations are only scanned and not reported as leaks
	 * (min_count is set to 0).
	 */
941
	if (kmemleak_enabled && ptr && !IS_ERR(ptr))
942 943
		for_each_possible_cpu(cpu)
			create_object((unsigned long)per_cpu_ptr(ptr, cpu),
944
				      size, 0, gfp);
945
	else if (kmemleak_early_log)
946 947 948 949
		log_early(KMEMLEAK_ALLOC_PERCPU, ptr, size, 0);
}
EXPORT_SYMBOL_GPL(kmemleak_alloc_percpu);

950 951 952 953 954 955
/**
 * kmemleak_free - unregister a previously registered object
 * @ptr:	pointer to beginning of the object
 *
 * This function is called from the kernel allocators when an object (memory
 * block) is freed (kmem_cache_free, kfree, vfree etc.).
956
 */
957
void __ref kmemleak_free(const void *ptr)
958 959 960
{
	pr_debug("%s(0x%p)\n", __func__, ptr);

961
	if (kmemleak_free_enabled && ptr && !IS_ERR(ptr))
962
		delete_object_full((unsigned long)ptr);
963
	else if (kmemleak_early_log)
964
		log_early(KMEMLEAK_FREE, ptr, 0, 0);
965 966 967
}
EXPORT_SYMBOL_GPL(kmemleak_free);

968 969 970 971 972 973 974 975
/**
 * kmemleak_free_part - partially unregister a previously registered object
 * @ptr:	pointer to the beginning or inside the object. This also
 *		represents the start of the range to be freed
 * @size:	size to be unregistered
 *
 * This function is called when only a part of a memory block is freed
 * (usually from the bootmem allocator).
976
 */
977
void __ref kmemleak_free_part(const void *ptr, size_t size)
978 979 980
{
	pr_debug("%s(0x%p)\n", __func__, ptr);

981
	if (kmemleak_enabled && ptr && !IS_ERR(ptr))
982
		delete_object_part((unsigned long)ptr, size);
983
	else if (kmemleak_early_log)
984
		log_early(KMEMLEAK_FREE_PART, ptr, size, 0);
985 986 987
}
EXPORT_SYMBOL_GPL(kmemleak_free_part);

988 989 990 991 992 993 994 995 996 997 998 999 1000
/**
 * kmemleak_free_percpu - unregister a previously registered __percpu object
 * @ptr:	__percpu pointer to beginning of the object
 *
 * This function is called from the kernel percpu allocator when an object
 * (memory block) is freed (free_percpu).
 */
void __ref kmemleak_free_percpu(const void __percpu *ptr)
{
	unsigned int cpu;

	pr_debug("%s(0x%p)\n", __func__, ptr);

1001
	if (kmemleak_free_enabled && ptr && !IS_ERR(ptr))
1002 1003 1004
		for_each_possible_cpu(cpu)
			delete_object_full((unsigned long)per_cpu_ptr(ptr,
								      cpu));
1005
	else if (kmemleak_early_log)
1006 1007 1008 1009
		log_early(KMEMLEAK_FREE_PERCPU, ptr, 0, 0);
}
EXPORT_SYMBOL_GPL(kmemleak_free_percpu);

1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
/**
 * kmemleak_update_trace - update object allocation stack trace
 * @ptr:	pointer to beginning of the object
 *
 * Override the object allocation stack trace for cases where the actual
 * allocation place is not always useful.
 */
void __ref kmemleak_update_trace(const void *ptr)
{
	struct kmemleak_object *object;
	unsigned long flags;

	pr_debug("%s(0x%p)\n", __func__, ptr);

	if (!kmemleak_enabled || IS_ERR_OR_NULL(ptr))
		return;

	object = find_and_get_object((unsigned long)ptr, 1);
	if (!object) {
#ifdef DEBUG
		kmemleak_warn("Updating stack trace for unknown object at %p\n",
			      ptr);
#endif
		return;
	}

	spin_lock_irqsave(&object->lock, flags);
	object->trace_len = __save_stack_trace(object->trace);
	spin_unlock_irqrestore(&object->lock, flags);

	put_object(object);
}
EXPORT_SYMBOL(kmemleak_update_trace);

1044 1045 1046 1047 1048 1049
/**
 * kmemleak_not_leak - mark an allocated object as false positive
 * @ptr:	pointer to beginning of the object
 *
 * Calling this function on an object will cause the memory block to no longer
 * be reported as leak and always be scanned.
1050
 */
1051
void __ref kmemleak_not_leak(const void *ptr)
1052 1053 1054
{
	pr_debug("%s(0x%p)\n", __func__, ptr);

1055
	if (kmemleak_enabled && ptr && !IS_ERR(ptr))
1056
		make_gray_object((unsigned long)ptr);
1057
	else if (kmemleak_early_log)
1058
		log_early(KMEMLEAK_NOT_LEAK, ptr, 0, 0);
1059 1060 1061
}
EXPORT_SYMBOL(kmemleak_not_leak);

1062 1063 1064 1065 1066 1067 1068 1069
/**
 * kmemleak_ignore - ignore an allocated object
 * @ptr:	pointer to beginning of the object
 *
 * Calling this function on an object will cause the memory block to be
 * ignored (not scanned and not reported as a leak). This is usually done when
 * it is known that the corresponding block is not a leak and does not contain
 * any references to other allocated memory blocks.
1070
 */
1071
void __ref kmemleak_ignore(const void *ptr)
1072 1073 1074
{
	pr_debug("%s(0x%p)\n", __func__, ptr);

1075
	if (kmemleak_enabled && ptr && !IS_ERR(ptr))
1076
		make_black_object((unsigned long)ptr);
1077
	else if (kmemleak_early_log)
1078
		log_early(KMEMLEAK_IGNORE, ptr, 0, 0);
1079 1080 1081
}
EXPORT_SYMBOL(kmemleak_ignore);

1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
/**
 * kmemleak_scan_area - limit the range to be scanned in an allocated object
 * @ptr:	pointer to beginning or inside the object. This also
 *		represents the start of the scan area
 * @size:	size of the scan area
 * @gfp:	kmalloc() flags used for kmemleak internal memory allocations
 *
 * This function is used when it is known that only certain parts of an object
 * contain references to other objects. Kmemleak will only scan these areas
 * reducing the number false negatives.
1092
 */
1093
void __ref kmemleak_scan_area(const void *ptr, size_t size, gfp_t gfp)
1094 1095 1096
{
	pr_debug("%s(0x%p)\n", __func__, ptr);

1097
	if (kmemleak_enabled && ptr && size && !IS_ERR(ptr))
1098
		add_scan_area((unsigned long)ptr, size, gfp);
1099
	else if (kmemleak_early_log)
1100
		log_early(KMEMLEAK_SCAN_AREA, ptr, size, 0);
1101 1102 1103
}
EXPORT_SYMBOL(kmemleak_scan_area);

1104 1105 1106 1107 1108 1109 1110 1111
/**
 * kmemleak_no_scan - do not scan an allocated object
 * @ptr:	pointer to beginning of the object
 *
 * This function notifies kmemleak not to scan the given memory block. Useful
 * in situations where it is known that the given object does not contain any
 * references to other objects. Kmemleak will not scan such objects reducing
 * the number of false negatives.
1112
 */
1113
void __ref kmemleak_no_scan(const void *ptr)
1114 1115 1116
{
	pr_debug("%s(0x%p)\n", __func__, ptr);

1117
	if (kmemleak_enabled && ptr && !IS_ERR(ptr))
1118
		object_no_scan((unsigned long)ptr);
1119
	else if (kmemleak_early_log)
1120
		log_early(KMEMLEAK_NO_SCAN, ptr, 0, 0);
1121 1122 1123
}
EXPORT_SYMBOL(kmemleak_no_scan);

1124 1125 1126 1127 1128 1129 1130 1131 1132 1133
/*
 * Update an object's checksum and return true if it was modified.
 */
static bool update_checksum(struct kmemleak_object *object)
{
	u32 old_csum = object->checksum;

	if (!kmemcheck_is_obj_initialized(object->pointer, object->size))
		return false;

1134
	kasan_disable_current();
1135
	object->checksum = crc32(0, (void *)object->pointer, object->size);
1136 1137
	kasan_enable_current();

1138 1139 1140
	return object->checksum != old_csum;
}

1141 1142
/*
 * Memory scanning is a long process and it needs to be interruptable. This
1143
 * function checks whether such interrupt condition occurred.
1144 1145 1146
 */
static int scan_should_stop(void)
{
1147
	if (!kmemleak_enabled)
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
		return 1;

	/*
	 * This function may be called from either process or kthread context,
	 * hence the need to check for both stop conditions.
	 */
	if (current->mm)
		return signal_pending(current);
	else
		return kthread_should_stop();

	return 0;
}

/*
 * Scan a memory block (exclusive range) for valid pointers and add those
 * found to the gray list.
 */
static void scan_block(void *_start, void *_end,
1167
		       struct kmemleak_object *scanned)
1168 1169 1170 1171
{
	unsigned long *ptr;
	unsigned long *start = PTR_ALIGN(_start, BYTES_PER_POINTER);
	unsigned long *end = _end - (BYTES_PER_POINTER - 1);
1172
	unsigned long flags;
1173

1174
	read_lock_irqsave(&kmemleak_lock, flags);
1175 1176
	for (ptr = start; ptr < end; ptr++) {
		struct kmemleak_object *object;
1177
		unsigned long pointer;
1178 1179 1180 1181

		if (scan_should_stop())
			break;

1182 1183 1184 1185 1186
		/* don't scan uninitialized memory */
		if (!kmemcheck_is_obj_initialized((unsigned long)ptr,
						  BYTES_PER_POINTER))
			continue;

1187
		kasan_disable_current();
1188
		pointer = *ptr;
1189
		kasan_enable_current();
1190

1191 1192 1193 1194 1195 1196 1197 1198 1199 1200
		if (pointer < min_addr || pointer >= max_addr)
			continue;

		/*
		 * No need for get_object() here since we hold kmemleak_lock.
		 * object->use_count cannot be dropped to 0 while the object
		 * is still present in object_tree_root and object_list
		 * (with updates protected by kmemleak_lock).
		 */
		object = lookup_object(pointer, 1);
1201 1202
		if (!object)
			continue;
1203
		if (object == scanned)
1204 1205 1206 1207 1208 1209 1210 1211
			/* self referenced, ignore */
			continue;

		/*
		 * Avoid the lockdep recursive warning on object->lock being
		 * previously acquired in scan_object(). These locks are
		 * enclosed by scan_mutex.
		 */
1212
		spin_lock_nested(&object->lock, SINGLE_DEPTH_NESTING);
1213 1214
		if (!color_white(object)) {
			/* non-orphan, ignored or new */
1215
			spin_unlock(&object->lock);
1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
			continue;
		}

		/*
		 * Increase the object's reference count (number of pointers
		 * to the memory block). If this count reaches the required
		 * minimum, the object's color will become gray and it will be
		 * added to the gray_list.
		 */
		object->count++;
1226
		if (color_gray(object)) {
1227 1228
			/* put_object() called when removing from gray_list */
			WARN_ON(!get_object(object));
1229
			list_add_tail(&object->gray_list, &gray_list);
1230
		}
1231 1232 1233 1234
		spin_unlock(&object->lock);
	}
	read_unlock_irqrestore(&kmemleak_lock, flags);
}
1235

1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247
/*
 * Scan a large memory block in MAX_SCAN_SIZE chunks to reduce the latency.
 */
static void scan_large_block(void *start, void *end)
{
	void *next;

	while (start < end) {
		next = min(start + MAX_SCAN_SIZE, end);
		scan_block(start, next, NULL);
		start = next;
		cond_resched();
1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260
	}
}

/*
 * Scan a memory block corresponding to a kmemleak_object. A condition is
 * that object->use_count >= 1.
 */
static void scan_object(struct kmemleak_object *object)
{
	struct kmemleak_scan_area *area;
	unsigned long flags;

	/*
1261 1262
	 * Once the object->lock is acquired, the corresponding memory block
	 * cannot be freed (the same lock is acquired in delete_object).
1263 1264 1265 1266 1267 1268 1269
	 */
	spin_lock_irqsave(&object->lock, flags);
	if (object->flags & OBJECT_NO_SCAN)
		goto out;
	if (!(object->flags & OBJECT_ALLOCATED))
		/* already freed object */
		goto out;
1270 1271 1272
	if (hlist_empty(&object->area_list)) {
		void *start = (void *)object->pointer;
		void *end = (void *)(object->pointer + object->size);
1273 1274 1275 1276 1277
		void *next;

		do {
			next = min(start + MAX_SCAN_SIZE, end);
			scan_block(start, next, object);
1278

1279 1280 1281
			start = next;
			if (start >= end)
				break;
1282 1283 1284 1285

			spin_unlock_irqrestore(&object->lock, flags);
			cond_resched();
			spin_lock_irqsave(&object->lock, flags);
1286
		} while (object->flags & OBJECT_ALLOCATED);
1287
	} else
1288
		hlist_for_each_entry(area, &object->area_list, node)
1289 1290
			scan_block((void *)area->start,
				   (void *)(area->start + area->size),
1291
				   object);
1292 1293 1294 1295
out:
	spin_unlock_irqrestore(&object->lock, flags);
}

1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
/*
 * Scan the objects already referenced (gray objects). More objects will be
 * referenced and, if there are no memory leaks, all the objects are scanned.
 */
static void scan_gray_list(void)
{
	struct kmemleak_object *object, *tmp;

	/*
	 * The list traversal is safe for both tail additions and removals
	 * from inside the loop. The kmemleak objects cannot be freed from
	 * outside the loop because their use_count was incremented.
	 */
	object = list_entry(gray_list.next, typeof(*object), gray_list);
	while (&object->gray_list != &gray_list) {
		cond_resched();

		/* may add new objects to the list */
		if (!scan_should_stop())
			scan_object(object);

		tmp = list_entry(object->gray_list.next, typeof(*object),
				 gray_list);

		/* remove the object from the list and release it */
		list_del(&object->gray_list);
		put_object(object);

		object = tmp;
	}
	WARN_ON(!list_empty(&gray_list));
}

1329 1330 1331 1332 1333 1334 1335 1336
/*
 * Scan data sections and all the referenced memory blocks allocated via the
 * kernel's standard allocators. This function must be called with the
 * scan_mutex held.
 */
static void kmemleak_scan(void)
{
	unsigned long flags;
1337
	struct kmemleak_object *object;
1338
	int i;
1339
	int new_leaks = 0;
1340

1341 1342
	jiffies_last_scan = jiffies;

1343 1344 1345 1346 1347 1348 1349 1350 1351 1352
	/* prepare the kmemleak_object's */
	rcu_read_lock();
	list_for_each_entry_rcu(object, &object_list, object_list) {
		spin_lock_irqsave(&object->lock, flags);
#ifdef DEBUG
		/*
		 * With a few exceptions there should be a maximum of
		 * 1 reference to any object at this point.
		 */
		if (atomic_read(&object->use_count) > 1) {
Joe Perches's avatar
Joe Perches committed
1353
			pr_debug("object->use_count = %d\n",
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367
				 atomic_read(&object->use_count));
			dump_object_info(object);
		}
#endif
		/* reset the reference count (whiten the object) */
		object->count = 0;
		if (color_gray(object) && get_object(object))
			list_add_tail(&object->gray_list, &gray_list);

		spin_unlock_irqrestore(&object->lock, flags);
	}
	rcu_read_unlock();

	/* data/bss scanning */
1368 1369
	scan_large_block(_sdata, _edata);
	scan_large_block(__bss_start, __bss_stop);
1370 1371 1372 1373

#ifdef CONFIG_SMP
	/* per-cpu sections scanning */
	for_each_possible_cpu(i)
1374 1375
		scan_large_block(__per_cpu_start + per_cpu_offset(i),
				 __per_cpu_end + per_cpu_offset(i));
1376 1377 1378
#endif

	/*
1379
	 * Struct page scanning for each node.
1380
	 */
1381
	get_online_mems();
1382
	for_each_online_node(i) {
1383 1384
		unsigned long start_pfn = node_start_pfn(i);
		unsigned long end_pfn = node_end_pfn(i);
1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395
		unsigned long pfn;

		for (pfn = start_pfn; pfn < end_pfn; pfn++) {
			struct page *page;

			if (!pfn_valid(pfn))
				continue;
			page = pfn_to_page(pfn);
			/* only scan if page is in use */
			if (page_count(page) == 0)
				continue;
1396
			scan_block(page, page + 1, NULL);
1397 1398
		}
	}
1399
	put_online_mems();
1400 1401

	/*
1402
	 * Scanning the task stacks (may introduce false negatives).
1403 1404
	 */
	if (kmemleak_stack_scan) {
1405 1406
		struct task_struct *p, *g;

1407
		read_lock(&tasklist_lock);
1408 1409
		do_each_thread(g, p) {
			scan_block(task_stack_page(p), task_stack_page(p) +
1410
				   THREAD_SIZE, NULL);
1411
		} while_each_thread(g, p);
1412 1413 1414 1415 1416
		read_unlock(&tasklist_lock);
	}

	/*
	 * Scan the objects already referenced from the sections scanned
1417
	 * above.
1418
	 */
1419
	scan_gray_list();
1420 1421

	/*
1422 1423
	 * Check for new or unreferenced objects modified since the previous
	 * scan and color them gray until the next scan.
1424 1425 1426 1427
	 */
	rcu_read_lock();
	list_for_each_entry_rcu(object, &object_list, object_list) {
		spin_lock_irqsave(&object->lock, flags);
1428 1429 1430 1431
		if (color_white(object) && (object->flags & OBJECT_ALLOCATED)
		    && update_checksum(object) && get_object(object)) {
			/* color it gray temporarily */
			object->count = object->min_count;
1432 1433 1434 1435 1436 1437
			list_add_tail(&object->gray_list, &gray_list);
		}
		spin_unlock_irqrestore(&object->lock, flags);
	}
	rcu_read_unlock();

1438 1439 1440 1441
	/*
	 * Re-scan the gray list for modified unreferenced objects.
	 */
	scan_gray_list();
1442

1443
	/*
1444
	 * If scanning was stopped do not report any new unreferenced objects.
1445
	 */
1446
	if (scan_should_stop())
1447 1448
		return;

1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
	/*
	 * Scanning result reporting.
	 */
	rcu_read_lock();
	list_for_each_entry_rcu(object, &object_list, object_list) {
		spin_lock_irqsave(&object->lock, flags);
		if (unreferenced_object(object) &&
		    !(object->flags & OBJECT_REPORTED)) {
			object->flags |= OBJECT_REPORTED;
			new_leaks++;
		}
		spin_unlock_irqrestore(&object->lock, flags);
	}
	rcu_read_unlock();

1464 1465 1466
	if (new_leaks) {
		kmemleak_found_leaks = true;

1467 1468
		pr_info("%d new suspected memory leaks (see "
			"/sys/kernel/debug/kmemleak)\n", new_leaks);
1469
	}
1470

1471 1472 1473 1474 1475 1476 1477 1478 1479 1480
}

/*
 * Thread function performing automatic memory scanning. Unreferenced objects
 * at the end of a memory scan are reported but only the first time.
 */
static int kmemleak_scan_thread(void *arg)
{
	static int first_run = 1;

Joe Perches's avatar
Joe Perches committed
1481
	pr_info("Automatic memory scanning thread started\n");
1482
	set_user_nice(current, 10);
1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497

	/*
	 * Wait before the first scan to allow the system to fully initialize.
	 */
	if (first_run) {
		first_run = 0;
		ssleep(SECS_FIRST_SCAN);
	}

	while (!kthread_should_stop()) {
		signed long timeout = jiffies_scan_wait;

		mutex_lock(&scan_mutex);
		kmemleak_scan();
		mutex_unlock(&scan_mutex);
1498

1499 1500 1501 1502 1503
		/* wait before the next scan */
		while (timeout && !kthread_should_stop())
			timeout = schedule_timeout_interruptible(timeout);
	}

Joe Perches's avatar
Joe Perches committed
1504
	pr_info("Automatic memory scanning thread ended\n");
1505 1506 1507 1508 1509 1510

	return 0;
}

/*
 * Start the automatic memory scanning thread. This function must be called
1511
 * with the scan_mutex held.
1512
 */
1513
static void start_scan_thread(void)
1514 1515 1516 1517 1518
{
	if (scan_thread)
		return;
	scan_thread = kthread_run(kmemleak_scan_thread, NULL, "kmemleak");
	if (IS_ERR(scan_thread)) {
Joe Perches's avatar
Joe Perches committed
1519
		pr_warning("Failed to create the scan thread\n");
1520 1521 1522 1523 1524 1525
		scan_thread = NULL;
	}
}

/*
 * Stop the automatic memory scanning thread. This function must be called
1526
 * with the scan_mutex held.
1527
 */
1528
static void stop_scan_thread(void)
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544
{
	if (scan_thread) {
		kthread_stop(scan_thread);
		scan_thread = NULL;
	}
}

/*
 * Iterate over the object_list and return the first valid object at or after
 * the required position with its use_count incremented. The function triggers
 * a memory scanning when the pos argument points to the first position.
 */
static void *kmemleak_seq_start(struct seq_file *seq, loff_t *pos)
{
	struct kmemleak_object *object;
	loff_t n = *pos;
1545 1546 1547 1548 1549
	int err;

	err = mutex_lock_interruptible(&scan_mutex);
	if (err < 0)
		return ERR_PTR(err);
1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570

	rcu_read_lock();
	list_for_each_entry_rcu(object, &object_list, object_list) {
		if (n-- > 0)
			continue;
		if (get_object(object))
			goto out;
	}
	object = NULL;
out:
	return object;
}

/*
 * Return the next object in the object_list. The function decrements the
 * use_count of the previous object and increases that of the next one.
 */
static void *kmemleak_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	struct kmemleak_object *prev_obj = v;
	struct kmemleak_object *next_obj = NULL;
1571
	struct kmemleak_object *obj = prev_obj;
1572 1573 1574

	++(*pos);

1575
	list_for_each_entry_continue_rcu(obj, &object_list, object_list) {
1576 1577
		if (get_object(obj)) {
			next_obj = obj;
1578
			break;
1579
		}
1580
	}
1581

1582 1583 1584 1585 1586 1587 1588 1589