[linux-block.git] / lib / stackdepot.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 * Generic stack depot for storing stack traces.
 *
 * Some debugging tools need to save stack traces of certain events which can
 * be later presented to the user. For example, KASAN needs to safe alloc and
 * free stacks for each object, but storing two stack traces per object
 * requires too much memory (e.g. SLUB_DEBUG needs 256 bytes per object for
 * that).
 *
 * Instead, stack depot maintains a hashtable of unique stacktraces. Since alloc
 * and free stacks repeat a lot, we save about 100x space.
 * Stacks are never removed from depot, so we store them contiguously one after
 * another in a contiguous memory allocation.
 *
 * Author: Alexander Potapenko <glider@google.com>
 * Copyright (C) 2016 Google, Inc.
 *
 * Based on code by Dmitry Chernenkov.
 */

#include <linux/gfp.h>
#include <linux/jhash.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/mutex.h>
#include <linux/percpu.h>
#include <linux/printk.h>
#include <linux/slab.h>
#include <linux/stacktrace.h>
#include <linux/stackdepot.h>
#include <linux/string.h>
#include <linux/types.h>
#include <linux/memblock.h>

#define DEPOT_STACK_BITS (sizeof(depot_stack_handle_t) * 8)

#define STACK_ALLOC_NULL_PROTECTION_BITS 1
#define STACK_ALLOC_ORDER 2 /* 'Slab' size order for stack depot, 4 pages */
#define STACK_ALLOC_SIZE (1LL << (PAGE_SHIFT + STACK_ALLOC_ORDER))
#define STACK_ALLOC_ALIGN 4
#define STACK_ALLOC_OFFSET_BITS (STACK_ALLOC_ORDER + PAGE_SHIFT - \
					STACK_ALLOC_ALIGN)
#define STACK_ALLOC_INDEX_BITS (DEPOT_STACK_BITS - \
		STACK_ALLOC_NULL_PROTECTION_BITS - STACK_ALLOC_OFFSET_BITS)
#define STACK_ALLOC_SLABS_CAP 8192
#define STACK_ALLOC_MAX_SLABS \
	(((1LL << (STACK_ALLOC_INDEX_BITS)) < STACK_ALLOC_SLABS_CAP) ? \
	 (1LL << (STACK_ALLOC_INDEX_BITS)) : STACK_ALLOC_SLABS_CAP)

/* The compact structure to store the reference to stacks. */
union handle_parts {
	depot_stack_handle_t handle;
	struct {
		u32 slabindex : STACK_ALLOC_INDEX_BITS;
		u32 offset : STACK_ALLOC_OFFSET_BITS;
		u32 valid : STACK_ALLOC_NULL_PROTECTION_BITS;
	};
};

struct stack_record {
	struct stack_record *next;	/* Link in the hashtable */
	u32 hash;			/* Hash in the hastable */
	u32 size;			/* Number of frames in the stack */
	union handle_parts handle;
	unsigned long entries[];	/* Variable-sized array of entries. */
};

static void *stack_slabs[STACK_ALLOC_MAX_SLABS];

static int depot_index;
static int next_slab_inited;
static size_t depot_offset;
static DEFINE_RAW_SPINLOCK(depot_lock);

static bool init_stack_slab(void **prealloc)
{
	if (!*prealloc)
		return false;
	/*
	 * This smp_load_acquire() pairs with smp_store_release() to
	 * |next_slab_inited| below and in depot_alloc_stack().
	 */
	if (smp_load_acquire(&next_slab_inited))
		return true;
	if (stack_slabs[depot_index] == NULL) {
		stack_slabs[depot_index] = *prealloc;
		*prealloc = NULL;
	} else {
		/* If this is the last depot slab, do not touch the next one. */
		if (depot_index + 1 < STACK_ALLOC_MAX_SLABS) {
			stack_slabs[depot_index + 1] = *prealloc;
			*prealloc = NULL;
		}
		/*
		 * This smp_store_release pairs with smp_load_acquire() from
		 * |next_slab_inited| above and in stack_depot_save().
		 */
		smp_store_release(&next_slab_inited, 1);
	}
	return true;
}

/* Allocation of a new stack in raw storage */
static struct stack_record *
depot_alloc_stack(unsigned long *entries, int size, u32 hash, void **prealloc)
{
	struct stack_record *stack;
	size_t required_size = struct_size(stack, entries, size);

	required_size = ALIGN(required_size, 1 << STACK_ALLOC_ALIGN);

	if (unlikely(depot_offset + required_size > STACK_ALLOC_SIZE)) {
		if (unlikely(depot_index + 1 >= STACK_ALLOC_MAX_SLABS)) {
			WARN_ONCE(1, "Stack depot reached limit capacity");
			return NULL;
		}
		depot_index++;
		depot_offset = 0;
		/*
		 * smp_store_release() here pairs with smp_load_acquire() from
		 * |next_slab_inited| in stack_depot_save() and
		 * init_stack_slab().
		 */
		if (depot_index + 1 < STACK_ALLOC_MAX_SLABS)
			smp_store_release(&next_slab_inited, 0);
	}
	init_stack_slab(prealloc);
	if (stack_slabs[depot_index] == NULL)
		return NULL;

	stack = stack_slabs[depot_index] + depot_offset;

	stack->hash = hash;
	stack->size = size;
	stack->handle.slabindex = depot_index;
	stack->handle.offset = depot_offset >> STACK_ALLOC_ALIGN;
	stack->handle.valid = 1;
	memcpy(stack->entries, entries, flex_array_size(stack, entries, size));
	depot_offset += required_size;

	return stack;
}

#define STACK_HASH_SIZE (1L << CONFIG_STACK_HASH_ORDER)
#define STACK_HASH_MASK (STACK_HASH_SIZE - 1)
#define STACK_HASH_SEED 0x9747b28c

static bool stack_depot_disable;
static struct stack_record **stack_table;

static int __init is_stack_depot_disabled(char *str)
{
	int ret;

	ret = kstrtobool(str, &stack_depot_disable);
	if (!ret && stack_depot_disable) {
		pr_info("Stack Depot is disabled\n");
		stack_table = NULL;
	}
	return 0;
}
early_param("stack_depot_disable", is_stack_depot_disabled);

/*
 * __ref because of memblock_alloc(), which will not be actually called after
 * the __init code is gone, because at that point slab_is_available() is true
 */
__ref int stack_depot_init(void)
{
	static DEFINE_MUTEX(stack_depot_init_mutex);

	mutex_lock(&stack_depot_init_mutex);
	if (!stack_depot_disable && !stack_table) {
		size_t size = (STACK_HASH_SIZE * sizeof(struct stack_record *));
		int i;

		if (slab_is_available()) {
			pr_info("Stack Depot allocating hash table with kvmalloc\n");
			stack_table = kvmalloc(size, GFP_KERNEL);
		} else {
			pr_info("Stack Depot allocating hash table with memblock_alloc\n");
			stack_table = memblock_alloc(size, SMP_CACHE_BYTES);
		}
		if (stack_table) {
			for (i = 0; i < STACK_HASH_SIZE;  i++)
				stack_table[i] = NULL;
		} else {
			pr_err("Stack Depot hash table allocation failed, disabling\n");
			stack_depot_disable = true;
			mutex_unlock(&stack_depot_init_mutex);
			return -ENOMEM;
		}
	}
	mutex_unlock(&stack_depot_init_mutex);
	return 0;
}
EXPORT_SYMBOL_GPL(stack_depot_init);

/* Calculate hash for a stack */
static inline u32 hash_stack(unsigned long *entries, unsigned int size)
{
	return jhash2((u32 *)entries,
		      array_size(size,  sizeof(*entries)) / sizeof(u32),
		      STACK_HASH_SEED);
}

/* Use our own, non-instrumented version of memcmp().
 *
 * We actually don't care about the order, just the equality.
 */
static inline
int stackdepot_memcmp(const unsigned long *u1, const unsigned long *u2,
			unsigned int n)
{
	for ( ; n-- ; u1++, u2++) {
		if (*u1 != *u2)
			return 1;
	}
	return 0;
}

/* Find a stack that is equal to the one stored in entries in the hash */
static inline struct stack_record *find_stack(struct stack_record *bucket,
					     unsigned long *entries, int size,
					     u32 hash)
{
	struct stack_record *found;

	for (found = bucket; found; found = found->next) {
		if (found->hash == hash &&
		    found->size == size &&
		    !stackdepot_memcmp(entries, found->entries, size))
			return found;
	}
	return NULL;
}

/**
 * stack_depot_snprint - print stack entries from a depot into a buffer
 *
 * @handle:	Stack depot handle which was returned from
 *		stack_depot_save().
 * @buf:	Pointer to the print buffer
 *
 * @size:	Size of the print buffer
 *
 * @spaces:	Number of leading spaces to print
 *
 * Return:	Number of bytes printed.
 */
int stack_depot_snprint(depot_stack_handle_t handle, char *buf, size_t size,
		       int spaces)
{
	unsigned long *entries;
	unsigned int nr_entries;

	nr_entries = stack_depot_fetch(handle, &entries);
	return nr_entries ? stack_trace_snprint(buf, size, entries, nr_entries,
						spaces) : 0;
}
EXPORT_SYMBOL_GPL(stack_depot_snprint);

/**
 * stack_depot_print - print stack entries from a depot
 *
 * @stack:		Stack depot handle which was returned from
 *			stack_depot_save().
 *
 */
void stack_depot_print(depot_stack_handle_t stack)
{
	unsigned long *entries;
	unsigned int nr_entries;

	nr_entries = stack_depot_fetch(stack, &entries);
	if (nr_entries > 0)
		stack_trace_print(entries, nr_entries, 0);
}
EXPORT_SYMBOL_GPL(stack_depot_print);

/**
 * stack_depot_fetch - Fetch stack entries from a depot
 *
 * @handle:		Stack depot handle which was returned from
 *			stack_depot_save().
 * @entries:		Pointer to store the entries address
 *
 * Return: The number of trace entries for this depot.
 */
unsigned int stack_depot_fetch(depot_stack_handle_t handle,
			       unsigned long **entries)
{
	union handle_parts parts = { .handle = handle };
	void *slab;
	size_t offset = parts.offset << STACK_ALLOC_ALIGN;
	struct stack_record *stack;

	*entries = NULL;
	if (!handle)
		return 0;

	if (parts.slabindex > depot_index) {
		WARN(1, "slab index %d out of bounds (%d) for stack id %08x\n",
			parts.slabindex, depot_index, handle);
		return 0;
	}
	slab = stack_slabs[parts.slabindex];
	if (!slab)
		return 0;
	stack = slab + offset;

	*entries = stack->entries;
	return stack->size;
}
EXPORT_SYMBOL_GPL(stack_depot_fetch);

/**
 * __stack_depot_save - Save a stack trace from an array
 *
 * @entries:		Pointer to storage array
 * @nr_entries:		Size of the storage array
 * @alloc_flags:	Allocation gfp flags
 * @can_alloc:		Allocate stack slabs (increased chance of failure if false)
 *
 * Saves a stack trace from @entries array of size @nr_entries. If @can_alloc is
 * %true, is allowed to replenish the stack slab pool in case no space is left
 * (allocates using GFP flags of @alloc_flags). If @can_alloc is %false, avoids
 * any allocations and will fail if no space is left to store the stack trace.
 *
 * If the stack trace in @entries is from an interrupt, only the portion up to
 * interrupt entry is saved.
 *
 * Context: Any context, but setting @can_alloc to %false is required if
 *          alloc_pages() cannot be used from the current context. Currently
 *          this is the case from contexts where neither %GFP_ATOMIC nor
 *          %GFP_NOWAIT can be used (NMI, raw_spin_lock).
 *
 * Return: The handle of the stack struct stored in depot, 0 on failure.
 */
depot_stack_handle_t __stack_depot_save(unsigned long *entries,
					unsigned int nr_entries,
					gfp_t alloc_flags, bool can_alloc)
{
	struct stack_record *found = NULL, **bucket;
	depot_stack_handle_t retval = 0;
	struct page *page = NULL;
	void *prealloc = NULL;
	unsigned long flags;
	u32 hash;

	/*
	 * If this stack trace is from an interrupt, including anything before
	 * interrupt entry usually leads to unbounded stackdepot growth.
	 *
	 * Because use of filter_irq_stacks() is a requirement to ensure
	 * stackdepot can efficiently deduplicate interrupt stacks, always
	 * filter_irq_stacks() to simplify all callers' use of stackdepot.
	 */
	nr_entries = filter_irq_stacks(entries, nr_entries);

	if (unlikely(nr_entries == 0) || stack_depot_disable)
		goto fast_exit;

	hash = hash_stack(entries, nr_entries);
	bucket = &stack_table[hash & STACK_HASH_MASK];

	/*
	 * Fast path: look the stack trace up without locking.
	 * The smp_load_acquire() here pairs with smp_store_release() to
	 * |bucket| below.
	 */
	found = find_stack(smp_load_acquire(bucket), entries,
			   nr_entries, hash);
	if (found)
		goto exit;

	/*
	 * Check if the current or the next stack slab need to be initialized.
	 * If so, allocate the memory - we won't be able to do that under the
	 * lock.
	 *
	 * The smp_load_acquire() here pairs with smp_store_release() to
	 * |next_slab_inited| in depot_alloc_stack() and init_stack_slab().
	 */
	if (unlikely(can_alloc && !smp_load_acquire(&next_slab_inited))) {
		/*
		 * Zero out zone modifiers, as we don't have specific zone
		 * requirements. Keep the flags related to allocation in atomic
		 * contexts and I/O.
		 */
		alloc_flags &= ~GFP_ZONEMASK;
		alloc_flags &= (GFP_ATOMIC | GFP_KERNEL);
		alloc_flags |= __GFP_NOWARN;
		page = alloc_pages(alloc_flags, STACK_ALLOC_ORDER);
		if (page)
			prealloc = page_address(page);
	}

	raw_spin_lock_irqsave(&depot_lock, flags);

	found = find_stack(*bucket, entries, nr_entries, hash);
	if (!found) {
		struct stack_record *new = depot_alloc_stack(entries, nr_entries, hash, &prealloc);

		if (new) {
			new->next = *bucket;
			/*
			 * This smp_store_release() pairs with
			 * smp_load_acquire() from |bucket| above.
			 */
			smp_store_release(bucket, new);
			found = new;
		}
	} else if (prealloc) {
		/*
		 * We didn't need to store this stack trace, but let's keep
		 * the preallocated memory for the future.
		 */
		WARN_ON(!init_stack_slab(&prealloc));
	}

	raw_spin_unlock_irqrestore(&depot_lock, flags);
exit:
	if (prealloc) {
		/* Nobody used this memory, ok to free it. */
		free_pages((unsigned long)prealloc, STACK_ALLOC_ORDER);
	}
	if (found)
		retval = found->handle.handle;
fast_exit:
	return retval;
}
EXPORT_SYMBOL_GPL(__stack_depot_save);

/**
 * stack_depot_save - Save a stack trace from an array
 *
 * @entries:		Pointer to storage array
 * @nr_entries:		Size of the storage array
 * @alloc_flags:	Allocation gfp flags
 *
 * Context: Contexts where allocations via alloc_pages() are allowed.
 *          See __stack_depot_save() for more details.
 *
 * Return: The handle of the stack struct stored in depot, 0 on failure.
 */
depot_stack_handle_t stack_depot_save(unsigned long *entries,
				      unsigned int nr_entries,
				      gfp_t alloc_flags)
{
	return __stack_depot_save(entries, nr_entries, alloc_flags, true);
}
EXPORT_SYMBOL_GPL(stack_depot_save);
Commit	Line	Data
1802d0be	1	// SPDX-License-Identifier: GPL-2.0-only
cd11016e AP	2	/*
	3	* Generic stack depot for storing stack traces.
	4	*
	5	* Some debugging tools need to save stack traces of certain events which can
	6	* be later presented to the user. For example, KASAN needs to safe alloc and
	7	* free stacks for each object, but storing two stack traces per object
	8	* requires too much memory (e.g. SLUB_DEBUG needs 256 bytes per object for
	9	* that).
	10	*
	11	* Instead, stack depot maintains a hashtable of unique stacktraces. Since alloc
	12	* and free stacks repeat a lot, we save about 100x space.
	13	* Stacks are never removed from depot, so we store them contiguously one after
9dbbc3b9	14	* another in a contiguous memory allocation.
cd11016e AP	15	*
	16	* Author: Alexander Potapenko <glider@google.com>
	17	* Copyright (C) 2016 Google, Inc.
	18	*
	19	* Based on code by Dmitry Chernenkov.
cd11016e AP	20	*/
	21
	22	#include <linux/gfp.h>
	23	#include <linux/jhash.h>
	24	#include <linux/kernel.h>
	25	#include <linux/mm.h>
2dba5eb1	26	#include <linux/mutex.h>
cd11016e AP	27	#include <linux/percpu.h>
	28	#include <linux/printk.h>
	29	#include <linux/slab.h>
	30	#include <linux/stacktrace.h>
	31	#include <linux/stackdepot.h>
	32	#include <linux/string.h>
	33	#include <linux/types.h>
e1fdc403	34	#include <linux/memblock.h>
cd11016e AP	35
	36	#define DEPOT_STACK_BITS (sizeof(depot_stack_handle_t) * 8)
	37
7c31190b	38	#define STACK_ALLOC_NULL_PROTECTION_BITS 1
cd11016e AP	39	#define STACK_ALLOC_ORDER 2 /* 'Slab' size order for stack depot, 4 pages */
	40	#define STACK_ALLOC_SIZE (1LL << (PAGE_SHIFT + STACK_ALLOC_ORDER))
	41	#define STACK_ALLOC_ALIGN 4
	42	#define STACK_ALLOC_OFFSET_BITS (STACK_ALLOC_ORDER + PAGE_SHIFT - \
	43	STACK_ALLOC_ALIGN)
7c31190b JK	44	#define STACK_ALLOC_INDEX_BITS (DEPOT_STACK_BITS - \
7c31190b JK	45	STACK_ALLOC_NULL_PROTECTION_BITS - STACK_ALLOC_OFFSET_BITS)
02754e0a	46	#define STACK_ALLOC_SLABS_CAP 8192
cd11016e AP	47	#define STACK_ALLOC_MAX_SLABS \
	48	(((1LL << (STACK_ALLOC_INDEX_BITS)) < STACK_ALLOC_SLABS_CAP) ? \
	49	(1LL << (STACK_ALLOC_INDEX_BITS)) : STACK_ALLOC_SLABS_CAP)
	50
	51	/* The compact structure to store the reference to stacks. */
	52	union handle_parts {
	53	depot_stack_handle_t handle;
	54	struct {
	55	u32 slabindex : STACK_ALLOC_INDEX_BITS;
	56	u32 offset : STACK_ALLOC_OFFSET_BITS;
7c31190b	57	u32 valid : STACK_ALLOC_NULL_PROTECTION_BITS;
cd11016e AP	58	};
	59	};
	60
	61	struct stack_record {
	62	struct stack_record next; / Link in the hashtable */
	63	u32 hash; /* Hash in the hastable */
	64	u32 size; /* Number of frames in the stack */
	65	union handle_parts handle;
3a2b67e6	66	unsigned long entries[]; /* Variable-sized array of entries. */
cd11016e AP	67	};
	68
	69	static void *stack_slabs[STACK_ALLOC_MAX_SLABS];
	70
	71	static int depot_index;
	72	static int next_slab_inited;
	73	static size_t depot_offset;
78564b94	74	static DEFINE_RAW_SPINLOCK(depot_lock);
cd11016e AP	75
	76	static bool init_stack_slab(void **prealloc)
	77	{
	78	if (!*prealloc)
	79	return false;
	80	/*
	81	* This smp_load_acquire() pairs with smp_store_release() to
	82	* \|next_slab_inited\| below and in depot_alloc_stack().
	83	*/
	84	if (smp_load_acquire(&next_slab_inited))
	85	return true;
	86	if (stack_slabs[depot_index] == NULL) {
	87	stack_slabs[depot_index] = *prealloc;
305e519c	88	*prealloc = NULL;
cd11016e	89	} else {
305e519c AP	90	/* If this is the last depot slab, do not touch the next one. */
	91	if (depot_index + 1 < STACK_ALLOC_MAX_SLABS) {
	92	stack_slabs[depot_index + 1] = *prealloc;
	93	*prealloc = NULL;
	94	}
cd11016e AP	95	/*
cd11016e AP	96	* This smp_store_release pairs with smp_load_acquire() from
ee050dc8	97	* \|next_slab_inited\| above and in stack_depot_save().
cd11016e AP	98	*/
	99	smp_store_release(&next_slab_inited, 1);
	100	}
cd11016e AP	101	return true;
	102	}
	103
	104	/* Allocation of a new stack in raw storage */
7f2b8818 ME	105	static struct stack_record *
7f2b8818 ME	106	depot_alloc_stack(unsigned long entries, int size, u32 hash, void *prealloc)
cd11016e	107	{
cd11016e	108	struct stack_record *stack;
3a2b67e6	109	size_t required_size = struct_size(stack, entries, size);
cd11016e AP	110
	111	required_size = ALIGN(required_size, 1 << STACK_ALLOC_ALIGN);
	112
	113	if (unlikely(depot_offset + required_size > STACK_ALLOC_SIZE)) {
	114	if (unlikely(depot_index + 1 >= STACK_ALLOC_MAX_SLABS)) {
	115	WARN_ONCE(1, "Stack depot reached limit capacity");
	116	return NULL;
	117	}
	118	depot_index++;
	119	depot_offset = 0;
	120	/*
	121	* smp_store_release() here pairs with smp_load_acquire() from
ee050dc8	122	* \|next_slab_inited\| in stack_depot_save() and
cd11016e AP	123	* init_stack_slab().
	124	*/
	125	if (depot_index + 1 < STACK_ALLOC_MAX_SLABS)
	126	smp_store_release(&next_slab_inited, 0);
	127	}
	128	init_stack_slab(prealloc);
	129	if (stack_slabs[depot_index] == NULL)
	130	return NULL;
	131
	132	stack = stack_slabs[depot_index] + depot_offset;
	133
	134	stack->hash = hash;
	135	stack->size = size;
	136	stack->handle.slabindex = depot_index;
	137	stack->handle.offset = depot_offset >> STACK_ALLOC_ALIGN;
7c31190b	138	stack->handle.valid = 1;
47e684aa	139	memcpy(stack->entries, entries, flex_array_size(stack, entries, size));
cd11016e AP	140	depot_offset += required_size;
	141
	142	return stack;
	143	}
	144
d2620936	145	#define STACK_HASH_SIZE (1L << CONFIG_STACK_HASH_ORDER)
cd11016e AP	146	#define STACK_HASH_MASK (STACK_HASH_SIZE - 1)
	147	#define STACK_HASH_SEED 0x9747b28c
	148
e1fdc403 VJ	149	static bool stack_depot_disable;
	150	static struct stack_record **stack_table;
	151
	152	static int __init is_stack_depot_disabled(char *str)
	153	{
64427985 VJ	154	int ret;
	155
	156	ret = kstrtobool(str, &stack_depot_disable);
	157	if (!ret && stack_depot_disable) {
e1fdc403 VJ	158	pr_info("Stack Depot is disabled\n");
	159	stack_table = NULL;
	160	}
	161	return 0;
	162	}
	163	early_param("stack_depot_disable", is_stack_depot_disabled);
	164
2dba5eb1 VB	165	/*
	166	* __ref because of memblock_alloc(), which will not be actually called after
	167	* the __init code is gone, because at that point slab_is_available() is true
	168	*/
	169	__ref int stack_depot_init(void)
e1fdc403	170	{
2dba5eb1 VB	171	static DEFINE_MUTEX(stack_depot_init_mutex);
	172
	173	mutex_lock(&stack_depot_init_mutex);
	174	if (!stack_depot_disable && !stack_table) {
e1fdc403 VJ	175	size_t size = (STACK_HASH_SIZE * sizeof(struct stack_record *));
	176	int i;
	177
2dba5eb1 VB	178	if (slab_is_available()) {
	179	pr_info("Stack Depot allocating hash table with kvmalloc\n");
	180	stack_table = kvmalloc(size, GFP_KERNEL);
	181	} else {
	182	pr_info("Stack Depot allocating hash table with memblock_alloc\n");
	183	stack_table = memblock_alloc(size, SMP_CACHE_BYTES);
	184	}
	185	if (stack_table) {
	186	for (i = 0; i < STACK_HASH_SIZE; i++)
	187	stack_table[i] = NULL;
	188	} else {
	189	pr_err("Stack Depot hash table allocation failed, disabling\n");
	190	stack_depot_disable = true;
	191	mutex_unlock(&stack_depot_init_mutex);
	192	return -ENOMEM;
	193	}
e1fdc403	194	}
2dba5eb1	195	mutex_unlock(&stack_depot_init_mutex);
e1fdc403 VJ	196	return 0;
e1fdc403 VJ	197	}
2dba5eb1	198	EXPORT_SYMBOL_GPL(stack_depot_init);
cd11016e AP	199
	200	/* Calculate hash for a stack */
	201	static inline u32 hash_stack(unsigned long *entries, unsigned int size)
	202	{
	203	return jhash2((u32 *)entries,
180644f8 GS	204	array_size(size, sizeof(*entries)) / sizeof(u32),
180644f8 GS	205	STACK_HASH_SEED);
cd11016e AP	206	}
cd11016e AP	207
a571b272 AP	208	/* Use our own, non-instrumented version of memcmp().
	209	*
	210	* We actually don't care about the order, just the equality.
	211	*/
	212	static inline
	213	int stackdepot_memcmp(const unsigned long u1, const unsigned long u2,
	214	unsigned int n)
	215	{
	216	for ( ; n-- ; u1++, u2++) {
	217	if (u1 != u2)
	218	return 1;
	219	}
	220	return 0;
	221	}
	222
cd11016e AP	223	/* Find a stack that is equal to the one stored in entries in the hash */
	224	static inline struct stack_record find_stack(struct stack_record bucket,
	225	unsigned long *entries, int size,
	226	u32 hash)
	227	{
	228	struct stack_record *found;
	229
	230	for (found = bucket; found; found = found->next) {
	231	if (found->hash == hash &&
	232	found->size == size &&
a571b272	233	!stackdepot_memcmp(entries, found->entries, size))
cd11016e	234	return found;
cd11016e AP	235	}
	236	return NULL;
	237	}
	238
0f68d45e IK	239	/**
	240	* stack_depot_snprint - print stack entries from a depot into a buffer
	241	*
	242	* @handle: Stack depot handle which was returned from
	243	* stack_depot_save().
	244	* @buf: Pointer to the print buffer
	245	*
	246	* @size: Size of the print buffer
	247	*
	248	* @spaces: Number of leading spaces to print
	249	*
	250	* Return: Number of bytes printed.
	251	*/
	252	int stack_depot_snprint(depot_stack_handle_t handle, char *buf, size_t size,
	253	int spaces)
	254	{
	255	unsigned long *entries;
	256	unsigned int nr_entries;
	257
	258	nr_entries = stack_depot_fetch(handle, &entries);
	259	return nr_entries ? stack_trace_snprint(buf, size, entries, nr_entries,
	260	spaces) : 0;
	261	}
	262	EXPORT_SYMBOL_GPL(stack_depot_snprint);
	263
505be481 IK	264	/**
	265	* stack_depot_print - print stack entries from a depot
	266	*
	267	* @stack: Stack depot handle which was returned from
	268	* stack_depot_save().
	269	*
	270	*/
	271	void stack_depot_print(depot_stack_handle_t stack)
	272	{
	273	unsigned long *entries;
	274	unsigned int nr_entries;
	275
	276	nr_entries = stack_depot_fetch(stack, &entries);
	277	if (nr_entries > 0)
	278	stack_trace_print(entries, nr_entries, 0);
	279	}
	280	EXPORT_SYMBOL_GPL(stack_depot_print);
	281
c0cfc337 TG	282	/**
	283	* stack_depot_fetch - Fetch stack entries from a depot
	284	*
	285	* @handle: Stack depot handle which was returned from
	286	* stack_depot_save().
	287	* @entries: Pointer to store the entries address
	288	*
	289	* Return: The number of trace entries for this depot.
	290	*/
	291	unsigned int stack_depot_fetch(depot_stack_handle_t handle,
	292	unsigned long **entries)
cd11016e AP	293	{
cd11016e AP	294	union handle_parts parts = { .handle = handle };
69866e15	295	void *slab;
cd11016e	296	size_t offset = parts.offset << STACK_ALLOC_ALIGN;
69866e15 AP	297	struct stack_record *stack;
	298
	299	*entries = NULL;
4d4712c1 IK	300	if (!handle)
	301	return 0;
	302
69866e15 AP	303	if (parts.slabindex > depot_index) {
	304	WARN(1, "slab index %d out of bounds (%d) for stack id %08x\n",
	305	parts.slabindex, depot_index, handle);
	306	return 0;
	307	}
	308	slab = stack_slabs[parts.slabindex];
	309	if (!slab)
	310	return 0;
	311	stack = slab + offset;
cd11016e	312
c0cfc337 TG	313	*entries = stack->entries;
	314	return stack->size;
	315	}
	316	EXPORT_SYMBOL_GPL(stack_depot_fetch);
	317
cd11016e	318	/**
11ac25c6	319	* __stack_depot_save - Save a stack trace from an array
c0cfc337 TG	320	*
	321	* @entries: Pointer to storage array
	322	* @nr_entries: Size of the storage array
	323	* @alloc_flags: Allocation gfp flags
11ac25c6 ME	324	* @can_alloc: Allocate stack slabs (increased chance of failure if false)
	325	*
	326	* Saves a stack trace from @entries array of size @nr_entries. If @can_alloc is
	327	* %true, is allowed to replenish the stack slab pool in case no space is left
	328	* (allocates using GFP flags of @alloc_flags). If @can_alloc is %false, avoids
	329	* any allocations and will fail if no space is left to store the stack trace.
	330	*
e9400660 ME	331	* If the stack trace in @entries is from an interrupt, only the portion up to
	332	* interrupt entry is saved.
	333	*
11ac25c6 ME	334	* Context: Any context, but setting @can_alloc to %false is required if
	335	* alloc_pages() cannot be used from the current context. Currently
	336	* this is the case from contexts where neither %GFP_ATOMIC nor
	337	* %GFP_NOWAIT can be used (NMI, raw_spin_lock).
cd11016e	338	*
11ac25c6	339	* Return: The handle of the stack struct stored in depot, 0 on failure.
cd11016e	340	*/
11ac25c6 ME	341	depot_stack_handle_t __stack_depot_save(unsigned long *entries,
	342	unsigned int nr_entries,
	343	gfp_t alloc_flags, bool can_alloc)
cd11016e	344	{
cd11016e	345	struct stack_record found = NULL, *bucket;
c0cfc337	346	depot_stack_handle_t retval = 0;
cd11016e AP	347	struct page *page = NULL;
cd11016e AP	348	void *prealloc = NULL;
c0cfc337 TG	349	unsigned long flags;
c0cfc337 TG	350	u32 hash;
cd11016e	351
e9400660 ME	352	/*
	353	* If this stack trace is from an interrupt, including anything before
	354	* interrupt entry usually leads to unbounded stackdepot growth.
	355	*
	356	* Because use of filter_irq_stacks() is a requirement to ensure
	357	* stackdepot can efficiently deduplicate interrupt stacks, always
	358	* filter_irq_stacks() to simplify all callers' use of stackdepot.
	359	*/
	360	nr_entries = filter_irq_stacks(entries, nr_entries);
	361
e1fdc403	362	if (unlikely(nr_entries == 0) \|\| stack_depot_disable)
cd11016e AP	363	goto fast_exit;
cd11016e AP	364
c0cfc337	365	hash = hash_stack(entries, nr_entries);
cd11016e AP	366	bucket = &stack_table[hash & STACK_HASH_MASK];
	367
	368	/*
	369	* Fast path: look the stack trace up without locking.
	370	* The smp_load_acquire() here pairs with smp_store_release() to
	371	* \|bucket\| below.
	372	*/
c0cfc337 TG	373	found = find_stack(smp_load_acquire(bucket), entries,
c0cfc337 TG	374	nr_entries, hash);
cd11016e AP	375	if (found)
	376	goto exit;
	377
	378	/*
	379	* Check if the current or the next stack slab need to be initialized.
	380	* If so, allocate the memory - we won't be able to do that under the
	381	* lock.
	382	*
	383	* The smp_load_acquire() here pairs with smp_store_release() to
	384	* \|next_slab_inited\| in depot_alloc_stack() and init_stack_slab().
	385	*/
11ac25c6	386	if (unlikely(can_alloc && !smp_load_acquire(&next_slab_inited))) {
cd11016e AP	387	/*
	388	* Zero out zone modifiers, as we don't have specific zone
	389	* requirements. Keep the flags related to allocation in atomic
	390	* contexts and I/O.
	391	*/
	392	alloc_flags &= ~GFP_ZONEMASK;
	393	alloc_flags &= (GFP_ATOMIC \| GFP_KERNEL);
87cc271d	394	alloc_flags \|= __GFP_NOWARN;
cd11016e AP	395	page = alloc_pages(alloc_flags, STACK_ALLOC_ORDER);
	396	if (page)
	397	prealloc = page_address(page);
	398	}
	399
78564b94	400	raw_spin_lock_irqsave(&depot_lock, flags);
cd11016e	401
c0cfc337	402	found = find_stack(*bucket, entries, nr_entries, hash);
cd11016e	403	if (!found) {
7f2b8818 ME	404	struct stack_record *new = depot_alloc_stack(entries, nr_entries, hash, &prealloc);
7f2b8818 ME	405
cd11016e AP	406	if (new) {
	407	new->next = *bucket;
	408	/*
	409	* This smp_store_release() pairs with
	410	* smp_load_acquire() from \|bucket\| above.
	411	*/
	412	smp_store_release(bucket, new);
	413	found = new;
	414	}
	415	} else if (prealloc) {
	416	/*
	417	* We didn't need to store this stack trace, but let's keep
	418	* the preallocated memory for the future.
	419	*/
	420	WARN_ON(!init_stack_slab(&prealloc));
	421	}
	422
78564b94	423	raw_spin_unlock_irqrestore(&depot_lock, flags);
cd11016e AP	424	exit:
	425	if (prealloc) {
	426	/* Nobody used this memory, ok to free it. */
	427	free_pages((unsigned long)prealloc, STACK_ALLOC_ORDER);
	428	}
	429	if (found)
	430	retval = found->handle.handle;
	431	fast_exit:
	432	return retval;
	433	}
11ac25c6 ME	434	EXPORT_SYMBOL_GPL(__stack_depot_save);
	435
	436	/**
	437	* stack_depot_save - Save a stack trace from an array
	438	*
	439	* @entries: Pointer to storage array
	440	* @nr_entries: Size of the storage array
	441	* @alloc_flags: Allocation gfp flags
	442	*
	443	* Context: Contexts where allocations via alloc_pages() are allowed.
	444	* See __stack_depot_save() for more details.
	445	*
	446	* Return: The handle of the stack struct stored in depot, 0 on failure.
	447	*/
	448	depot_stack_handle_t stack_depot_save(unsigned long *entries,
	449	unsigned int nr_entries,
	450	gfp_t alloc_flags)
	451	{
	452	return __stack_depot_save(entries, nr_entries, alloc_flags, true);
	453	}
c0cfc337	454	EXPORT_SYMBOL_GPL(stack_depot_save);