[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/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);

int __init stack_depot_init(void)
{
	if (!stack_depot_disable) {
		size_t size = (STACK_HASH_SIZE * sizeof(struct stack_record *));
		int i;

		stack_table = memblock_alloc(size, size);
		for (i = 0; i < STACK_HASH_SIZE;  i++)
			stack_table[i] = NULL;
	}
	return 0;
}

/* 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_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.
 *
 * 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 (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>
	26	#include <linux/percpu.h>
	27	#include <linux/printk.h>
	28	#include <linux/slab.h>
	29	#include <linux/stacktrace.h>
	30	#include <linux/stackdepot.h>
	31	#include <linux/string.h>
	32	#include <linux/types.h>
e1fdc403	33	#include <linux/memblock.h>
cd11016e AP	34
	35	#define DEPOT_STACK_BITS (sizeof(depot_stack_handle_t) * 8)
	36
7c31190b	37	#define STACK_ALLOC_NULL_PROTECTION_BITS 1
cd11016e AP	38	#define STACK_ALLOC_ORDER 2 /* 'Slab' size order for stack depot, 4 pages */
	39	#define STACK_ALLOC_SIZE (1LL << (PAGE_SHIFT + STACK_ALLOC_ORDER))
	40	#define STACK_ALLOC_ALIGN 4
	41	#define STACK_ALLOC_OFFSET_BITS (STACK_ALLOC_ORDER + PAGE_SHIFT - \
	42	STACK_ALLOC_ALIGN)
7c31190b JK	43	#define STACK_ALLOC_INDEX_BITS (DEPOT_STACK_BITS - \
7c31190b JK	44	STACK_ALLOC_NULL_PROTECTION_BITS - STACK_ALLOC_OFFSET_BITS)
02754e0a	45	#define STACK_ALLOC_SLABS_CAP 8192
cd11016e AP	46	#define STACK_ALLOC_MAX_SLABS \
	47	(((1LL << (STACK_ALLOC_INDEX_BITS)) < STACK_ALLOC_SLABS_CAP) ? \
	48	(1LL << (STACK_ALLOC_INDEX_BITS)) : STACK_ALLOC_SLABS_CAP)
	49
	50	/* The compact structure to store the reference to stacks. */
	51	union handle_parts {
	52	depot_stack_handle_t handle;
	53	struct {
	54	u32 slabindex : STACK_ALLOC_INDEX_BITS;
	55	u32 offset : STACK_ALLOC_OFFSET_BITS;
7c31190b	56	u32 valid : STACK_ALLOC_NULL_PROTECTION_BITS;
cd11016e AP	57	};
	58	};
	59
	60	struct stack_record {
	61	struct stack_record next; / Link in the hashtable */
	62	u32 hash; /* Hash in the hastable */
	63	u32 size; /* Number of frames in the stack */
	64	union handle_parts handle;
3a2b67e6	65	unsigned long entries[]; /* Variable-sized array of entries. */
cd11016e AP	66	};
	67
	68	static void *stack_slabs[STACK_ALLOC_MAX_SLABS];
	69
	70	static int depot_index;
	71	static int next_slab_inited;
	72	static size_t depot_offset;
78564b94	73	static DEFINE_RAW_SPINLOCK(depot_lock);
cd11016e AP	74
	75	static bool init_stack_slab(void **prealloc)
	76	{
	77	if (!*prealloc)
	78	return false;
	79	/*
	80	* This smp_load_acquire() pairs with smp_store_release() to
	81	* \|next_slab_inited\| below and in depot_alloc_stack().
	82	*/
	83	if (smp_load_acquire(&next_slab_inited))
	84	return true;
	85	if (stack_slabs[depot_index] == NULL) {
	86	stack_slabs[depot_index] = *prealloc;
305e519c	87	*prealloc = NULL;
cd11016e	88	} else {
305e519c AP	89	/* If this is the last depot slab, do not touch the next one. */
	90	if (depot_index + 1 < STACK_ALLOC_MAX_SLABS) {
	91	stack_slabs[depot_index + 1] = *prealloc;
	92	*prealloc = NULL;
	93	}
cd11016e AP	94	/*
cd11016e AP	95	* This smp_store_release pairs with smp_load_acquire() from
ee050dc8	96	* \|next_slab_inited\| above and in stack_depot_save().
cd11016e AP	97	*/
	98	smp_store_release(&next_slab_inited, 1);
	99	}
cd11016e AP	100	return true;
	101	}
	102
	103	/* Allocation of a new stack in raw storage */
7f2b8818 ME	104	static struct stack_record *
7f2b8818 ME	105	depot_alloc_stack(unsigned long entries, int size, u32 hash, void *prealloc)
cd11016e	106	{
cd11016e	107	struct stack_record *stack;
3a2b67e6	108	size_t required_size = struct_size(stack, entries, size);
cd11016e AP	109
	110	required_size = ALIGN(required_size, 1 << STACK_ALLOC_ALIGN);
	111
	112	if (unlikely(depot_offset + required_size > STACK_ALLOC_SIZE)) {
	113	if (unlikely(depot_index + 1 >= STACK_ALLOC_MAX_SLABS)) {
	114	WARN_ONCE(1, "Stack depot reached limit capacity");
	115	return NULL;
	116	}
	117	depot_index++;
	118	depot_offset = 0;
	119	/*
	120	* smp_store_release() here pairs with smp_load_acquire() from
ee050dc8	121	* \|next_slab_inited\| in stack_depot_save() and
cd11016e AP	122	* init_stack_slab().
	123	*/
	124	if (depot_index + 1 < STACK_ALLOC_MAX_SLABS)
	125	smp_store_release(&next_slab_inited, 0);
	126	}
	127	init_stack_slab(prealloc);
	128	if (stack_slabs[depot_index] == NULL)
	129	return NULL;
	130
	131	stack = stack_slabs[depot_index] + depot_offset;
	132
	133	stack->hash = hash;
	134	stack->size = size;
	135	stack->handle.slabindex = depot_index;
	136	stack->handle.offset = depot_offset >> STACK_ALLOC_ALIGN;
7c31190b	137	stack->handle.valid = 1;
47e684aa	138	memcpy(stack->entries, entries, flex_array_size(stack, entries, size));
cd11016e AP	139	depot_offset += required_size;
	140
	141	return stack;
	142	}
	143
d2620936	144	#define STACK_HASH_SIZE (1L << CONFIG_STACK_HASH_ORDER)
cd11016e AP	145	#define STACK_HASH_MASK (STACK_HASH_SIZE - 1)
	146	#define STACK_HASH_SEED 0x9747b28c
	147
e1fdc403 VJ	148	static bool stack_depot_disable;
	149	static struct stack_record **stack_table;
	150
	151	static int __init is_stack_depot_disabled(char *str)
	152	{
64427985 VJ	153	int ret;
	154
	155	ret = kstrtobool(str, &stack_depot_disable);
	156	if (!ret && stack_depot_disable) {
e1fdc403 VJ	157	pr_info("Stack Depot is disabled\n");
	158	stack_table = NULL;
	159	}
	160	return 0;
	161	}
	162	early_param("stack_depot_disable", is_stack_depot_disabled);
	163
	164	int __init stack_depot_init(void)
	165	{
	166	if (!stack_depot_disable) {
	167	size_t size = (STACK_HASH_SIZE * sizeof(struct stack_record *));
	168	int i;
	169
	170	stack_table = memblock_alloc(size, size);
	171	for (i = 0; i < STACK_HASH_SIZE; i++)
	172	stack_table[i] = NULL;
	173	}
	174	return 0;
	175	}
cd11016e AP	176
	177	/* Calculate hash for a stack */
	178	static inline u32 hash_stack(unsigned long *entries, unsigned int size)
	179	{
	180	return jhash2((u32 *)entries,
180644f8 GS	181	array_size(size, sizeof(*entries)) / sizeof(u32),
180644f8 GS	182	STACK_HASH_SEED);
cd11016e AP	183	}
cd11016e AP	184
a571b272 AP	185	/* Use our own, non-instrumented version of memcmp().
	186	*
	187	* We actually don't care about the order, just the equality.
	188	*/
	189	static inline
	190	int stackdepot_memcmp(const unsigned long u1, const unsigned long u2,
	191	unsigned int n)
	192	{
	193	for ( ; n-- ; u1++, u2++) {
	194	if (u1 != u2)
	195	return 1;
	196	}
	197	return 0;
	198	}
	199
cd11016e AP	200	/* Find a stack that is equal to the one stored in entries in the hash */
	201	static inline struct stack_record find_stack(struct stack_record bucket,
	202	unsigned long *entries, int size,
	203	u32 hash)
	204	{
	205	struct stack_record *found;
	206
	207	for (found = bucket; found; found = found->next) {
	208	if (found->hash == hash &&
	209	found->size == size &&
a571b272	210	!stackdepot_memcmp(entries, found->entries, size))
cd11016e	211	return found;
cd11016e AP	212	}
	213	return NULL;
	214	}
	215
505be481 IK	216	/**
	217	* stack_depot_print - print stack entries from a depot
	218	*
	219	* @stack: Stack depot handle which was returned from
	220	* stack_depot_save().
	221	*
	222	*/
	223	void stack_depot_print(depot_stack_handle_t stack)
	224	{
	225	unsigned long *entries;
	226	unsigned int nr_entries;
	227
	228	nr_entries = stack_depot_fetch(stack, &entries);
	229	if (nr_entries > 0)
	230	stack_trace_print(entries, nr_entries, 0);
	231	}
	232	EXPORT_SYMBOL_GPL(stack_depot_print);
	233
c0cfc337 TG	234	/**
	235	* stack_depot_fetch - Fetch stack entries from a depot
	236	*
	237	* @handle: Stack depot handle which was returned from
	238	* stack_depot_save().
	239	* @entries: Pointer to store the entries address
	240	*
	241	* Return: The number of trace entries for this depot.
	242	*/
	243	unsigned int stack_depot_fetch(depot_stack_handle_t handle,
	244	unsigned long **entries)
cd11016e AP	245	{
cd11016e AP	246	union handle_parts parts = { .handle = handle };
69866e15	247	void *slab;
cd11016e	248	size_t offset = parts.offset << STACK_ALLOC_ALIGN;
69866e15 AP	249	struct stack_record *stack;
	250
	251	*entries = NULL;
4d4712c1 IK	252	if (!handle)
	253	return 0;
	254
69866e15 AP	255	if (parts.slabindex > depot_index) {
	256	WARN(1, "slab index %d out of bounds (%d) for stack id %08x\n",
	257	parts.slabindex, depot_index, handle);
	258	return 0;
	259	}
	260	slab = stack_slabs[parts.slabindex];
	261	if (!slab)
	262	return 0;
	263	stack = slab + offset;
cd11016e	264
c0cfc337 TG	265	*entries = stack->entries;
	266	return stack->size;
	267	}
	268	EXPORT_SYMBOL_GPL(stack_depot_fetch);
	269
cd11016e	270	/**
11ac25c6	271	* __stack_depot_save - Save a stack trace from an array
c0cfc337 TG	272	*
	273	* @entries: Pointer to storage array
	274	* @nr_entries: Size of the storage array
	275	* @alloc_flags: Allocation gfp flags
11ac25c6 ME	276	* @can_alloc: Allocate stack slabs (increased chance of failure if false)
	277	*
	278	* Saves a stack trace from @entries array of size @nr_entries. If @can_alloc is
	279	* %true, is allowed to replenish the stack slab pool in case no space is left
	280	* (allocates using GFP flags of @alloc_flags). If @can_alloc is %false, avoids
	281	* any allocations and will fail if no space is left to store the stack trace.
	282	*
	283	* Context: Any context, but setting @can_alloc to %false is required if
	284	* alloc_pages() cannot be used from the current context. Currently
	285	* this is the case from contexts where neither %GFP_ATOMIC nor
	286	* %GFP_NOWAIT can be used (NMI, raw_spin_lock).
cd11016e	287	*
11ac25c6	288	* Return: The handle of the stack struct stored in depot, 0 on failure.
cd11016e	289	*/
11ac25c6 ME	290	depot_stack_handle_t __stack_depot_save(unsigned long *entries,
	291	unsigned int nr_entries,
	292	gfp_t alloc_flags, bool can_alloc)
cd11016e	293	{
cd11016e	294	struct stack_record found = NULL, *bucket;
c0cfc337	295	depot_stack_handle_t retval = 0;
cd11016e AP	296	struct page *page = NULL;
cd11016e AP	297	void *prealloc = NULL;
c0cfc337 TG	298	unsigned long flags;
c0cfc337 TG	299	u32 hash;
cd11016e	300
e1fdc403	301	if (unlikely(nr_entries == 0) \|\| stack_depot_disable)
cd11016e AP	302	goto fast_exit;
cd11016e AP	303
c0cfc337	304	hash = hash_stack(entries, nr_entries);
cd11016e AP	305	bucket = &stack_table[hash & STACK_HASH_MASK];
	306
	307	/*
	308	* Fast path: look the stack trace up without locking.
	309	* The smp_load_acquire() here pairs with smp_store_release() to
	310	* \|bucket\| below.
	311	*/
c0cfc337 TG	312	found = find_stack(smp_load_acquire(bucket), entries,
c0cfc337 TG	313	nr_entries, hash);
cd11016e AP	314	if (found)
	315	goto exit;
	316
	317	/*
	318	* Check if the current or the next stack slab need to be initialized.
	319	* If so, allocate the memory - we won't be able to do that under the
	320	* lock.
	321	*
	322	* The smp_load_acquire() here pairs with smp_store_release() to
	323	* \|next_slab_inited\| in depot_alloc_stack() and init_stack_slab().
	324	*/
11ac25c6	325	if (unlikely(can_alloc && !smp_load_acquire(&next_slab_inited))) {
cd11016e AP	326	/*
	327	* Zero out zone modifiers, as we don't have specific zone
	328	* requirements. Keep the flags related to allocation in atomic
	329	* contexts and I/O.
	330	*/
	331	alloc_flags &= ~GFP_ZONEMASK;
	332	alloc_flags &= (GFP_ATOMIC \| GFP_KERNEL);
87cc271d	333	alloc_flags \|= __GFP_NOWARN;
cd11016e AP	334	page = alloc_pages(alloc_flags, STACK_ALLOC_ORDER);
	335	if (page)
	336	prealloc = page_address(page);
	337	}
	338
78564b94	339	raw_spin_lock_irqsave(&depot_lock, flags);
cd11016e	340
c0cfc337	341	found = find_stack(*bucket, entries, nr_entries, hash);
cd11016e	342	if (!found) {
7f2b8818 ME	343	struct stack_record *new = depot_alloc_stack(entries, nr_entries, hash, &prealloc);
7f2b8818 ME	344
cd11016e AP	345	if (new) {
	346	new->next = *bucket;
	347	/*
	348	* This smp_store_release() pairs with
	349	* smp_load_acquire() from \|bucket\| above.
	350	*/
	351	smp_store_release(bucket, new);
	352	found = new;
	353	}
	354	} else if (prealloc) {
	355	/*
	356	* We didn't need to store this stack trace, but let's keep
	357	* the preallocated memory for the future.
	358	*/
	359	WARN_ON(!init_stack_slab(&prealloc));
	360	}
	361
78564b94	362	raw_spin_unlock_irqrestore(&depot_lock, flags);
cd11016e AP	363	exit:
	364	if (prealloc) {
	365	/* Nobody used this memory, ok to free it. */
	366	free_pages((unsigned long)prealloc, STACK_ALLOC_ORDER);
	367	}
	368	if (found)
	369	retval = found->handle.handle;
	370	fast_exit:
	371	return retval;
	372	}
11ac25c6 ME	373	EXPORT_SYMBOL_GPL(__stack_depot_save);
	374
	375	/**
	376	* stack_depot_save - Save a stack trace from an array
	377	*
	378	* @entries: Pointer to storage array
	379	* @nr_entries: Size of the storage array
	380	* @alloc_flags: Allocation gfp flags
	381	*
	382	* Context: Contexts where allocations via alloc_pages() are allowed.
	383	* See __stack_depot_save() for more details.
	384	*
	385	* Return: The handle of the stack struct stored in depot, 0 on failure.
	386	*/
	387	depot_stack_handle_t stack_depot_save(unsigned long *entries,
	388	unsigned int nr_entries,
	389	gfp_t alloc_flags)
	390	{
	391	return __stack_depot_save(entries, nr_entries, alloc_flags, true);
	392	}
c0cfc337	393	EXPORT_SYMBOL_GPL(stack_depot_save);