zephyr/kernel/mempool.c
Andy Ross 4c63af8434 mem_pool: Don't check level_empty() before breaking a block
This test was just wrong.  If the current thread did not race with any
others during the allocation process, then the result will be false
because it was detected so earlier in the function.  If we did race,
then sure: it might be true now if someone snuck in and freed a block.
But so what?  We already have the block we want to break.  The
behavior in the code as written was to early-exit from the break loop,
returning a buffer that was larger than the one requested (though
otherwise benign -- we wouldn't leak, just waste memory).  No idea
what I was thinking.

Thanks to Du Quanwen for the diagnosis.

Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
2017-07-31 09:14:59 -07:00

409 lines
9.5 KiB
C

/*
* Copyright (c) 2017 Intel Corporation
*
* SPDX-License-Identifier: Apache-2.0
*/
#include <kernel.h>
#include <ksched.h>
#include <wait_q.h>
#include <init.h>
#include <string.h>
/* Linker-defined symbols bound the static pool structs */
extern struct k_mem_pool _k_mem_pool_list_start[];
extern struct k_mem_pool _k_mem_pool_list_end[];
s64_t _tick_get(void);
static struct k_mem_pool *get_pool(int id)
{
return &_k_mem_pool_list_start[id];
}
static int pool_id(struct k_mem_pool *pool)
{
return pool - &_k_mem_pool_list_start[0];
}
static void *block_ptr(struct k_mem_pool *p, size_t lsz, int block)
{
return p->buf + lsz * block;
}
static int block_num(struct k_mem_pool *p, void *block, int sz)
{
return (block - p->buf) / sz;
}
static bool level_empty(struct k_mem_pool *p, int l)
{
return sys_dlist_is_empty(&p->levels[l].free_list);
}
/* Places a 32 bit output pointer in word, and an integer bit index
* within that word as the return value
*/
static int get_bit_ptr(struct k_mem_pool *p, int level, int bn, u32_t **word)
{
u32_t *bitarray = level <= p->max_inline_level ?
&p->levels[level].bits : p->levels[level].bits_p;
*word = &bitarray[bn / 32];
return bn & 0x1f;
}
static void set_free_bit(struct k_mem_pool *p, int level, int bn)
{
u32_t *word;
int bit = get_bit_ptr(p, level, bn, &word);
*word |= (1<<bit);
}
static void clear_free_bit(struct k_mem_pool *p, int level, int bn)
{
u32_t *word;
int bit = get_bit_ptr(p, level, bn, &word);
*word &= ~(1<<bit);
}
/* Returns all four of the free bits for the specified blocks
* "partners" in the bottom 4 bits of the return value
*/
static int partner_bits(struct k_mem_pool *p, int level, int bn)
{
u32_t *word;
int bit = get_bit_ptr(p, level, bn, &word);
return (*word >> (4*(bit / 4))) & 0xf;
}
static size_t buf_size(struct k_mem_pool *p)
{
return p->n_max * p->max_sz;
}
static bool block_fits(struct k_mem_pool *p, void *block, size_t bsz)
{
return (block + bsz - 1 - p->buf) < buf_size(p);
}
static void init_mem_pool(struct k_mem_pool *p)
{
int i;
size_t buflen = p->n_max * p->max_sz, sz = p->max_sz;
u32_t *bits = p->buf + buflen;
sys_dlist_init(&p->wait_q);
for (i = 0; i < p->n_levels; i++) {
int nblocks = buflen / sz;
sys_dlist_init(&p->levels[i].free_list);
if (nblocks < 32) {
p->max_inline_level = i;
} else {
p->levels[i].bits_p = bits;
bits += (nblocks + 31)/32;
}
sz = _ALIGN4(sz / 4);
}
for (i = 0; i < p->n_max; i++) {
void *block = block_ptr(p, p->max_sz, i);
sys_dlist_append(&p->levels[0].free_list, block);
set_free_bit(p, 0, i);
}
}
int init_static_pools(struct device *unused)
{
ARG_UNUSED(unused);
struct k_mem_pool *p;
for (p = _k_mem_pool_list_start; p < _k_mem_pool_list_end; p++) {
init_mem_pool(p);
}
return 0;
}
SYS_INIT(init_static_pools, PRE_KERNEL_1, CONFIG_KERNEL_INIT_PRIORITY_OBJECTS);
/* A note on synchronization: all manipulation of the actual pool data
* happens in one of alloc_block()/free_block() or break_block(). All
* of these transition between a state where the caller "holds" a
* block pointer that is marked used in the store and one where she
* doesn't (or else they will fail, e.g. if there isn't a free block).
* So that is the basic operation that needs synchronization, which we
* can do piecewise as needed in small one-block chunks to preserve
* latency. At most (in free_block) a single locked operation
* consists of four bit sets and dlist removals. If the overall
* allocation operation fails, we just free the block we have (putting
* a block back into the list cannot fail) and return failure.
*/
static void *alloc_block(struct k_mem_pool *p, int l, size_t lsz)
{
sys_dnode_t *block;
int key = irq_lock();
block = sys_dlist_get(&p->levels[l].free_list);
if (block) {
clear_free_bit(p, l, block_num(p, block, lsz));
}
irq_unlock(key);
return block;
}
static void free_block(struct k_mem_pool *p, int level, size_t *lsizes, int bn)
{
int i, key, lsz = lsizes[level];
void *block = block_ptr(p, lsz, bn);
key = irq_lock();
set_free_bit(p, level, bn);
if (level && partner_bits(p, level, bn) == 0xf) {
for (i = 0; i < 4; i++) {
int b = (bn & ~3) + i;
clear_free_bit(p, level, b);
if (b != bn &&
block_fits(p, block_ptr(p, lsz, b), lsz)) {
sys_dlist_remove(block_ptr(p, lsz, b));
}
}
irq_unlock(key);
free_block(p, level-1, lsizes, bn / 4); /* tail recursion! */
return;
}
if (block_fits(p, block, lsz)) {
sys_dlist_append(&p->levels[level].free_list, block);
}
irq_unlock(key);
}
/* Takes a block of a given level, splits it into four blocks of the
* next smaller level, puts three into the free list as in
* free_block() but without the need to check adjacent bits or
* recombine, and returns the remaining smaller block.
*/
static void *break_block(struct k_mem_pool *p, void *block,
int l, size_t *lsizes)
{
int i, bn, key;
key = irq_lock();
bn = block_num(p, block, lsizes[l]);
for (i = 1; i < 4; i++) {
int lbn = 4*bn + i;
int lsz = lsizes[l + 1];
void *block2 = (lsz * i) + (char *)block;
set_free_bit(p, l + 1, lbn);
if (block_fits(p, block2, lsz)) {
sys_dlist_append(&p->levels[l + 1].free_list, block2);
}
}
irq_unlock(key);
return block;
}
static int pool_alloc(struct k_mem_pool *p, struct k_mem_block *block,
size_t size)
{
size_t lsizes[p->n_levels];
int i, alloc_l = -1, free_l = -1, from_l;
void *blk = NULL;
/* Walk down through levels, finding the one from which we
* want to allocate and the smallest one with a free entry
* from which we can split an allocation if needed. Along the
* way, we populate an array of sizes for each level so we
* don't need to waste RAM storing it.
*/
lsizes[0] = _ALIGN4(p->max_sz);
for (i = 0; i < p->n_levels; i++) {
if (i > 0) {
lsizes[i] = _ALIGN4(lsizes[i-1] / 4);
}
if (lsizes[i] < size) {
break;
}
alloc_l = i;
if (!level_empty(p, i)) {
free_l = i;
}
}
if (alloc_l < 0 || free_l < 0) {
block->data = NULL;
return -ENOMEM;
}
/* Iteratively break the smallest enclosing block... */
blk = alloc_block(p, free_l, lsizes[free_l]);
if (!blk) {
/* This can happen if we race with another allocator.
* It's OK, just back out and the timeout code will
* retry. Note mild overloading: -EAGAIN isn't for
* propagation to the caller, it's to tell the loop in
* k_mem_pool_alloc() to try again synchronously. But
* it means exactly what it says.
*/
return -EAGAIN;
}
for (from_l = free_l; from_l < alloc_l; from_l++) {
blk = break_block(p, blk, from_l, lsizes);
}
/* ... until we have something to return */
block->data = blk;
block->id.pool = pool_id(p);
block->id.level = alloc_l;
block->id.block = block_num(p, block->data, lsizes[alloc_l]);
return 0;
}
int k_mem_pool_alloc(struct k_mem_pool *p, struct k_mem_block *block,
size_t size, s32_t timeout)
{
int ret, key;
s64_t end = 0;
__ASSERT(!(_is_in_isr() && timeout != K_NO_WAIT), "");
if (timeout > 0) {
end = _tick_get() + _ms_to_ticks(timeout);
}
while (1) {
ret = pool_alloc(p, block, size);
if (ret == 0 || timeout == K_NO_WAIT ||
ret == -EAGAIN || (ret && ret != -ENOMEM)) {
return ret;
}
key = irq_lock();
_pend_current_thread(&p->wait_q, timeout);
_Swap(key);
if (timeout != K_FOREVER) {
timeout = end - _tick_get();
if (timeout < 0) {
break;
}
}
}
return -EAGAIN;
}
void k_mem_pool_free(struct k_mem_block *block)
{
int i, key, need_sched = 0;
struct k_mem_pool *p = get_pool(block->id.pool);
size_t lsizes[p->n_levels];
/* As in k_mem_pool_alloc(), we build a table of level sizes
* to avoid having to store it in precious RAM bytes.
* Overhead here is somewhat higher because free_block()
* doesn't inherently need to traverse all the larger
* sublevels.
*/
lsizes[0] = _ALIGN4(p->max_sz);
for (i = 1; i <= block->id.level; i++) {
lsizes[i] = _ALIGN4(lsizes[i-1] / 4);
}
free_block(get_pool(block->id.pool), block->id.level,
lsizes, block->id.block);
/* Wake up anyone blocked on this pool and let them repeat
* their allocation attempts
*/
key = irq_lock();
while (!sys_dlist_is_empty(&p->wait_q)) {
struct k_thread *th = (void *)sys_dlist_peek_head(&p->wait_q);
_unpend_thread(th);
_abort_thread_timeout(th);
_ready_thread(th);
need_sched = 1;
}
if (need_sched && !_is_in_isr()) {
_reschedule_threads(key);
} else {
irq_unlock(key);
}
}
#if (CONFIG_HEAP_MEM_POOL_SIZE > 0)
/*
* Heap is defined using HEAP_MEM_POOL_SIZE configuration option.
*
* This module defines the heap memory pool and the _HEAP_MEM_POOL symbol
* that has the address of the associated memory pool struct.
*/
K_MEM_POOL_DEFINE(_heap_mem_pool, 64, CONFIG_HEAP_MEM_POOL_SIZE, 1, 4);
#define _HEAP_MEM_POOL (&_heap_mem_pool)
void *k_malloc(size_t size)
{
struct k_mem_block block;
/*
* get a block large enough to hold an initial (hidden) block
* descriptor, as well as the space the caller requested
*/
size += sizeof(struct k_mem_block);
if (k_mem_pool_alloc(_HEAP_MEM_POOL, &block, size, K_NO_WAIT) != 0) {
return NULL;
}
/* save the block descriptor info at the start of the actual block */
memcpy(block.data, &block, sizeof(struct k_mem_block));
/* return address of the user area part of the block to the caller */
return (char *)block.data + sizeof(struct k_mem_block);
}
void k_free(void *ptr)
{
if (ptr != NULL) {
/* point to hidden block descriptor at start of block */
ptr = (char *)ptr - sizeof(struct k_mem_block);
/* return block to the heap memory pool */
k_mem_pool_free(ptr);
}
}
#endif