堆内存管理算法的实现
头文件
/* * MemoryManage.h * * Created on: 2020年2月22日 * Author: LuYonglei */ #ifndef SRC_MEMORYMANAGE_H_ #define SRC_MEMORYMANAGE_H_ #include <stdlib.h> #include <stdint.h> #include "task.h" /* 字节对齐数 字节对齐掩码 * 8 0x0007 * 4 0x0003 * 2 0x0001 * 1 0x0000 */ #define BYTE_ALIGNMENT 8 //字节对齐数 #define BYTE_ALIGNMENT_MASK 0x0007 //字节对齐掩码 #define TOTAL_HEAP_SIZE 1000 //堆大小 #define MIN_BLOCK_SIZE 32 //最小内存块大小 #define HEAP_BITS_PER_BYTE 8 //堆中每个字节拥有的位数 //定义内存控制块 typedef struct memory_control_block { struct memory_control_block *nextUsableBlock; //下一个空闲块的地址 size_t blockSize; //空闲块的大小 } MCB; void* los_malloc(size_t wantedSize); //动态分配一块内存，内存大小为 wantedSize Bytes void los_free(void *addressToBeFree); //释放一块动态分配了的内存 size_t los_get_usable_heap_size(void); //获取当前未分配的内存堆大小 size_t los_get_ever_min_usable_heap_size(void); //获取未分配的内存堆的历史最小值 //以下函数为调试系统所用，实际使用中可删除 void OS_MemoryUsableInfo(void); //获取当前系统内存可分配情况 #endif /* SRC_MEMORYMANAGE_H_ */
源代码文件
/* * MemoryManage.c * * Created on: 2020年2月22日 * Author: LuYonglei */ #include "MemoryManage.h" static uint8_t Heap[TOTAL_HEAP_SIZE]; //编译器生成可用堆，堆的大小为TOTAL_HEAP_SIZE Bytes static MCB heapStart; //堆的起始 static MCB *heapEnd; //堆的末尾 static size_t usableBytesRemaining; //当前剩余可用字节数 static size_t usableEverMinRemaining; //历史最小剩余可用字节数 static size_t adjustedHeapSize; //经过对齐调整后的堆的大小 static size_t blockAllocatedBit; //用一个位来标记内存块是否被分配 static size_t MCB_size = sizeof(MCB); //内存控制块大小 static void los_heap_init(void); //系统堆初始化函数，只在第一次动态分配内存时调用 static void los_insert_block_into_usable_list(MCB *toBeInsertedBlock); //把一块内存插入空闲内存链表 void* los_malloc(size_t wantedSize) { //动态分配一块内存，内存大小为 wantedSize Bytes MCB *currentBlock; MCB *previousBlock; MCB *newBlock; void *addressReturn = NULL; los_suspend_all_tasks(); //调度器锁打开 { //如果是第一次分配内存就调用堆初始化函数 if (heapEnd == NULL) los_heap_init(); if (wantedSize > 0) { wantedSize += MCB_size; //如果需要进行内存对齐 if ((wantedSize & BYTE_ALIGNMENT_MASK) != 0) wantedSize = (wantedSize + BYTE_ALIGNMENT) & (~BYTE_ALIGNMENT_MASK); } //此处判断wantedSize是否超出限制 if ((wantedSize & blockAllocatedBit) == 0) { //若没有超出限制 if ((wantedSize > 0) && (wantedSize <= usableBytesRemaining)) { //如果申请的内存大小大于0,且系统剩余内存可以满足分配 previousBlock = &heapStart; currentBlock = heapStart.nextUsableBlock; while ((currentBlock->blockSize < wantedSize) && (currentBlock->nextUsableBlock != NULL)) { previousBlock = currentBlock; currentBlock = currentBlock->nextUsableBlock; } if (currentBlock != heapEnd) { addressReturn = (void*) ((size_t) (previousBlock->nextUsableBlock) + MCB_size); //将当前空闲内存块从内存空闲链表移除 previousBlock->nextUsableBlock = currentBlock->nextUsableBlock; if ((currentBlock->blockSize - wantedSize) > MIN_BLOCK_SIZE) { //如果这个内存块剩余的部分还可以再利用，就创建新的空闲内存块 newBlock = (MCB*) ((size_t) currentBlock + wantedSize); //更新新的内存块的大小 newBlock->blockSize = currentBlock->blockSize - wantedSize; //改变原来的内存块的大小 currentBlock->blockSize = wantedSize; //将新的空闲内存块插入到空闲内存块链表中去 los_insert_block_into_usable_list(newBlock); } //更新剩余内存总大小 usableBytesRemaining -= currentBlock->blockSize; //如果当前剩余内存总大小小于历史剩余内存总大小的最小值，就更新历史值 if (usableBytesRemaining < usableEverMinRemaining) usableEverMinRemaining = usableBytesRemaining; //此处将内存块的最高位设置为1,标记内存块已经被申请使用 currentBlock->blockSize |= blockAllocatedBit; currentBlock->nextUsableBlock = NULL; } } } else { //此处申请内存过大,超出size_t/2 } //此处做trace操作 } los_resume_all_tasks(); //调度器锁关闭 return addressReturn; } void los_free(void *addressToBeFree) { uint8_t *baseAddressToBeFree = (uint8_t*) addressToBeFree; MCB *blockToBeFree; //判断释放的地址是否为空 if (baseAddressToBeFree != NULL) { //经过偏移计算出内存控制块的基地址 baseAddressToBeFree -= MCB_size; blockToBeFree = (MCB*) baseAddressToBeFree; //判断内存快是否已经被分配使用，如果是就是放内存块 if ((blockToBeFree->blockSize & blockAllocatedBit) != 0) { if (blockToBeFree->nextUsableBlock == NULL) { //将内存块标识为空闲 blockToBeFree->blockSize &= ~blockAllocatedBit; los_suspend_all_tasks(); { usableBytesRemaining += blockToBeFree->blockSize; //此处做trace操作 los_insert_block_into_usable_list(blockToBeFree); } los_resume_all_tasks(); } } } } static void los_heap_init(void) { //系统堆的初始化函数 MCB *firstUsableBlock; uint8_t *alignedHeapBaseAddress; size_t totalHeapSize = TOTAL_HEAP_SIZE; //进行内存对齐操作 size_t newHeapBaseAdress = (size_t) Heap; //获取堆数祖的地址 if ((newHeapBaseAdress & BYTE_ALIGNMENT_MASK) != 0) { newHeapBaseAdress = (newHeapBaseAdress + BYTE_ALIGNMENT) & (~BYTE_ALIGNMENT_MASK); //内存对齐后,堆的大小发生变化 totalHeapSize -= newHeapBaseAdress - (size_t) Heap; } alignedHeapBaseAddress = (uint8_t*) newHeapBaseAdress; //初始化空闲链表头部 heapStart.nextUsableBlock = (MCB*) alignedHeapBaseAddress; heapStart.blockSize = (size_t) 0; //初始化链表尾部 newHeapBaseAdress = ((size_t) alignedHeapBaseAddress) + totalHeapSize; //获取堆尾部地址 newHeapBaseAdress -= MCB_size; newHeapBaseAdress &= ~((size_t) BYTE_ALIGNMENT_MASK); //完成heapEnd分配并进行内存对齐后的堆尾部地址 //初始化链表尾部 heapEnd = (MCB*) newHeapBaseAdress; heapEnd->nextUsableBlock = NULL; heapEnd->blockSize = (size_t) 0; //将当前所有内存插入空闲内存块链表 firstUsableBlock = (MCB*) alignedHeapBaseAddress; firstUsableBlock->blockSize = newHeapBaseAdress - (size_t) firstUsableBlock; firstUsableBlock->nextUsableBlock = heapEnd; //更新统计变量 adjustedHeapSize = firstUsableBlock->blockSize; //内存对齐后系统管理的堆的大小 usableEverMinRemaining = firstUsableBlock->blockSize; usableBytesRemaining = firstUsableBlock->blockSize; //内存块分配标志设置 blockAllocatedBit = ((size_t) 1) << ((sizeof(size_t) * HEAP_BITS_PER_BYTE) - 1); } static void los_insert_block_into_usable_list(MCB *toBeInsertedBlock) { //把一块内存插入空闲内存链表 MCB *blockIterator; uint8_t *tempAddress; //首先找到一个和toBeInsertedBlock相邻的前一个空闲内存 blockIterator = &heapStart; while ((blockIterator->nextUsableBlock) < toBeInsertedBlock) blockIterator = blockIterator->nextUsableBlock; tempAddress = (uint8_t*) blockIterator; //如果前一个内存的尾部恰好是toBeInsertedBlock的头部，那么这两块空闲内存是连续的，可以合并 if ((tempAddress + blockIterator->blockSize) == (uint8_t*) toBeInsertedBlock) { //将toBeInsertedBlock合并到blockIterator中 blockIterator->blockSize += toBeInsertedBlock->blockSize; toBeInsertedBlock = blockIterator; } //判断是否和后面的空闲内存相邻 tempAddress = (uint8_t*) toBeInsertedBlock; if ((tempAddress + toBeInsertedBlock->blockSize) == (uint8_t*) blockIterator->nextUsableBlock) { if (blockIterator->nextUsableBlock != heapEnd) { //如果后面的内存块的不是heapEnd //将后面的内存块合入toBeInsertedBlock toBeInsertedBlock->blockSize += blockIterator->nextUsableBlock->blockSize; toBeInsertedBlock->nextUsableBlock = blockIterator->nextUsableBlock->nextUsableBlock; } else { //如果后面的内存块的是heapEnd，则只改变指针 toBeInsertedBlock->nextUsableBlock = heapEnd; } } else { //如果和后面的内存块不相邻，就只能插入链表了 toBeInsertedBlock->nextUsableBlock = blockIterator->nextUsableBlock; } //判断前面是否已经合并，如果没有合并就要更新链表 if (toBeInsertedBlock != blockIterator) { blockIterator->nextUsableBlock = toBeInsertedBlock; } } size_t los_get_usable_heap_size(void) { return usableBytesRemaining; } size_t los_get_ever_min_usable_heap_size(void) { return usableEverMinRemaining; } void OS_MemoryUsableInfo(void) { MCB *start = &heapStart; MCB *end = heapEnd; start = start->nextUsableBlock; printf( "------------------------------------------------------------------\n"); printf("当前可用内存大小为：%d\n", usableBytesRemaining); printf("历史最小可用大小：%d\n", usableEverMinRemaining); while (start != end) { printf("空闲内存块地址：%p ，空闲内存块大小 %d ，空闲内存块结束地址：%p\n", (size_t) start, start->blockSize, ((size_t) start) + start->blockSize); start = start->nextUsableBlock; } printf( "------------------------------------------------------------------\n"); }
