FreeRTOS 中的 heap 5 內存管理,相對於 heap 4《FreeRTOS --(5)內存管理 heap4》 只增加了對非連續內存區域的管理,什麼叫非連續區域內存呢?比如一款芯片,它即支持了內部的 RAM,也支持了外掛 RAM,那麼這兩個內存就可能在地址上不是連續的,比如 RAM1、RAM2、RAM3,如下所示:
針對這種情況,就可以使用 heap 5 來管理;
不同於之前的 heap 管理,heap 5 引入了一個結構體來管理這些非連續的區域:
typedef struct HeapRegion
{
/* The start address of a block of memory that will be part of the heap.*/
uint8_t *pucStartAddress;
/* The size of the block of memory in bytes. */
size_t xSizeInBytes;
} HeapRegion_t;
一個塊連續的內存用一個 HeapRegion_t 表示,多個連續內存通過 HeapRegion_t 數組的形式組織成爲了所有的內存;
heap 5 要求,在調用正式的內存分配函數之前,需要定義 HeapRegion,並調用 vPortDefineHeapRegions 來初始化它;
官方給了一個 demo:
/* Define the start address and size of the three RAM regions. */
#define RAM1_START_ADDRESS ( ( uint8_t * ) 0x00010000 )
#define RAM1_SIZE ( 65 * 1024 )
#define RAM2_START_ADDRESS ( ( uint8_t * ) 0x00020000 )
#define RAM2_SIZE ( 32 * 1024 )
#define RAM3_START_ADDRESS ( ( uint8_t * ) 0x00030000 )
#define RAM3_SIZE ( 32 * 1024 )
/* Create an array of HeapRegion_t definitions, with an index for each of the three
RAM regions, and terminating the array with a NULL address. The HeapRegion_t
structures must appear in start address order, with the structure that contains the
lowest start address appearing first. */
const HeapRegion_t xHeapRegions[] =
{
{ RAM1_START_ADDRESS, RAM1_SIZE },
{ RAM2_START_ADDRESS, RAM2_SIZE },
{ RAM3_START_ADDRESS, RAM3_SIZE },
{ NULL, 0 } /* Marks the end of the array. */
};
int main( void ) {
/* Initialize heap_5. */
vPortDefineHeapRegions( xHeapRegions );
/* Add application code here. */
}
如果有 3 個 RAM 區域的話,那麼這樣去定義他們;
需要注意的幾點是:
1、定義 HeapRegion_t 數組的時候,最後一定要定義成爲 NULL 和 0,這樣接口才知道這是終點;
2、被定義的 RAM 區域,都會去參與內存管理;
那麼問題就來了,在真實使用的時候,有可能你很難去定義部分 RAM 的 Start Address!
比如,一款芯片,它告訴你,它的第一塊 RAM 有 64KB,第二塊 RAM 有 32KB,第三塊 RAM 有 32KB,那麼你顯然不能夠直接將這些內存信息,按照上面 demo 代碼的形式,定義到這個表格中,因爲在編譯階段,有可能你相關的代碼和數據等等(.text,.data,.bss,等)都會佔用一部分的 RAM,但凡是定義到這個 HeapRegion_t 數組表格的,都會參與內存管理的行爲,這顯然是我們不願意的;
也就是說,比如你芯片的 RAM 起始地址 0x2000_0000,你編譯你的 Source Code 後,相關的代碼和數據要佔用 20KB,也就是 0x5000;那麼你定義的 RAM1_START_ADDRESS 起始地址,就必須要大於 0x2000_0000+0x5000,這樣纔不會踩到你的其他數據;但是呢?你總不可能每次編譯完都去改你的這個表吧?這是很痛苦的事情;
所有考慮到實際的使用,官方給出參考 demo 的方法是:
/* Define the start address and size of the two RAM regions not used by the
linker. */
#define RAM2_START_ADDRESS ( ( uint8_t * ) 0x00020000 )
#define RAM2_SIZE ( 32 * 1024 )
#define RAM3_START_ADDRESS ( ( uint8_t * ) 0x00030000 )
#define RAM3_SIZE ( 32 * 1024 )
/* Declare an array that will be part of the heap used by heap_5. The array will be
placed in RAM1 by the linker. */
#define RAM1_HEAP_SIZE ( 30 * 1024 )
static uint8_t ucHeap[ RAM1_HEAP_SIZE ];
/* Create an array of HeapRegion_t definitions. Whereas in Listing 6 the first entry
described all of RAM1, so heap_5 will have used all of RAM1, this time the first
entry only describes the ucHeap array, so heap_5 will only use the part of RAM1 that
contains the ucHeap array. The HeapRegion_t structures must still appear in start
address order, with the structure that contains the lowest start address appearing
first. */
const HeapRegion_t xHeapRegions[] =
{
{ ucHeap, RAM1_HEAP_SIZE },
{ RAM2_START_ADDRESS, RAM2_SIZE },
{ RAM3_START_ADDRESS, RAM3_SIZE },
{ NULL, 0 } /* Marks the end of the array. */
};
即,將鏈接的代碼數據,根據鏈接器(Linker)配置後,這些都放置在第一段的區域,ucHeap 也放在一樣的地方,這樣就避免去根據 map 文件去硬編碼這個表格;
通過 beyond compare 可以知道,heap 5 和 heap 4 的代碼在分配內存的 pvPortMalloc,和釋放內存的 vPortFree,以及插入節點合併空閒內存 prvInsertBlockIntoFreeList 的部分,幾乎完全一樣,唯一不一樣的地方在於:
heap 4 的內存初始化用的是 prvHeapInit
heap 5 的內存初始化用的是 vPortDefineHeapRegions
那我們就來看看這個 vPortDefineHeapRegions 的實現:
void vPortDefineHeapRegions( const HeapRegion_t * const pxHeapRegions )
{
BlockLink_t *pxFirstFreeBlockInRegion = NULL, *pxPreviousFreeBlock;
size_t xAlignedHeap;
size_t xTotalRegionSize, xTotalHeapSize = 0;
BaseType_t xDefinedRegions = 0;
size_t xAddress;
const HeapRegion_t *pxHeapRegion;
/* Can only call once! */
configASSERT( pxEnd == NULL );
pxHeapRegion = &( pxHeapRegions[ xDefinedRegions ] );
while( pxHeapRegion->xSizeInBytes > 0 )
{
xTotalRegionSize = pxHeapRegion->xSizeInBytes;
/* Ensure the heap region starts on a correctly aligned boundary. */
xAddress = ( size_t ) pxHeapRegion->pucStartAddress;
if( ( xAddress & portBYTE_ALIGNMENT_MASK ) != 0 )
{
xAddress += ( portBYTE_ALIGNMENT - 1 );
xAddress &= ~portBYTE_ALIGNMENT_MASK;
/* Adjust the size for the bytes lost to alignment. */
xTotalRegionSize -= xAddress - ( size_t ) pxHeapRegion->pucStartAddress;
}
xAlignedHeap = xAddress;
/* Set xStart if it has not already been set. */
if( xDefinedRegions == 0 )
{
/* xStart is used to hold a pointer to the first item in the list of
free blocks. The void cast is used to prevent compiler warnings. */
xStart.pxNextFreeBlock = ( BlockLink_t * ) xAlignedHeap;
xStart.xBlockSize = ( size_t ) 0;
}
else
{
/* Should only get here if one region has already been added to the
heap. */
configASSERT( pxEnd != NULL );
/* Check blocks are passed in with increasing start addresses. */
configASSERT( xAddress > ( size_t ) pxEnd );
}
/* Remember the location of the end marker in the previous region, if
any. */
pxPreviousFreeBlock = pxEnd;
/* pxEnd is used to mark the end of the list of free blocks and is
inserted at the end of the region space. */
xAddress = xAlignedHeap + xTotalRegionSize;
xAddress -= xHeapStructSize;
xAddress &= ~portBYTE_ALIGNMENT_MASK;
pxEnd = ( BlockLink_t * ) xAddress;
pxEnd->xBlockSize = 0;
pxEnd->pxNextFreeBlock = NULL;
/* To start with there is a single free block in this region that is
sized to take up the entire heap region minus the space taken by the
free block structure. */
pxFirstFreeBlockInRegion = ( BlockLink_t * ) xAlignedHeap;
pxFirstFreeBlockInRegion->xBlockSize = xAddress - ( size_t ) pxFirstFreeBlockInRegion;
pxFirstFreeBlockInRegion->pxNextFreeBlock = pxEnd;
/* If this is not the first region that makes up the entire heap space
then link the previous region to this region. */
if( pxPreviousFreeBlock != NULL )
{
pxPreviousFreeBlock->pxNextFreeBlock = pxFirstFreeBlockInRegion;
}
xTotalHeapSize += pxFirstFreeBlockInRegion->xBlockSize;
/* Move onto the next HeapRegion_t structure. */
xDefinedRegions++;
pxHeapRegion = &( pxHeapRegions[ xDefinedRegions ] );
}
xMinimumEverFreeBytesRemaining = xTotalHeapSize;
xFreeBytesRemaining = xTotalHeapSize;
/* Check something was actually defined before it is accessed. */
configASSERT( xTotalHeapSize );
/* Work out the position of the top bit in a size_t variable. */
xBlockAllocatedBit = ( ( size_t ) 1 ) << ( ( sizeof( size_t ) * heapBITS_PER_BYTE ) - 1 );
}
經過一些對齊操作,將 demo 中的 3 塊內存通過鏈表的方式掛接起來了,只不過內存地址不連續而已,但是對於連續的內存地址中,照樣在釋放的時候,可以進行合併操作;