5 fsmc 扩展外部SRAM，读写测试通过，使用 rt_malloc /rt_memheap_alloc 申请动态内存时死机

发布于 2024-11-18 14:54:24 浏览：362 订阅该版

如题，能够读写到0x68000000之后的内存情况，但是在申请外部sram的动态内存时就会溢出，日志显示如下
 
![screenshot_image.png](https://oss-club.rt-thread.org/uploads/20241118/81303a528c9fd21b28bc3e3c387b623e.png)

![screenshot_image.png](https://oss-club.rt-thread.org/uploads/20241118/324179efb71eec66fbcea2665a0de7fe.png.webp)
我的初始化配置如下
```C
static int rt_hw_sram_init(void)
{
    int result = RT_EOK;
    rt_pin_mode(VCC_SRAM_EN, PIN_MODE_OUTPUT);
    rt_pin_write(VCC_SRAM_EN, PIN_HIGH);

FSMC_NORSRAM_TimingTypeDef Timing = {0};

/** Perform the SRAM2 memory initialization sequence
    */
    hsram.Instance = FSMC_NORSRAM_DEVICE;
    hsram.Extended = FSMC_NORSRAM_EXTENDED_DEVICE;
    /* hsram.Init */
    hsram.Init.NSBank = FSMC_NORSRAM_BANK3;
    hsram.Init.DataAddressMux = FSMC_DATA_ADDRESS_MUX_DISABLE;
    hsram.Init.MemoryType = FSMC_MEMORY_TYPE_SRAM;
#if SRAM_DATA_WIDTH == 8
    hsram.Init.MemoryDataWidth = FSMC_NORSRAM_MEM_BUS_WIDTH_8;
#elif SRAM_DATA_WIDTH == 16
    hsram.Init.MemoryDataWidth = FSMC_NORSRAM_MEM_BUS_WIDTH_16;
#else
    hsram.Init.MemoryDataWidth = FSMC_NORSRAM_MEM_BUS_WIDTH_32;
#endif
    hsram.Init.BurstAccessMode = FSMC_BURST_ACCESS_MODE_DISABLE;
    hsram.Init.WaitSignalPolarity = FSMC_WAIT_SIGNAL_POLARITY_LOW;
    hsram.Init.WrapMode = FSMC_WRAP_MODE_DISABLE;
    hsram.Init.WaitSignalActive = FSMC_WAIT_TIMING_BEFORE_WS;
    hsram.Init.WriteOperation = FSMC_WRITE_OPERATION_ENABLE;
    hsram.Init.WaitSignal = FSMC_WAIT_SIGNAL_DISABLE;
    hsram.Init.ExtendedMode = FSMC_EXTENDED_MODE_DISABLE;
    hsram.Init.AsynchronousWait = FSMC_ASYNCHRONOUS_WAIT_DISABLE;
    hsram.Init.WriteBurst = FSMC_WRITE_BURST_DISABLE;
    hsram.Init.PageSize = FSMC_PAGE_SIZE_NONE;

/* Timing */
    Timing.AddressSetupTime = 0;
    Timing.AddressHoldTime = 0;
    Timing.DataSetupTime = 8;
    Timing.BusTurnAroundDuration = 0;
    Timing.CLKDivision = 0;
    Timing.DataLatency = 0;  //WJF 0->3
    Timing.AccessMode = FSMC_ACCESS_MODE_A;
    /* ExtTiming */

if (HAL_SRAM_Init(&hsram, &Timing, &Timing) != HAL_OK)
    {
        LOG_E("SRAM init failed!");
        result = -RT_ERROR;
    }
    else
    {
        LOG_D("sram init success, mapped at 0x%X, size is %d bytes, data width is %d", SRAM_BANK_ADDR, SRAM_SIZE, SRAM_DATA_WIDTH);
#ifdef RT_USING_MEMHEAP_AS_HEAP
        /* If RT_USING_MEMHEAP_AS_HEAP is enabled, SRAM is initialized to the heap */
       rt_memheap_init(&sram_heap, "sram", (void *)SRAM_BANK_ADDR, SRAM_SIZE);
       LOG_D("%s\n", sram_heap.parent.name);
#endif
    }

return result;
}
INIT_BOARD_EXPORT(rt_hw_sram_init);
```
在宏定义阶段添加了RT_USING_MEMHEAP_AS_HEAP 
![screenshot_image.png](https://oss-club.rt-thread.org/uploads/20241118/c3b6d19041621e6dc438d646969298b7.png.webp)
在初始化阶段有使用rt_memheap_init配置堆
```c
  rt_memheap_init(&sram_heap, "sram", (void *)SRAM_BANK_ADDR, SRAM_SIZE);
```
能够读到外挂的sram ，也能够堆这段内存进行读写操作
![screenshot_image.png](https://oss-club.rt-thread.org/uploads/20241118/51a3775f924c898bafad4c990208cd66.png)

![screenshot_Pasted image 20241115131912.png](https://oss-club.rt-thread.org/uploads/20241118/ebdf460a6a9a1f82e014648951d78333.png)
有和 @crystal266 的文章里所讲的，是在测试完读写后 重新复位，再进行申请动态内存的操作。
然而，当代码跑进rt_memheap_alloc中时，会直接溢出

![screenshot_image.png](https://oss-club.rt-thread.org/uploads/20241118/99e6c880cf5e88a229e5e7b34c59920f.png.webp)

求助 ，在这种情况下，该怎么继续排查