STM32F407 4GB NAND Flash ECC computation while read and write are different, but the data is the same.

[hypnotriod]

This is the code I'm using to read the page of 16 data chunks of 256 bytes each:

#define K9xxG08_ECC_Enable()			FSMC_NAND_DEVICE->PCR2 |= FSMC_PCR2_ECCEN
#define K9xxG08_ECC_Disable()		        FSMC_NAND_DEVICE->PCR2 &= ~FSMC_PCR2_ECCEN
#define K9xxG08_FIFO_Wait()			while ((FSMC_NAND_DEVICE->SR2 & FSMC_SR2_FEMPT) != FSMC_SR2_FEMPT){};
#define K9xxG08_ECC_Read()			(FSMC_NAND_DEVICE->ECCR2 & K9xxG08_ECC_MASK)
 
void K9xxG08_ReadPage(uint8_t *pBuffer, uint32_t pageAddress)
{
	uint32_t i;
	uint32_t endAddr;
	uint32_t ecc;
	uint32_t eccBuff[K9LBG08_ECC_CHUNKS_NUM];
 
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | CMD_AREA))  = K9xxG08_CMD_READ0; 
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_1st_CYCLE(K9xxG08_PageSize);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_2nd_CYCLE(K9xxG08_PageSize);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_1st_CYCLE(pageAddress);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_2nd_CYCLE(pageAddress);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_3rd_CYCLE(pageAddress);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | CMD_AREA))  = K9xxG08_CMD_READ1;
	
	for (i = 0; i != K9xxG08_EccChunksNum; i++)
	{
		((uint8_t *)&eccBuff[i])[0] = (*(__IO uint8_t *)K9xxG08_FSMC_NAND_DEVICE);
		((uint8_t *)&eccBuff[i])[1] = (*(__IO uint8_t *)K9xxG08_FSMC_NAND_DEVICE);
		((uint8_t *)&eccBuff[i])[2] = (*(__IO uint8_t *)K9xxG08_FSMC_NAND_DEVICE);
		((uint8_t *)&eccBuff[i])[3] = 0;
	}
	
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | CMD_AREA))  = K9xxG08_CMD_READ0; 
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = 0;
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = 0;
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_1st_CYCLE(pageAddress);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_2nd_CYCLE(pageAddress);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_3rd_CYCLE(pageAddress);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | CMD_AREA))  = K9xxG08_CMD_READ1;
	
	for (i = 0; i != K9xxG08_EccChunksNum; i++)
	{
		endAddr = ((uint32_t)pBuffer) + K9xxG08_ECC_CHUNK_SIZE;
		
		K9xxG08_ECC_Enable();
		
		while( ((uint32_t)pBuffer) != endAddr)
			*(uint8_t *)pBuffer++ = *(__IO uint8_t *)K9xxG08_FSMC_NAND_DEVICE;
 
		K9xxG08_FIFO_Wait();
		ecc = K9xxG08_ECC_Read();
		K9xxG08_ECC_Disable();
		
		if (ecc != eccBuff[i])
		{
			performErrorCorrection(ecc, eccBuff[i], pBuffer - K9xxG08_ECC_CHUNK_SIZE);
		}
	}
}

The issue is that I'm getting wrong ecc computed value for 30% of data chunks read. (But the data read is ok)

In other side when I'm writing data chunks to the page, I get 100% correct ecc computed for each chunk:

uint8_t K9xxG08_WritePage(uint8_t *pBuffer, uint32_t pageAddress)
{
	uint32_t i;
        uint32_t status = 0x00;
	uint32_t endAddr;
	uint32_t ecc;
	uint8_t *pEccArea;
  
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | CMD_AREA))  = K9xxG08_CMD_WRITE0;
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = 0;
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = 0;
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_1st_CYCLE(pageAddress);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_2nd_CYCLE(pageAddress);
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | ADDR_AREA)) = K9xxG08_3rd_CYCLE(pageAddress);
	
	pEccArea = pBuffer + K9xxG08_PageSize;
	for (i = 0; i != K9xxG08_EccChunksNum; i++)
	{
		endAddr = ((uint32_t)pBuffer) + K9xxG08_ECC_CHUNK_SIZE;
		
		K9xxG08_ECC_Enable();
		
		while( ((uint32_t)pBuffer) != endAddr)
			*(__IO uint8_t *)K9xxG08_FSMC_NAND_DEVICE = *(uint8_t *)pBuffer++;
		
		K9xxG08_FIFO_Wait();
		ecc = K9xxG08_ECC_Read();
		K9xxG08_ECC_Disable();
		
		*pEccArea++ = ((uint8_t *)&ecc)[0];
		*pEccArea++ = ((uint8_t *)&ecc)[1];
		*pEccArea++ = ((uint8_t *)&ecc)[2];
	}
	
	endAddr = ((uint32_t)pBuffer) + K9xxG08_SpareAreaSize;
	
	while( ((uint32_t)pBuffer) != endAddr)
		*(__IO uint8_t *)K9xxG08_FSMC_NAND_DEVICE = *(uint8_t *)pBuffer++;
 
	*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | CMD_AREA)) = K9xxG08_CMD_WRITE1;
	
	while((status & K9xxG08_STATUS_READY) != K9xxG08_STATUS_READY)
	{
		*(__IO uint8_t *)((uint32_t)(K9xxG08_FSMC_NAND_DEVICE | CMD_AREA)) = K9xxG08_CMD_STATUS;
		status = *(__IO uint8_t *)K9xxG08_FSMC_NAND_DEVICE;
	}
	
	return (status & K9xxG08_STATUS_FAIL);
}

So what is wrong with the algorithm? Or should I know something I don't know?