SPI驱动使用直接调用内核函数是否合适呢？

硬件平台：GD32F4xx

在使用SPI驱动时，发现应用层调用SPI配置，其实还是直接调用spi_core.c内的函数，而不是通过通用的read write control接口

且如果采用目前的read write接口不适用，因为通常的SPI通讯都是发送完一个地址之后读取数据，但是现在的read和write都是单独的纯写或者纯读

rt_size_t rt_spi_transfer(struct rt_spi_device *device,

const void *send_buf,

void *recv_buf,

rt_size_t length)

{

rt_err_t result;

struct rt_spi_message message;

RT_ASSERT(device != RT_NULL);

RT_ASSERT(device->bus != RT_NULL);

result = rt_mutex_take(&(device->bus->lock), RT_WAITING_FOREVER);

if (result == RT_EOK)

{

if (device->bus->owner != device)

{

/* not the same owner as current, re-configure SPI bus */

result = device->bus->ops->configure(device, &device->config);

if (result == RT_EOK)

{

/* set SPI bus owner */

device->bus->owner = device;

}

else

{

/* configure SPI bus failed */

rt_set_errno(-RT_EIO);

result = 0;

goto __exit;

}

}

/* initial message */

message.send_buf = send_buf;

message.recv_buf = recv_buf;

message.length = length;

message.cs_take = 1;

message.cs_release = 1;

message.next = RT_NULL;

/* transfer message */

result = device->bus->ops->xfer(device, &message);

if (result == 0)

{

rt_set_errno(-RT_EIO);

goto __exit;

}

}

else

{

rt_set_errno(-RT_EIO);

return 0;

}

__exit:

rt_mutex_release(&(device->bus->lock));

return result;

}

个人觉得可以修改spi_dev.c和spi_core.c来完善此驱动，大家觉得有必要吗？