stm32或gd32移植libcanard实现UAVCAN协议

一、源码下载

1、git下载

点击我下载

2、csdn下载

自己上传的点击下载

二、源码移植

我自己是使用rt-thread操作系统移植的。但是不局限与操作系统,裸机也可以。

1、首先将源码加入到工程

在这里插入图片描述

2、分别实现一个内存的分配与释放函数,他是一个指针函数,原型为typedef void* (*CanardMemoryAllocate)(CanardInstance* ins, size_t amount);

static void* mem_allocate(CanardInstance* const canard, const size_t amount)
{
    
    
    (void) canard;
    return rt_malloc(amount);
}
static void mem_free(CanardInstance* const canard, void* const pointer)
{
    
    
    (void) canard;
    rt_free(pointer);
}

3、初始化canard

void slave_comm_init()
{
    
    
	canard = canardInit(&mem_allocate, &mem_free);
	canard.node_id = savePara.id; 
	txQueue = canardTxInit(1536,  CANARD_MTU_CAN_CLASSIC);  // Set MTU = 64 bytes. There is also CANARD_MTU_CAN_CLASSIC.	
}	

canard.node_id 设置本机id
canardTxInit(1536, CANARD_MTU_CAN_CLASSIC); 初始化发送队列,1536为大小。CANARD_MTU_CAN_CLASSIC表示使用的是普通的can,数据最大为8个字节,CANARD_MTU_CAN_FD表示使用的是can fd。

3、实现发送函数

void slave_comm_tx_process()
{
    
    
	for (const CanardTxQueueItem* ti = NULL; (ti = canardTxPeek(&txQueue)) != NULL;)  // Peek at the top of the queue.
	{
    
    
		if ((0U == ti->tx_deadline_usec) || (ti->tx_deadline_usec > rt_tick_get_millisecond()*1000))  // Check the deadline.
		{
    
    
			if (!slave_send_ext(ti->frame.extended_can_id,(void *)ti->frame.payload,ti->frame.payload_size))               // Send the frame over this redundant CAN iface.
			{
    
    
				break;                             // If the driver is busy, break and retry later.
			}
		}

		// After the frame is transmitted or if it has timed out while waiting, pop it from the queue and deallocate:
		canard.memory_free(&canard, canardTxPop(&txQueue, ti));
	}		
}

canard协议将发送的包处理完后会写入到队列中,canardTxPeek(&txQueue))从队列中取出数据,slave_send_ext(ti->frame.extended_can_id,(void *)ti->frame.payload,ti->frame.payload_size))为硬件发送函数,调用can发送。注意:UAVCAN使用的是扩展帧id
硬件发送函数为:

rt_inline uint8_t slave_send_ext(uint32_t id,uint8_t *sendBuf,uint8_t len)
{
    
    
	struct rt_can_msg txMsg = {
    
    0};
	
	txMsg.id = 	id;
	txMsg.ide = RT_CAN_EXTID;
	txMsg.rtr = RT_CAN_DTR;
	txMsg.len = len;
	for(rt_uint8_t i=0;i<len;i++)
	{
    
    
		txMsg.data[i] = sendBuf[i];
	}
	
	return rt_device_write(slaveDev,0,&txMsg,sizeof(txMsg));
	
}

RT_CAN_EXTID表示使用扩展帧

4、实现can硬件接收处理函数

void slave_comm_rx_process()
{
    
    
	CanardRxTransfer transfer;
	CanardFrame receivedFrame;
	struct rt_can_msg canRxMsg = {
    
    0};
	uint32_t rxTimestampUsec;
	int8_t result;
	
	while(rt_mq_recv(&slave_rec_msgq,&canRxMsg,sizeof(canRxMsg),RT_WAITING_NO) == RT_EOK)
	{
    
    
	
		receivedFrame.extended_can_id = canRxMsg.id;
		receivedFrame.payload_size = canRxMsg.len;
		receivedFrame.payload = canRxMsg.data;
		
		rxTimestampUsec = rt_tick_get_millisecond()*1000;
		
		result = canardRxAccept(&canard,
								rxTimestampUsec,          // When the frame was received, in microseconds.
								&receivedFrame,            // The CAN frame received from the bus.
								0,  // If the transport is not redundant, use 0.
								&transfer,
								NULL);
		if (result < 0)
		{
    
    
			// An error has occurred: either an argument is invalid or we've ran out of memory.
			// It is possible to statically prove that an out-of-memory will never occur for a given application if
			// the heap is sized correctly; for background, refer to the Robson's Proof and the documentation for O1Heap.
			// Reception of an invalid frame is NOT an error.
		}
		else if (result == 1)
		{
    
    
			void process_received_transfer(const uint8_t index,CanardRxTransfer* const transfer);
			process_received_transfer(0, &transfer);  // A transfer has been received, process it.
			canard.memory_free(&canard, transfer.payload);                  // Deallocate the dynamic memory afterwards.
		}
		else
		{
    
    
			// Nothing to do.
			// The received frame is either invalid or it's a non-last frame of a multi-frame transfer.
			// Reception of an invalid frame is NOT reported as an error because it is not an error.
		}
	}	
}

实现方法是,在can的中断中接收到数据放入slave_rec_msgq队列,

	struct rt_can_msg rxMsg = {
    
    0};
	rt_device_read(slaveDev,0,&rxMsg,sizeof(rxMsg));
	rt_mq_send(&slave_rec_msgq, &rxMsg, sizeof(rxMsg));	

然后协议从队列中读取数据处理。
将can数据转为canard支持的数据类型。

		receivedFrame.extended_can_id = canRxMsg.id;
		receivedFrame.payload_size = canRxMsg.len;
		receivedFrame.payload = canRxMsg.data;

当协议接收到完整的一帧数据后返回result等于1,自己处理接收到的数据。

process_received_transfer(0, &transfer);  // A transfer has been received, process it.

实现为:

void process_received_transfer(const uint8_t index,CanardRxTransfer* const transfer)
{
    
    
	LOG_D("slave rec id:%d size:%d",transfer->metadata.remote_node_id,transfer->payload_size);
	if(transfer->metadata.remote_node_id == canard.node_id)
	{
    
    
		slavePackDef *p = (slavePackDef *)transfer->payload;
		recCmd = p->packCmd;
		
	}
		
}

数据保存在transfer->payload中。执行完毕后释放内存:canard.memory_free(&canard, transfer.payload);

5、订阅消息

在canard协议中消息有三种类型,分别是: CanardTransferKindMessage``CanardTransferKindResponseCanardTransferKindRequest
区别在于:
CanardTransferKindMessage :广播,从发布者到所有订阅者。
CanardTransferKindResponse:点对点,从服务器到客户端。
CanardTransferKindRequest:点对点,从客户端到服务器。

一般来说从机为服务端,主机为客户端。

void slave_control_init()
{
    
    
	(void) canardRxSubscribe(&canard,   // Subscribe to an arbitrary service response.
							 CanardTransferKindResponse,  // Specify that we want service responses, not requests.
							 SLAVE_RESPONSE_PORT_ID,       // The Service-ID whose responses we will receive.
							 1536,      // The extent (see above).
							 CANARD_DEFAULT_TRANSFER_ID_TIMEOUT_USEC,
							 &responseSubscription);
	
}

以上订阅了一个CanardTransferKindResponse类型的消息,Service-ID为SLAVE_RESPONSE_PORT_ID

#define SLAVE_RESPONSE_PORT_ID 123

函数原型为:

int8_t canardRxSubscribe(CanardInstance* const       ins,
                         const CanardTransferKind    transfer_kind,
                         const CanardPortID          port_id,
                         const size_t                extent,
                         const CanardMicrosecond     transfer_id_timeout_usec,
                         CanardRxSubscription* const out_subscription);

参数解析:
ins:canard的一个实例,就是上面初始化的那个变量。
transfer_kind:消息类型,上面说的三个。
port_id:消息id。
extent:定义了传输有效载荷存储器缓冲器的大小。
transfer_id_timeout_usec:默认传输ID超时值定义。
out_subscription:如果已根据请求创建新订阅,则返回值为1。
如果在调用函数时存在此类订阅,则返回值为0。在这种情况下,终止现有订阅,然后在其位置创建新订阅。挂起的传输可能会丢失。
如果任何输入参数无效,则返回值为否定的无效参数错误。

三、应用层数据发送

static void send_data()
{
    
    
	static uint8_t messageTransferId = 0; 
	const CanardTransferMetadata transferMetadata = {
    
    
		.priority       = CanardPriorityNominal,
		.transfer_kind  = CanardTransferKindResponse,
		.port_id        = SLAVE_RESPONSE_PORT_ID,     
		.remote_node_id = id,      
		.transfer_id    = messageTransferId,
	};
	
	uint8_t sendBuf[100];
	for(uint8_t i=0;i<sizeof(sendBuf);i++)
	{
    
    
		sendBuf[i] = i;
	}
	
	
	++messageTransferId; transmission on this subject.
	int32_t result = canardTxPush(&txQueue,              
								  &canard,
								  0,   
								  &transferMetadata,
								  sizeof(sendBuf),                  
								  sendBuf);
	if (result < 0)
	{
    
    
		LOG_W("slave cmd send failed!");
	}	
}

需要注意的是:

	const CanardTransferMetadata transferMetadata = {
    
    
		.priority       = CanardPriorityNominal,
		.transfer_kind  = CanardTransferKindResponse,
		.port_id        = SLAVE_RESPONSE_PORT_ID,     // This is the subject-ID.
		.remote_node_id = id,       // Messages cannot be unicast, so use UNSET.
		.transfer_id    = messageTransferId,
	};

transfer_kind 需要和上面订阅的消息类型一样才能接收成功。
messageTransferId:每次发送都需要自加1。
port_id:也需要和订阅消息port_id一样。

四、移植成功源码下载

点击我下载

猜你喜欢

转载自blog.csdn.net/qq_15181569/article/details/131415387