AXI-Stream Data FIFO仿真调试

1、 s_axis_tready出现异常情况

如图所示,一开始s_axis_tready是正常的,但是之后开始写入数据时,s_axis_tready没拉起来。

检查了一遍,然后经过测试发现,读数据端的时钟无效,然后造成s_axis_tready没有拉起来,把读数据的时钟换成正常时钟后,s_axis_tready也变正常了。

2、axis_wr_data_count计数问题

 在观察axis_wr_data_count计数时,我发现一开始的读数出现几次重复的0到1计数,百思不得其解,结果如下图所示。

 后来去做了一下常规FIFO IP核的仿真,发现结果也是一样

然后我翻了一下正点原子的文档,最后发现解析如下:

 

这是FIFO的First Word Fall Through模式,Xilinx官方文档PG057的解释如下:

The first word fall through maintains a similar behavior with embedded/interconnect register. The empty gives a latency of one more cycle as compared to selecting only one register, before rd_en signal is initiated. The next output is latched when rd_en initiates without any additional latency for BRAM as shown in Figure 3-22

 

以下是一个简单的AXI4-Stream FIFO写数据的VHDL实现,可以用于仿真或实际硬件实现: ```vhdl library IEEE; use IEEE.STD_LOGIC_1164.ALL; use IEEE.NUMERIC_STD.ALL; entity axi4_stream_fifo is generic ( DATA_WIDTH : integer := 32; -- 数据宽度 ADDR_WIDTH : integer := 6; -- 地址宽度 MAX_DEPTH : integer := 64 -- 最大深度 ); port ( -- AXI4-Stream接口 s_axis_tdata : in std_logic_vector(DATA_WIDTH-1 downto 0); s_axis_tvalid : in std_logic; s_axis_tready : out std_logic; -- AXI接口 s_axi_awaddr : in std_logic_vector(ADDR_WIDTH-1 downto 0); s_axi_awvalid : in std_logic; s_axi_awready : out std_logic; s_axi_wdata : in std_logic_vector(DATA_WIDTH-1 downto 0); s_axi_wvalid : in std_logic; s_axi_wready : out std_logic; s_axi_bresp : out std_logic_vector(1 downto 0); s_axi_bvalid : out std_logic; s_axi_bready : in std_logic ); end entity; architecture rtl of axi4_stream_fifo is -- FIFO深度计数器 signal count : integer range 0 to MAX_DEPTH-1 := 0; -- FIFO存储器 type fifo_mem_t is array (0 to MAX_DEPTH-1) of std_logic_vector(DATA_WIDTH-1 downto 0); signal fifo_mem : fifo_mem_t := (others => (others => '0')); begin -- AXI4-Stream接口写数据 write_data: process (s_axis_tdata, s_axis_tvalid, s_axis_tready) is begin if (s_axis_tvalid = '1' and s_axis_tready = '1') then -- 数据写入FIFO fifo_mem(count) <= s_axis_tdata; count <= count + 1; end if; end process; -- AXI接口响应写请求 respond_write: process (s_axi_awaddr, s_axi_awvalid, s_axi_awready, s_axi_wdata, s_axi_wvalid, s_axi_wready) is begin if (s_axi_awvalid = '1' and s_axi_awready = '1' and s_axi_wvalid = '1' and s_axi_wready = '1') then -- 写入FIFO的地址为当前深度 s_axi_awaddr <= std_logic_vector(to_unsigned(count-1, ADDR_WIDTH)); -- 写入数据 fifo_mem(count-1) <= s_axi_wdata; -- 计数器加1 count <= count + 1; -- 响应写请求 s_axi_bresp <= "00"; s_axi_bvalid <= '1'; end if; end process; -- AXI接口读请求 read_request: process (s_axi_awaddr, s_axi_awvalid, s_axi_awready) is begin if (s_axi_awvalid = '1' and s_axi_awready = '1') then -- 读请求的地址为0 s_axi_awaddr <= (others => '0'); -- 响应读请求 s_axi_bresp <= "00"; s_axi_bvalid <= '1'; end if; end process; -- AXI接口读数据 read_data: process (s_axi_araddr, s_axi_arvalid, s_axi_arready) is begin if (s_axi_arvalid = '1' and s_axi_arready = '1') then -- 读取FIFO的第一个数据 s_axi_rdata <= fifo_mem(0); -- 读取后计数器减1 count <= count - 1; -- 响应读请求 s_axi_rvalid <= '1'; end if; end process; end architecture; ``` 请注意,这只是一个简单的实现,没有考虑流控制和错误处理。在实际应用中,您可能需要更多的功能和保护。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值