fdata引脚在Slave FIFO的时间点的数据损坏

你好，
我经历了从我发送数据到了终点，到达fdata引脚在Slave FIFO的时间点的数据损坏。让我解释一下我的设置。
我开始按照同步Slave FIFO流的例子。所以我有Altera MAX V CPLD作为主FIFO和FX2的配置为一个两端Slave FIFO。端点0x02为大批量的终点和0x86配置为中断和轮询间隔端点1。你可以看到这里的描述符HTTPS：/ /依据。GitHub。COM /一闻到/ 9390416 # file-dscr-asm-l95。
FX2被配置为自动提交两个端点。我利用CPLD提供外部时钟的FIFO的时钟。时钟正以10MHz的速度运行。你可以看到这里的HTTPS配置：/ /依据。GitHub。COM /一闻到/ 9390416 # file-fx2-c-l205。
我的申请是基于流的例子包含在SDK。排队几个异步传输在终点等待CPLD发送一个数据包。http：/ /依据。GitHub。COM /一闻到/ 9390416 # file-form-cs-l1081显示beginxfer呼叫和HTTPS：/ /依据。GitHub。COM /一闻到/ 9390416 # file-form-cs-l1156显示endxfer呼叫。这都是抄袭的例子并没有改变。
变化的analyzeframedata HTTPS做开始：/ /依据。GitHub。COM /一闻到/ 9390416 # file-form-cs-l1160。如果帧包含一个指定的命令我想发送一个数据包在了终点。我正在使用同步API来实现这一点。你可以在这里看到我的xferdata了HTTPS端点电话：/ /依据。GitHub。COM /一闻到/ 9390416 # file-form-cs-l1372。这就是数据破坏发生的地方。它不是一直发生的，而是随机的。我用我的逻辑分析仪闻fdata [7:0]引脚的FX2看看它给我的CPLD和通常有一个字节是不正确的。
有人看到了我的方式使用API吗？在这种情况下使用同步XFEDATA API是错误的吗？我应该为我的输出端点创建一个异步请求吗？
如有任何建议，将不胜感激。
谢谢，劳尔
P.S）我用一个中断端点，因为我需要保持尽可能低的延迟。我有1毫秒pkt_end CPLD发送信号的FX2和当我开始写数据到CPLD之间的时间下。



     以下为原文
  Hello,
    I'm experiencing some data corruption from the point I send data to the OUT endpoint and by the time it reaches the fdata pins on the slave FIFO. Let me explain my setup.
    I started by following the Synchronous Slave FIFO Streaming Example. So I have an Altera MAX V CPLD acting as a Master FIFO and the FX2 configured as a Slave FIFO with two endpoints. Endpoint 0x02 is configured as an OUT BULK end point and 0x86 is configured as an IN INTERRUPT endpoint with polling interval of 1. You can see the descriptor here https://gist.github.com/ismell/9390416#file-dscr-asm-L95.
    The FX2 is configured to auto commit both endpoints. I am using an external clock provided by the CPLD as the FIFO clock. The clock is running at 10MHz. You can see the configuration here https://gist.github.com/ismell/9390416#file-fx2-c-L205.
    My application is based off the Streaming Example included in the SDK. It queues up a few async transfers on the IN endpoint and waits for the CPLD to send a packet. https://gist.github.com/ismell/9390416#file-form-cs-L1081 shows the BeginXfer call and https://gist.github.com/ismell/9390416#file-form-cs-L1156 shows the EndXfer call. This is all copied from the example and has not been changed.
    The changes do start with analyzeFrameData at https://gist.github.com/ismell/9390416#file-form-cs-L1160. If the frame contains one of the specified commands I want to send a packet on the OUT endpoint. I'm using the synchronous API to do this. You can see my XferData call on the OUT endpoint here https://gist.github.com/ismell/9390416#file-form-cs-L1372. This is where the data corruption is happening. It doesn't happen all the time, but randomly. I have used my logic analyzer to sniff the fdata[7:0] pins on the FX2 to see what it is sending to my CPLD and there is usually one byte that is incorrect.
    Does anyone see anything wrong with the way I'm using the API? Is using the synchronous XferData API wrong in this case? Should I be creating an async request for my OUT endpoint?
    Any suggestions would be greatly appreciated.
    Thanks,
Raul
     
    p.s) I'm using an Interrupt IN endpoint because I need to keep latency as low as possible. I have under 1ms between the time the CPLD sends the pkt_end signal to the FX2 and when I have to start writing data back to the CPLD.