How to Work Around Xlnk being used in https://github.com/Xilinx/LSTM-PYNQ

Support
1

Hi, I am trying to use the PYNQ board for Optical Character Recognition (OCR). I found this git repo that does exactly what I am looking for; however, it is a few years old and is still using Xlnk.

So, when I try to clone the github, I get the following error in the terminal:

File "/tmp/pip-req-build-h3_p2379/lstm/lstm.py", line 36, in <module>  
  from pynq import Overlay, PL, Xlnk
ImportError: cannot import name 'Xlnk' from 'pynq' (/usr/local/lib/python3.8/dist-packages/pynq/\_\_init__.py)

Does anyone have any suggestions/tips on how to work around this?

I considered going back to SD image v2.6; however, I am having a difficult time finding the image file. I tried this site; however, when you click PYNQ-Z2 v2.6.0 SDCard image, nothing happens.

Thanks!

2

Depending on what gets used from Xlnk, some of that has been deprecated and now appears in the pynq library.

https://pynq.readthedocs.io/en/latest/changelog.html

  • Removed Xlnk allocator and libraries for edge devices (replaced by XRT allocation)
3

It uses:

 def cleanup(self):
        xlnk = Xlnk()
        xlnk.xlnk_reset()

I am unsure how to work around this using allocate. I know that I should use del buffer but I do not know of all the specific buffers I would need to delete.

4

right. xlnk_reset() resets all the CMA buffers allocated across the system.

Now I think you use “del” but you probably have to call it per-buffer.

So maybe
def cleanup(self):
del buffer_name1
del buffer_name2

as a brute force method

5

I came across this while looking into OCR implementations on PYNQ. It’s interesting to see how older projects can become difficult to run when libraries like Xlnk are deprecated. Updating the memory allocation and cleanup part seems like an important step when trying to keep an older OCR project working on a newer PYNQ setup.