Further testing shows the "return without updating the array" issue is not related to multi-threading and not related to load. I added a catch exception to my program to make it a soft-failure. It didn't happen in an 80 million transfers test, but then did happen overnight in a single thread, one-access-every-few-seconds test.Did you see any SPI collision issues or a xfer2 return without updating the result array with the Pi 5?
I had a strange SPI soft failure that I believe might have been from 2 simultaneous Python spidev xfer2() requests in 64-bit PiOS Bookworm on a Pi5 4GB. One of the programs caught what is claimed was “No SPI Response” in a 16-bit transfer using:
With the "soft-failure" coding, I can live with this SPI on RPi5 issue, so I'm not going to investigate further.
Statistics: Posted by amcdonley — Mon Apr 01, 2024 3:03 pm