fixes: busio, SPI OS error 5 for mimxrt10xx

[janderit]

I ran into an issue trying to connect an MCP2515 CAN bus controller via SPI to a Teensy 4.0 (using adafruit_mcp2515) and might have a simple solution to this and possibly related issues #3763 and #3055.

After a few CAN messages transferred, there is an exception with the following information.

common_hal_busio_spi_read: status 400
Traceback (most recent call last):
  File "code.py", line 49, in <module>
  File "code.py", line 44, in <module>
  File "/lib/adafruit_mcp2515/canio/__init__.py", line 102, in receive
  File "/lib/adafruit_mcp2515/__init__.py", line 355, in unread_message_count
  File "/lib/adafruit_mcp2515/__init__.py", line 421, in _read_from_rx_buffers
  File "/lib/adafruit_mcp2515/__init__.py", line 386, in _read_rx_buffer
  File "/lib/adafruit_mcp2515/__init__.py", line 386, in _read_rx_buffer
OSError: [Errno 5] Input/output error

It seems that SPI might occasionally be busy. The SPI drivers (e.g. MIMXRT10xx_SDK/.../MIMXRT1062/drivers/fsl_lpspi.c ) then return kStatus_LPSPI_Busy, which in busio mimxrt10xx/common-hal/busio/SPI.c translates into false result, thus resulting in OS Error 5.

Attempting a simple busy waiting loop in case of kStatus_LPSPI_Busy instead seems to do the trick. The problem seems to be resolved and I so far did not see more than a single additional iteration of the idle loop.

However, since I do not know the details of how SPI works here, I have no idea whether this simple solution is too simple. So if someone can asses this, please comment.

[tannewt]

Would you mind changing it to do a fixed number of retries? I'm worried about a hang as-is. Thanks for looking into this!

[janderit]

Would you mind changing it to do a fixed number of retries? I'm worried about a hang as-is. Thanks for looking into this!

Sure, will do.

[tannewt]

Thank you!