Fix: Determine if CAN TX is ongoing by reading from the TXBnCTRL register.

[aentinger]

Possible alternate solution for #54, without busy waiting.

@pepeRossRobotics can you please test? 🙏

[github-actions]

Memory usage change @ f66a9ec

Board	flash	%	RAM for global variables	%
arduino:mbed_edge:edge_control	0 - 0	0.0 - 0.0	0 - 0	0.0 - 0.0
arduino:mbed_nano:nano33ble	0 - 0	0.0 - 0.0	0 - 0	0.0 - 0.0
arduino:mbed_nano:nanorp2040connect	0 - 0	0.0 - 0.0	0 - 0	0.0 - 0.0
arduino:mbed_portenta:envie_m4	0 - 0	0.0 - 0.0	0 - 0	0.0 - 0.0
arduino:mbed_portenta:envie_m7	🔺 0 - +64	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkr1000	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkrfox1200	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkrgsm1400	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkrnb1500	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkrvidor4000	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkrwan1300	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkrwan1310	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkrwifi1010	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
arduino:samd:mkrzero	🔺 0 - +36	0.0 - +0.01	0 - 0	0.0 - 0.0
esp32:esp32:esp32	🔺 0 - +40	0.0 - 0.0	0 - 0	0.0 - 0.0
rp2040:rp2040:rpipico	🔺 0 - +32	0.0 - 0.0	0 - 0	0.0 - 0.0

Click for full report table

Board	examples/MCP2515-CAN-Sniffer flash	%	examples/MCP2515-CAN-Sniffer RAM for global variables	%	examples/MCP2515-Loopback flash	%	examples/MCP2515-Loopback RAM for global variables	%
arduino:mbed_edge:edge_control	0	0.0	0	0.0	0	0.0	0	0.0
arduino:mbed_nano:nano33ble	0	0.0	0	0.0	0	0.0	0	0.0
arduino:mbed_nano:nanorp2040connect	0	0.0	0	0.0	0	0.0	0	0.0
arduino:mbed_portenta:envie_m4	0	0.0	0	0.0	0	0.0	0	0.0
arduino:mbed_portenta:envie_m7	0	0.0	0	0.0	64	0.01	0	0.0
arduino:samd:mkr1000	0	0.0	0	0.0	36	0.01	0	0.0
arduino:samd:mkrfox1200	0	0.0	0	0.0	36	0.01	0	0.0
arduino:samd:mkrgsm1400	0	0.0	0	0.0	36	0.01	0	0.0
arduino:samd:mkrnb1500	0	0.0	0	0.0	36	0.01	0	0.0
arduino:samd:mkrvidor4000	0	0.0	0	0.0	36	0.01	0	0.0
arduino:samd:mkrwan1300	0	0.0	0	0.0	36	0.01	0	0.0
arduino:samd:mkrwan1310	0	0.0	0	0.0	36	0.01	0	0.0
arduino:samd:mkrwifi1010	0	0.0	0	0.0	36	0.01	0	0.0
arduino:samd:mkrzero	0	0.0	0	0.0	36	0.01	0	0.0
esp32:esp32:esp32	0	0.0	0	0.0	40	0.0	0	0.0
rp2040:rp2040:rpipico	0	0.0	0	0.0	32	0.0	0	0.0

Click for full report CSV

Board,examples/MCP2515-CAN-Sniffer<br>flash,%,examples/MCP2515-CAN-Sniffer<br>RAM for global variables,%,examples/MCP2515-Loopback<br>flash,%,examples/MCP2515-Loopback<br>RAM for global variables,%
arduino:mbed_edge:edge_control,0,0.0,0,0.0,0,0.0,0,0.0
arduino:mbed_nano:nano33ble,0,0.0,0,0.0,0,0.0,0,0.0
arduino:mbed_nano:nanorp2040connect,0,0.0,0,0.0,0,0.0,0,0.0
arduino:mbed_portenta:envie_m4,0,0.0,0,0.0,0,0.0,0,0.0
arduino:mbed_portenta:envie_m7,0,0.0,0,0.0,64,0.01,0,0.0
arduino:samd:mkr1000,0,0.0,0,0.0,36,0.01,0,0.0
arduino:samd:mkrfox1200,0,0.0,0,0.0,36,0.01,0,0.0
arduino:samd:mkrgsm1400,0,0.0,0,0.0,36,0.01,0,0.0
arduino:samd:mkrnb1500,0,0.0,0,0.0,36,0.01,0,0.0
arduino:samd:mkrvidor4000,0,0.0,0,0.0,36,0.01,0,0.0
arduino:samd:mkrwan1300,0,0.0,0,0.0,36,0.01,0,0.0
arduino:samd:mkrwan1310,0,0.0,0,0.0,36,0.01,0,0.0
arduino:samd:mkrwifi1010,0,0.0,0,0.0,36,0.01,0,0.0
arduino:samd:mkrzero,0,0.0,0,0.0,36,0.01,0,0.0
esp32:esp32:esp32,0,0.0,0,0.0,40,0.0,0,0.0
rp2040:rp2040:rpipico,0,0.0,0,0.0,32,0.0,0,0.0

[pepeRossRobotics]

This is with my home brew wait:

This is with this commit's code:

As you can see the amount of errors is around double (the up-time of both nodes was around 7 minutes), but the the modification of the performance of the code is orders of magnitude better than the master branch in that regard.

I think this fix performs very similar, but it is more elegant and efficient than my version, so happy to use it instead of mine.

What do you think?

[aentinger]

Thank you very much for the test. I'd prefer this variant to your PR, since it does not contain any busy-wait. If this is fine with you I'll merge this and close the other. As for the remaining errors ... possibly the CAN timings are too tight leading to frame errors due to out-of-margin bit-timings? Possibly also the SPI clock is too high for the MCP2515 resulting in bit errors during transmission? Those timing issues are hard to pin down ...

[pepeRossRobotics]

Agree with you. Thanks for taking the time to check this and come up with a better solution.