Mpfs improve isr and error handling

[jlaitine]

Summary

This re-writes the ISR handling of can driver to improve real-time response and avoid lock-ups.

Also RX data buffer handling is made safe, the frame consistency is checked on every read from the RX buffer.

Impact

Fixes hangs

Testing

On-desk trials

[jlaitine]

Now I guess it is ready for review

[jlaitine]

Some more testing, looks promising. The input from @vnopanen is that the drone has gone through flight test, and running on table without crashes/jams.

On desk another unit w. saluki-pi has gone through 16 hours of uav communication:
saluki> uptime
15:56:58 up 15:56, load average: 0.00, 0.00, 0.00
saluki> uavcan status
Pool allocator status:
Capacity hard/soft: 500/250 blocks
Reserved: 26 blocks
Allocated: 18 blocks

UAVCAN node status:
Internal failures: 0
Transfer errors: 0
RX transfers: 1378305
TX transfers: 22184190

CAN1 status:
HW errors: 0
IO errors: 0
RX frames: 1837892
TX frames: 22126786
CAN2 status:
HW errors: 0
IO errors: 0
RX frames: 1837892
TX frames: 22126786

ESC outputs:
INFO [mixer_module] Param prefix: UAVCAN_EC
Channel Configuration:
Channel 0: func: 101, value: 65535, failsafe: 65535, disarmed: 65535, min: 1, max: 8191
Channel 1: func: 102, value: 65535, failsafe: 65535, disarmed: 65535, min: 1, max: 8191
Channel 2: func: 103, value: 65535, failsafe: 65535, disarmed: 65535, min: 1, max: 8191
Channel 3: func: 104, value: 65535, failsafe: 65535, disarmed: 65535, min: 1, max: 8191
Channel 4: func: 0, value: 65535, failsafe: 65535, disarmed: 65535, min: 1, max: 8191
Channel 5: func: 0, value: 65535, failsafe: 65535, disarmed: 65535, min: 1, max: 8191
Channel 6: func: 0, value: 65535, failsafe: 65535, disarmed: 65535, min: 1, max: 8191
Channel 7: func: 0, value: 65535, failsafe: 65535, disarmed: 65535, min: 1, max: 8191
Servo outputs:
INFO [mixer_module] Param prefix: UAVCAN_SV
Channel Configuration:
Channel 0: func: 0, value: 0, failsafe: 500, disarmed: 500, min: 0, max: 1000
Channel 1: func: 0, value: 0, failsafe: 500, disarmed: 500, min: 0, max: 1000
Channel 2: func: 0, value: 0, failsafe: 500, disarmed: 500, min: 0, max: 1000
Channel 3: func: 0, value: 0, failsafe: 500, disarmed: 500, min: 0, max: 1000
Channel 4: func: 0, value: 0, failsafe: 500, disarmed: 500, min: 0, max: 1000
Channel 5: func: 0, value: 0, failsafe: 500, disarmed: 500, min: 0, max: 1000
Channel 6: func: 0, value: 0, failsafe: 500, disarmed: 500, min: 0, max: 1000
Channel 7: func: 0, value: 0, failsafe: 500, disarmed: 500, min: 0, max: 1000

Online nodes (Node ID, Health, Mode):
120 OK OPERAT
121 OK OPERAT
122 OK OPERAT
123 OK OPERAT

uavcan: cycle time: 50490729 events, 3433106204us elapsed, 67.99us avg, min 14us max 62608us 215.078us rms
uavcan: cycle interval: 50490729 events, 1137.46us avg, min 18us max 180918us 693.081us rms

[haitomatic]

good simplification for the interrupt handling routine 👍 . I didnt find any questionable changes that could affect the previous functionality. Only one test that we need to conduct is https://github.com/tiiuae/mpfs_canfd_uavcan_example/tree/canfd_driver_test

it tests diff bitrate and classic / canfd payloads. the UAVCAN test only uses classic CAN.

[jlaitine]

Merging this now, as it is fixing the rl issues found. Need to still run the tests @haitomatic pointed out, and if any issues are found, open new jira tickets of of those

[haitomatic]

yes, @vnopanen is conducting that test. Will open jira ticket if any issue found