search

PX4 / PX4-Autopilot

PX4 Autopilot Software
https://px4.io
BSD 3-Clause "New" or "Revised" License
8.49k stars 13.51k forks source link

PixRacer and HereFlow UAVCAN driver #16334

Closed taileron closed 2 years ago

taileron commented 3 years ago

Describe the bug PixRacer doesn´t recognise UAVCAN devices like HereFlow various devices work perfect with X2.1-777 (e.g. HereFlow as range and flow on id 125)

To Reproduce Steps to reproduce the behavior: UAVCAN parameters successfully enabled dmesg enabled to display start procedure uavcan status displayed firstly I thought the "china" PixRacer has a pcb design issue but a new r15 one directly from MRO behaves identical the cirquit diagram shows CL and CH pins exchanged between X2.1 and PixRacer so I tried it to change them in the sensor

Expected behavior flow and range/data should appear in QGC

Log Files and Screenshots

nsh> uavcan status
Pool allocator status:
    Capacity hard/soft: 500/250 blocks
    Reserved:  30 blocks
    Allocated: 4 blocks

UAVCAN node status:
    Internal failures: 0
    Transfer errors:   0
    RX transfers:      0
    TX transfers:      1653

CAN1 status:
    HW errors: 795
    IO errors: 833
    RX frames: 87
    TX frames: 1892

control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Switched to rate_ctrl work queue: 0
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0
INFO  [mixer_module] Channel Configuration:
INFO  [mixer_module] Channel 0: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 1: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 2: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 3: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 4: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 5: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 6: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 7: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 8: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 9: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 10: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 11: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 12: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 13: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 14: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
INFO  [mixer_module] Channel 15: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191

Sensor 'baro':
devname: /dev/uavcan/baro
channel 0: empty
channel 1: empty
channel 2: empty
channel 3: empty

Sensor 'mag':
devname: /dev/uavcan/mag
channel 0: empty
channel 1: empty
channel 2: empty
channel 3: empty

Sensor 'gnss':
devname: /dev/gnss
channel 0: empty
channel 1: empty
channel 2: empty
channel 3: empty
uavcan: gnss: rtcm pub: 0 events, 0.00us avg, min 0us max 0us 0.000us rms

Sensor 'flow':
devname: /dev/flow
channel 0: empty
channel 1: empty
channel 2: empty
channel 3: empty

Sensor 'battery':
devname: /dev/battery
channel 0: empty
channel 1: empty
channel 2: empty
channel 3: empty

Sensor 'airspeed':
devname: /dev/airspeed
channel 0: empty
channel 1: empty
channel 2: empty
channel 3: empty

Sensor 'differential_pressure':
devname: /dev/airspeed
channel 0: empty
channel 1: empty
channel 2: empty
channel 3: empty

Sensor 'rangefinder':
devname: /dev/range_finder
channel 0: empty
channel 1: empty
channel 2: empty
channel 3: empty

Online nodes (Node ID, Health, Mode):

uavcan: cycle time: 24877 events, 706708us elapsed, 28.41us avg, min 14us max 11330us 106.940us rms
uavcan: cycle interval: 24877 events, 2769.66us avg, min 32us max 155828us 1326.074us rms

(here is GCC 7.2.1 but with GCC 9.3.1 it behaves identical)

nsh> dmesg
sercon: Registering CDC/ACM serial driver
sercon: Successfully registered the CDC/ACM serial driver
HW arch: PX4_FMU_V4
FW git-hash: 615333badb289aa77285bcfad071cc6a70b13632
FW version: 1.11.0 0 (17498112)
FW git-branch: master
OS: NuttX
OS version: Release 10.0.0 (167772415)
OS git-hash: 904a602c74dc08a100b5c2bd490807de19e73e10
Build datetime: Nov 20 2020 19:08:18
Build uri: localhost
Toolchain: GNU GCC, 7.2.1 20170904 (release) [ARM/embedded-7-branch revision 255204]
PX4GUID: 00010000000036313737333951050051002d
MCU: STM32F42x, rev. V
INFO  [param] selected parameter default file /fs/mtd_params
Board defaults: /etc/init.d/rc.board_defaults
INFO  [dataman] Unknown restart, data manager file '/fs/microsd/dataman' size is 362560 bytes
Board sensors: /etc/init.d/rc.board_sensors
ms5611 #0 on SPI bus 2 (devid=0x3d)
WARN  [SPI_I2C] hmc5883: no instance started (no device on bus?)
lis3mdl #0 on SPI bus 1 (devid=0x5)
icm20602 #0 on SPI bus 1 (devid=0x38)
mpu9250 #0 on SPI bus 1 (devid=0x24)
INFO  [uavcan] Node ID 1, bitrate 1000000
INFO  [uavcan] sensor bridge 'baro' init ok
INFO  [uavcan] sensor bridge 'mag' init ok
INFO  [uavcan] sensor bridge 'gnss' init ok
INFO  [uavcan] sensor bridge 'flow' init ok
INFO  [uavcan] sensor bridge 'battery' init ok
INFO  [uavcan] sensor bridge 'airspeed' init ok
INFO  [uavcan] sensor bridge 'differential_pressure' init ok
INFO  [uavcan] sensor bridge 'rangefinder' init ok
ERROR [uavcan] CentralizedServer init: -13
ERROR [uavcan] Node init failed: -13
ERROR [uavcan] Firmware Server Failed to Start -13
Board extras: /etc/init.d/rc.board_mavlink
INFO  [mavlink] mode: Config, data rate: 800000 B/s on /dev/ttyACM0 @ 57600B
INFO  [mavlink] mode: Normal, data rate: 20000 B/s on /dev/ttyS0 @ 921600B
Starting Main GPS on /dev/ttyS3
Starting MAVLink on /dev/ttyS1
INFO  [mavlink] mode: Normal, data rate: 20000 B/s on /dev/ttyS1 @ 921600B
INFO  [init] Mixer: /etc/mixers/vtol_convergence.main.mix on /dev/pwm_output0
ekf2 [581:100]
Board extras: /etc/init.d/rc.board_extras
INFO  [logger] logger started (mode=all)

NuttShell (NSH) NuttX-10.0.0

nsh> INFO  [ecl/EKF] reset position to last known position
INFO  [ecl/EKF] reset velocity to zero
INFO  [ecl/EKF] 13508753: EKF aligned, (baro hgt, IMU buf: 18, OBS buf: 14)
INFO  [mavlink] Starting mavlink shell
LorenzMeier commented 3 years ago

Do you have an update on this?

taileron commented 3 years ago

@LorenzMeier I was able to do some more research, the problem arises as soon as telemetry 1 or 2 is used ... I always had one of them in use, of course. The PixRacer only recognises UAVCAN when TEL 1 and TEL 2 are switched off. If one of the telemetry UARTS is activated, UAVCAN also blocks the nsh shell. ... and all hw pins are in the right order.

taileron commented 3 years ago

@LorenzMeier For me, the deactivation of the ESP 01, which can be plugged into the PixRacer, was the final solution (mavlink start in rc.board_mavlink instead of TTYS0 -> TTYS2). A single Mavlink instance is tolerated by UAVCAN and it does not matter whether the telemetry unit is connected or not. Maybe there are not enough resources to use everything at the same time.

LorenzMeier commented 3 years ago

This basically means that you are running out of RAM.

taileron commented 3 years ago

@LorenzMeier Unfortunately, another round is now coming ... with the (China) PixRacer, uavcan works when the autostart Mavlink instances are removed. Unfortunately, with the current MRO PixRacer, uavcan can no longer be made to work at all. In principle, only the sensors and the MCU version are different.

PixRacer (China)

PX4GUID: 0001000000003131373130365117004f0031
MCU: STM32F42x, rev. 3

ms5611 #0 on SPI bus 2 (devid=0x3d)
hmc5883 #0 on SPI bus 1 (devid=0x1)
icm20608g #0 on SPI bus 1 (devid=0x3a)
mpu9250 #0 on SPI bus 1 (devid=0x24)
INFO  [uavcan] Node ID 1, bitrate 1000000
uavcan_flow channel 124 class instance 0 ok
uavcan_rangefinder adding channel 124...
uavcan_rangefinder channel 124 class instance 0 ok

free
             total       used       free    largest
Umem:       223360     210688      12672      10816
Prog:      2097152    1835008     262144          0

PixRacer (MRO)

PX4GUID: 00010000000036313737333951050051002d
MCU: STM32F42x, rev. V

ms5611 #0 on SPI bus 2 (devid=0x3d)
lis3mdl #0 on SPI bus 1 (devid=0x5)
icm20602 #0 on SPI bus 1 (devid=0x38)
mpu9250 #0 on SPI bus 1 (devid=0x24)
INFO  [uavcan] Node ID 1, bitrate 1000000
ERROR [uavcan] CentralizedServer init: -13
ERROR [uavcan] Node init failed: -13
ERROR [uavcan] Firmware Server Failed to Start -13

free
             total       used       free    largest
Umem:       223360     195664      27696      24832
Prog:      2097152    1835008     262144          0
taileron commented 3 years ago

I found it after all - it was a small file error on the SD media and when I repaired it on the Mac, the usual index folders were created again. Checked the SD card in a Windows computer for file errors and repaired it and deleted the Mac index folders (which may contain file names longer than 32/40 characters). After that, UAVCAN worked.
Please remove lines 113 - 137 from RCS again, as they cause the real problem to be overlooked.

taileron commented 3 years ago

In the end, there is always just not enough free ram. Depending on the problems with the file system or my larger mixers. If it is reduced at some point, it works. However, something always has to be removed in the code, because only the Mavlink USB connection or the Mavlink shell is sufficient for nothing to work and in order to be able to use only USB for monitoring the UAVCAN components, many other things have to be removed in the code beforehand.

dagar commented 3 years ago

The uavcan driver/module size is increasing because it always allocates everything. We should add a parameter so that you can be more selective about which subscribers (sensors) and publishers (ESCs, beep, etc) are active. https://github.com/PX4/PX4-Autopilot/tree/master/src/drivers/uavcan/sensors

For mavlink USB autostart potentially helps in some situations. https://github.com/PX4/PX4-Autopilot/pull/16180 Without USB connected it saves a ton of memory by not running the mavlink USB instance or the lower level support (sercon, etc). Obviously you'll still get into trouble when trying to use both Mavlink USB + heavy UAVCAN simultaneously, but at least it makes real usage viable.

taileron commented 3 years ago

flow is only monitored by USB not by wifi telemetry If that were not the case, USB would not be necessary at all. Is it intentional that optical flow via telemetry is not monitored? UAVCAN_ENABLE = Sensors Autom Conf still blocks further functions like logging or pwm output so does´t left enough resources for the rest.

Would it help to start individual sensors manually e.g. for testing in the mvlnk console to save ram-space. How would be the complete syntax to start flow and rangefinder in this case (HereFlow): Sensor 'flow': name: uavcan_flow channel 0: node id 125 --> instance 0 Sensor 'rangefinder': name: uavcan_rangefinder channel 0: node id 125 --> instance 0

taileron commented 3 years ago

@dagar with only one mavlink instance I could do flight tests with UAVCAN and PixRacer. But then there is no way to get the logging simultaneously. With 4 k log buffer the log process doesn´t start at all, with 9 k it starts but always stops at some point and then the SD file system is no longer there. #17442 The buffer scheme of 1.12 also requires somewhat faster SD cards, the 1Gb that came with the PicRacers that were good enough for years definitely don't work anymore.

dagar commented 3 years ago

@dagar with only one mavlink instance I could do flight tests with UAVCAN and PixRacer.

We should make UAVCAN more configurable so that it's not handling everything by default. https://github.com/PX4/PX4-Autopilot/blob/d62204856cf0711af9ca82a399aac529783c412b/src/drivers/uavcan/sensors/sensor_bridge.cpp#L58-L78

How much memory does uavcan shrink save?

taileron commented 3 years ago

17533 solved all issues about losing the sd file system. The logging procedure itself now works again with all SD cards.

And there are now several successful ways to free up enough of the F4xx's little ram space. However, there remains an insurmountable problem, with 4K log buffer (reduced when UAVCAN is activated), the logger does not start at all: logger status INFO [logger] Running in mode: all INFO [logger] Number of subscriptions: 150 (4800 bytes) INFO [logger] Full File Logging Running: INFO [logger] Log file: /fs/microsd/log/2021-05-12/16_55_26.ulg INFO [logger] Wrote 17.17 KiB (avg 0.14 KiB/s) INFO [logger] Since last status: dropouts: 0 (max len: 0.000 s), max used buffer: 0 / 4396 B nsh> logger status INFO [logger] Running in mode: all INFO [logger] Number of subscriptions: 150 (4800 bytes) INFO [logger] Full File Logging Running: INFO [logger] Log file: /fs/microsd/log/2021-05-12/16_55_26.ulg INFO [logger] Wrote 17.17 KiB (avg 0.13 KiB/s) INFO [logger] Since last status: dropouts: 0 (max len: 0.000 s), max used buffer: 0 / 4396 B nsh>

It starts to work with the setting 5k and with 6k there are almost no dropouts: nsh> logger status INFO [logger] Running in mode: all INFO [logger] Number of subscriptions: 150 (4800 bytes) INFO [logger] Full File Logging Running: INFO [logger] Log file: /fs/microsd/log/2021-05-12/17_14_45.ulg INFO [logger] Wrote 0.48 MiB (avg 42.06 KiB/s) INFO [logger] Since last status: dropouts: 12 (max len: 0.061 s), max used buffer: 5068 / 5120 B

nsh> logger status INFO [logger] Running in mode: all INFO [logger] Number of subscriptions: 150 (4800 bytes) INFO [logger] Full File Logging Running: INFO [logger] Log file: /fs/microsd/log/2021-05-12/17_17_49.ulg INFO [logger] Wrote 0.38 MiB (avg 44.10 KiB/s) INFO [logger] Since last status: dropouts: 2 (max len: 0.006 s), max used buffer: 5730 / 6144 B

free ram with 6k log buffer and UAVCAN-HereFlow: free total used free largest Umem: 231440 211040 20400 17792 Prog: 2097152 1835008 262144 0

rc.board_arch_defaults should set the logger_buf to 5k or 6k for the case of enabled UAVCAN

taileron commented 3 years ago

https://review.px4.io/plot_app?log=8f164e5c-26b1-41d0-a0db-658fdab419ac

UAVCAN reduced (sensor bridge) to flow and range, log buffer set to 6K All uavcan battery has to be removed in sensor bridge manually as well. As soon as uavcan is enabled, the (analog) power module gives wrong values. With this conditions it is still needed to remove USB (serdis) to have enough ram. 2k console buffer is used for dmesg uart 1 and 3 tx buffer 1200 (some of my Picracers use ESP8266 at telemetry 2)

what could be the cause of UART 4 low on stack ?

taileron commented 3 years ago

very similar with: (USB mavlink disconnected / serdis - 2k console buffer) PixFalcon UAVCAN disabled (12k log buffer): https://review.px4.io/plot_app?log=40662abb-7ff9-49a9-9485-b2ccc3675e0a free total used free largest Umem: 230176 181952 48224 46752 Prog: 2097152 1966080 131072 0

PixFalcon UAVCAN enabled(2)shrinked to range+flow (6k log buffer): https://review.px4.io/plot_app?log=9f130349-3a83-42a1-a3c6-da1f72e2de68 free total used free largest Umem: 230176 217120 13056 12720 Prog: 2097152 1966080 131072 0

uavcan status Pool allocator status: Capacity hard/soft: 500/250 blocks Reserved: 151 blocks Allocated: 55 blocks

UAVCAN node status: Internal failures: 0 Transfer errors: 0 RX transfers: 0 TX transfers: 1440

CAN1 status: HW errors: 21402 IO errors: 22768 RX frames: 0 TX frames: 224 CAN2 status: HW errors: 21398 IO errors: 22764 RX frames: 0 TX frames: 224

control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms INFO [mixer_module] Switched to rate_ctrl work queue: 0 INFO [mixer_module] Mixer loaded: no INFO [mixer_module] Driver instance: 0 INFO [mixer_module] Channel Configuration: INFO [mixer_module] Channel 0: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 1: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 2: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 3: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 4: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 5: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 6: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 7: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 8: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 9: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 10: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 11: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 12: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 13: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 14: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191 INFO [mixer_module] Channel 15: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191

Sensor 'flow': name: uavcan_flow channel 0: empty channel 1: empty channel 2: empty channel 3: empty

Sensor 'rangefinder': name: uavcan_rangefinder channel 0: empty channel 1: empty channel 2: empty channel 3: empty

Online nodes (Node ID, Health, Mode):

uavcan: cycle time: 19504 events, 1418993us elapsed, 72.75us avg, min 15us max 6220us 159.212us rms uavcan: cycle interval: 19504 events, 2967.72us avg, min 44us max 14303us 717.062us rms

PixFalcon UAVCAN enabled(2) all sensors: free total used free largest Umem: 230176 219424 10752 10224 Prog: 2097152 1966080 131072 0

PixFalcon UAVCAN enabled(3)range+flow+actuators: free total used free largest Umem: 230176 220016 10160 9328 Prog: 2097152 1966080 131072 0

taileron commented 3 years ago

16180 brings by far the best result 25k free ram with all necessary modules active.

https://review.px4.io/plot_app?log=df3b8469-d0b6-42a1-ad26-29a6ddc9744f

taileron commented 3 years ago

16180 hereflow flight test

https://review.px4.io/plot_app?log=b10a8303-e03b-486d-9fa3-8adecaee8bfe

taileron commented 3 years ago

with #16180 merged this issue can be closed

taileron commented 3 years ago

16180 now merged

dagar commented 3 years ago

Thanks for the followup @taileron.

taileron commented 3 years ago

I have now tested the master again and it uses aprox 16K more ram with the same settings. Flying is no longer possible despite USB auto connect and deactivation of all not needed uavcan sensors. So far I have not been able to determine where the 16K has gone. free total used free largest nused nfree Umem: 226688 219488 7200 6912 947 18 Prog: 2097152 1835008 262144 0 23 2

dagar commented 3 years ago

It might be the new modules for multicopter autotune and magnetometer bias estimation.

dagar commented 3 years ago

I'll also have changes to make uavcan a bit more modular as well.

dagar commented 3 years ago

UAVCAN - https://github.com/PX4/PX4-Autopilot/pull/18471

I'm thinking we should change the F4 default to not run certain modules like mc_autotune_attitude_control by default.

taileron commented 3 years ago

pr-uavcan_parameters currently doesn't compile successful - but I think the idea for it is great

After sensor_bridge.cpp manual disabled unused sensors, reduce all uart buffers expect the wifi telemetry to 300 disabled autotune, osd, dshot, sd-stress it can be flown again

free

           total       used       free    largest  nused  nfree
        
Umem:     231584     216688      14896      14576    905     19
        
Prog:    2097152    1835008     262144          0     23      2

uavcan status
Pool allocator status:
    Capacity hard/soft: 500/250 blocks
    Reserved:  79 blocks
    Allocated: 13 blocks

UAVCAN node status:
    Internal failures: 0
    Transfer errors:   0
    RX transfers:      269
    TX transfers:      3360

CAN1 status:
    HW errors: 43147
    IO errors: 46141
    RX frames: 1034
    TX frames: 933

ESC outputs:
INFO  [mixer_module] Param prefix: UAVCAN_EC
control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0
Channel Configuration:
Channel 0: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 1: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 2: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 3: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 4: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 5: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 6: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 7: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191

Servo outputs:
INFO  [mixer_module] Param prefix: UAVCAN_SV
control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0

Channel Configuration:

Channel 0: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Channel 1: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Channel 2: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Channel 3: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Channel 4: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Channel 5: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Channel 6: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Channel 7: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0



Sensor 'flow':
name: uavcan_flow

channel 0: node id 125 --> instance 0



Sensor 'rangefinder':
name: uavcan_rangefinder

channel 0: node id 125 --> instance 0



Online nodes (Node ID, Health, Mode):
     125 OK         OPERAT    

uavcan: cycle time: 48247 events, 2907623us elapsed, 60.27us avg, min 13us max 4487us 147.270us rms
uavcan: cycle interval: 48247 events, 2900.32us avg, min 34us max 281373us 1646.370us rms
taileron commented 3 years ago

The amount of free ram tested with #18471 on a Spare PixRacer is identical to that bracketed in sensors-bridge.cpp, but the great thing about it is that it can be set by parameter in the respective FC, as a reduction does not necessarily have to take place with an F7/H7. I will also test this on the plane, as last time the standard Bat Mon had no more values.

dagar commented 3 years ago

There will be more to come, I'll break off the other builtin drivers/uavcan features to be enabled with UAVCAN_PUB_* parameters.

taileron commented 3 years ago

There could still be some hidden potential for further savings in Ram.

dagar commented 3 years ago

There could still be some hidden potential for further savings in Ram.

Yes there's quite a bit.

farzadsw commented 2 years ago

I have a similar problem with pixracer and uavcan (optical flow), the uavcan does not start at the boot (UAVCAN_ENABLE: sensor auto config) and when I start it manually it doesn't detect the hereflow sensor.

NuttShell (NSH) NuttX-10.2.0
nsh> uavcan status
ERROR [uavcan] application not running
nsh> free
                   total       used       free    largest  nused  nfree
        Umem:     227440     169872      57568      57424    721     10
        Prog:    2097152    1966080     131072          0     24      1
nsh> uavcan start fw
INFO  [uavcan] Node ID 1, bitrate 1000000
nsh> free
                   total       used       free    largest  nused  nfree
        Umem:     227440     185872      41568      40048    795     10
        Prog:    2097152    1966080     131072          0     24      1
nsh> uavcan status
Pool allocator status:
    Capacity hard/soft: 500/250 blocks
    Reserved:  47 blocks
    Allocated: 30 blocks

UAVCAN node status:
    Internal failures: 0
    Transfer errors:   0
    RX transfers:      0
    TX transfers:      228

CAN1 status:

ESC outputs:
INFO  [mixer_module] Param prefix: UAVCAN_EC
control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0
Channel Configuration:
Channel 0: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 1: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 2: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 3: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 4: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 5: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 6: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 7: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Servo outputs:
INFO  [mixer_module] Param prefix: UAVCAN_SV
control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0
Channel Configuration:
Channel 0: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 1: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 2: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 3: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 4: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 5: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 6: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 7: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Sensor 'gnss':
name: uavcan_gnss
uavcan: gnss: rtcm pub: 0 events, 0.00us avg, min 0us max 0us 0.000us rms

Sensor 'mag':
name: uavcan_mag

Online nodes (Node ID, Health, Mode):

uavcan: cycle time: 2949 events, 102759us elapsed, 34.85us avg, min 17us max 1743us 87.697us rms
uavcan: cycle interval: 2949 events, 2998.91us avg, min 1475us max 4557us 410.817us rms

my hw:

nsh> ver all
HW arch: PX4_FMU_V4
FW git-hash: ad70f5eef92671df35e9614d0574d30c114afb18
FW version: 1.13.0 0 (17629184)
FW git-branch: kame.v2
OS: NuttX
OS version: Release 8.2.0 (134349055)
OS git-hash: d4c06e9dfb7bb14d0f19ef47b94d838b1caf50f3
Build datetime: Dec  9 2021 00:00:41
Build uri: localhost
Build variant: default
Toolchain: GNU GCC, 6.3.1 20170620 (release) [ARM/embedded-6-branch revision 249437]
PX4GUID: 00010000000020343856434e5003004a0042
MCU: STM32F42x, rev. 5

Also, I already diabled mavlinks (except for USB) and some unused modules to free the memory (SD_STRESS, MC_AUTOTUNE_ATTITUDE_CONTROL, OSD, IRLOCK)

I'm relatively new to the code base, what am I missing?

taileron commented 2 years ago

In the PX4 version used, with setting "uavcan sensors automatic" the individual UAVCAN sensor modules can be explicitly selected in sub-parameters, as GPS and MAG (is the default) can be deactivated but FLOW and RANGE can be activated.

farzadsw commented 2 years ago

Thank you for the quick reply and the tip! I activated sub-parameters, but still, it doesn't start at boot and also when I run it manually, it has no data:

nsh> uavcan start fw
nsh> uavcan status
Pool allocator status:
    Capacity hard/soft: 500/250 blocks
    Reserved:  57 blocks
    Allocated: 24 blocks

UAVCAN node status:
    Internal failures: 0
    Transfer errors:   0
    RX transfers:      0
    TX transfers:      295

CAN1 status:

ESC outputs:
INFO  [mixer_module] Param prefix: UAVCAN_EC
control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0
Channel Configuration:
Channel 0: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 1: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 2: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 3: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 4: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 5: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 6: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 7: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Servo outputs:
INFO  [mixer_module] Param prefix: UAVCAN_SV
control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0
Channel Configuration:
Channel 0: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 1: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 2: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 3: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 4: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 5: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 6: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 7: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Sensor 'flow':
name: uavcan_flow

Sensor 'gnss':
name: uavcan_gnss
uavcan: gnss: rtcm pub: 0 events, 0.00us avg, min 0us max 0us 0.000us rms

Sensor 'mag':
name: uavcan_mag

Sensor 'rangefinder':
name: uavcan_rangefinder

Online nodes (Node ID, Health, Mode):

uavcan: cycle time: 3768 events, 277577us elapsed, 73.67us avg, min 17us max 105503us 1720.974us rms
uavcan: cycle interval: 3768 events, 3083.66us avg, min 364us max 121697us 3100.292us rms
nsh> free
                   total       used       free    largest  nused  nfree
        Umem:     227440     199616      27824      23024    815     12
        Prog:    2097152    1966080     131072          0     24      1

I also already tried formatting the SD card and swapping the H/L wires. (and also the sensor is fine, it works with cube-orange)

taileron commented 2 years ago

Sorry, I have only now been able to test it again, but the sensor starts to stop working when the ram is less than 4k free. If only Flow and Range are activated, UAVCAN (Sensors automatic config) consumes approx. 32k, a Mavlink instance (also the USB connection) approx. 15k. This (>51000) must be free beforehand. With "uavcan start fw / all" the sensors are not recognised here either, but approx. 17k less ram is required. I need it for VTOL and for me it was necessary to reduce some UART buffers in defconfig.

uavcan status Pool allocator status: Capacity hard/soft: 500/250 blocks Reserved: 27 blocks Allocated: 17 blocks

UAVCAN node status: Internal failures: 0 Transfer errors: 0 RX transfers: 1014 TX transfers: 1103

CAN1 status: HW errors: 432 IO errors: 432 RX frames: 3831 TX frames: 1309 . . Channel 7: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Sensor 'flow': name: uavcan_flow channel 0: node id 125 --> instance 0

Sensor 'rangefinder': name: uavcan_rangefinder channel 0: node id 125 --> instance 0

Online nodes (Node ID, Health, Mode): 125 OK OPERAT

uavcan: cycle time: 19138 events, 1406326us elapsed, 73.48us avg, min 14us max 9584us 203.516us rms uavcan: cycle interval: 19138 events, 2465.77us avg, min 32us max 29747us 1366.323us rms free total used free largest nused nfree Umem: 232736 222736 10000 5712 931 17 Prog: 2097152 1966080 131072 0 24 1

farzadsw commented 2 years ago

Thank you for the reply, Since I need Mavlink over USB, I tried to free up more ram by disabling some other modules (logger, camera,...). But still didn't manage to make it work:

NuttShell (NSH) NuttX-10.2.0
nsh> free
                   total       used       free    largest  nused  nfree
        Umem:     227840     174272      53568      52368    777     15
        Prog:    2097152    1835008     262144          0     23      2
nsh> uavcan status
ERROR [uavcan] application not running
nsh> uavcan start fw
INFO  [uavcan] Node ID 1, bitrate 1000000
nsh> uavcan status
Pool allocator status:
    Capacity hard/soft: 500/250 blocks
    Reserved:  44 blocks
    Allocated: 26 blocks

UAVCAN node status:
    Internal failures: 0
    Transfer errors:   0
    RX transfers:      0
    TX transfers:      131

CAN1 status:

ESC outputs:
INFO  [mixer_module] Param prefix: UAVCAN_EC
control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0
Channel Configuration:
Channel 0: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 1: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 2: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 3: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 4: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 5: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 6: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Channel 7: value: 0, failsafe: 0, disarmed: 65535, min: 1, max: 8191
Servo outputs:
INFO  [mixer_module] Param prefix: UAVCAN_SV
control latency: 0 events, 0us elapsed, 0.00us avg, min 0us max 0us 0.000us rms
INFO  [mixer_module] Mixer loaded: no
INFO  [mixer_module] Driver instance: 0
Channel Configuration:
Channel 0: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 1: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 2: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 3: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 4: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 5: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 6: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0
Channel 7: value: 0, failsafe: 0, disarmed: 0, min: 0, max: 0

Sensor 'flow':
name: uavcan_flow

Sensor 'rangefinder':
name: uavcan_rangefinder

Online nodes (Node ID, Health, Mode):

uavcan: cycle time: 1591 events, 53125us elapsed, 33.39us avg, min 16us max 1777us 81.354us rms
uavcan: cycle interval: 1591 events, 2997.98us avg, min 1218us max 4791us 450.188us rms
nsh> free
                   total       used       free    largest  nused  nfree
        Umem:     227840     191984      35856      35664    883     12
        Prog:    2097152    1835008     262144          0     23      2

I think it would be better for me to switch to another FC with more ram for now.

I will put my parameters here, just in case it is helpful:

x + UAVCAN_ENABLE [716,1608] : 2
x   UAVCAN_ESC_IDLT [717,1609] : 1
x   UAVCAN_LGT_ANTCL [718,1610] : 2
x   UAVCAN_LGT_LAND [719,1611] : 0
x   UAVCAN_LGT_NAV [720,1612] : 3
x   UAVCAN_LGT_STROB [721,1613] : 1
x   UAVCAN_NODE_ID [722,1614] : 1
x   UAVCAN_RNG_MAX [723,1615] : 200.0000
x   UAVCAN_RNG_MIN [724,1616] : 0.3000
x   UAVCAN_SUB_ASPD [725,1617] : 0
x   UAVCAN_SUB_BARO [726,1618] : 0
x   UAVCAN_SUB_BAT [727,1619] : 0
x   UAVCAN_SUB_DPRES [728,1620] : 0
x + UAVCAN_SUB_FLOW [729,1621] : 1
x + UAVCAN_SUB_GPS [730,1622] : 0
x   UAVCAN_SUB_ICE [731,1623] : 0
x   UAVCAN_SUB_IMU [732,1624] : 0
x + UAVCAN_SUB_MAG [733,1625] : 0
x + UAVCAN_SUB_RNG [734,1626] : 1
    UAVCAN_SV_DIS1 [-1,1627] : 500
    UAVCAN_SV_DIS2 [-1,1628] : 500
taileron commented 2 years ago

i think for this ram limited fc you should no longer expect to be able to do vtol and uavcan sensors at the same time. The PX4 code is growing and therefore consuming more and more resources, so I would close this now.