https://toitlang.org/ logo
Powered by
Watchdogs & jag monitor
# help
a
Im playing around with watchdogs for one of the first times in toit. When the watchdog times out, jag monitor disconnects (probably fine) however I stop seeing any output from the ESP, when reconnecting, even though my container is still running / has restarted
Copy code
[watchdog] ERROR: watchdog too late {id: lightbug/otl/link}
[toit] INFO: entering deep sleep for 50ms
Error: Port has been closed
~/dev/lb/io/toit (main) jag monitor --attach
Starting serial monitor of port '/dev/ttyACM0' ...
What (if anything) am I doing wrong?
Oh intesting, when re installing the container, I then get this
Copy code
Starting serial monitor of port '/dev/ttyACM0' ...
(234138) task_wdt:  - sys_evt (CPU 0)
E (234138) task_wdt: Tasks currently running:
E (234138) task_wdt: CPU 0: IDLE
E (234138) task_wdt: Aborting.


Core  0 register dump:
MEPC    : 0x40811dfc  RA      : 0x4211bb34  SP      : 0x408303c0  GP      : 0x40823204
TP      : 0x40830400  T0      : 0x00000000  T1      : 0x40811b82  T2      : 0x00000000
S0/FP   : 0x4082c000  S1      : 0x40826674  A0      : 0x00000000  A1      : 0x00000000
A2      : 0x00000000  A3      : 0x00000000  A4      : 0x00000000  A5      : 0x00000000
A6      : 0x00000000  A7      : 0x00000000  S2      : 0x4082c000  S3      : 0x00000000
S4      : 0x00000000  S5      : 0x00000000  S6      : 0x00000000  S7      : 0x00000000
S8      : 0x00000000  S9      : 0x00000000  S10     : 0x00000000  S11     : 0x00000000
T3      : 0x00000000  T4      : 0x00000000  T5      : 0x00000000  T6      : 0x00000000
MSTATUS : 0x00000000  MTVEC   : 0x00000000  MCAUSE  : 0x00000000  MTVAL   : 0x00000000
MHARTID : 0x00000000
And the esp restarts
k
If you don't feed the watchdog in time, the system will restart.
But it is a little weird that you lose the serial monitoring in the process. Is this on a c6?
a
yup, on a c6!
It touches on another thing I wanted to look at in jag actually, such as making the watch command have a --reattach argument for example
unrelated, i'd love to add the concept of "wifi profiles" that can remember multiple creds for the flash command, so I dont have to change them up when im moving locations 😄
k
I think the serial monitoring only disconnects due to resets on some devices. I seem to remember that it is related to something about how much of the UART support is done in software versus hardware. I've had devices where very early crashes in the software stack made it incredibly hard to use the UART for flashing, because the software side of things would get interrupted by the crashes.
Not entirely sure about the c6, but I think I've struggled with an s3 for a few unfun days.
Is this because you can't keep feeding the watchdog from a container that has been shut down? And there is no cleaning up happening at shutdown that stops the watchdog?
a
potentially, I saw the advise that I should be running the watchdogs in a seperate container (which I am not currently doing) so likely I just need to use the watchdogs better
k
The high-level watchdog support is built on top of a low-level watchdog provided by ESP-IDF. We feed the low-level watchdog if the high-level watchdog are properly fed and we've noticed it. If the service provider runs along side other software it may be starved, which will lead to the low-level watchdog not being fed even though you've done your best to feed the high-level ones.
That is why we recommend running the low-level feeding provider separately, so the likelihood of that happening is low.
The loop that feeds the low-level watchdog - unless we've noticed a high-level watchdog that hasn't been fed - is here: https://github.com/toitware/toit-watchdog/blob/main/src/provider.toit#L111
a
re jag monitor, this is another one I get, where I just can't get it to close if something has gone wrong
Copy code
jag monitor
Starting serial monitor of port '/dev/ttyACM0' ...
^C^C
killing it even seems to take some crazy ammount of time
PreviousNext
Powered by