Random Hard Fault resets on Duet 3 Mini WiFi

omtek

@dc42 Logged a memory difference message with the new build. Printer was idle. No reset as of yet.

*** Memory difference at line 2228 offset 12: original 0d0a0d39 copy 20032958, original changed, copy ok, fix=yes

Logged another memory difference message. Still no reset.

*** Memory difference at line 2228 offset 60: original 2001882c copy 0d0a0d39, original ok, copy changed, fix=no

Exerqtor

@dc42 Haven't had time to look at this before now (like I mentioned in the email), but do you have a debug version based on 3.6 as well?

dc42

@Exerqtor I'll prepare a 3.6 version with the debug in it today.

dc42

@Exerqtor I've put a 3.6.0-beta.1+debug build for the Duet 3 Mini at https://www.dropbox.com/scl/fo/01s6pikmcusa2e3vu0rkh/APdYxDkXDSIWqeQl0R__Gag?rlkey=iwd3zb2dk3ggjw7ihcy2whg5h&dl=0. The instructions are slightly different from the 3.5.3 version:

Install it
Preferably, connect a PC running a terminal emulator to the USB port, check it is communicating with the Duet, and execute M111 P8 S1 to enable debug messages
Leave it running with at least one DWC session connected
Report any "Mem diff" messages that are displayed on the terminal emulator
If you get any unexpected resets, post a M122 report

If you don't wish to connect a terminal emulator, then don't run M111 P8 S1 and it won't attempt to output the debug messages to USB. It will still log up to 4 memory difference reports internally, but only if the board doesn't reset. Running M122 will report these messages just before the "Storage" section of the report, and clear them.

dc42

@omtek thanks, that's useful information. The first memory difference would likely have resulted in a reset in previous firmware versions, but the memory check detected and fixed it. The second memory difference was harmless.

Exerqtor

@omtek said in Random Hard Fault resets on Duet 3 Mini WiFi:

@dc42 Logged a memory difference message with the new build. Printer was idle. No reset as of yet.

Out of curiosity, how long had you had the logging going before it threw these messages?

omtek

@Exerqtor With both debug builds I logged a memory difference message within a couple of hours. The first debug build was under an hour, the second took maybe 1-2 hours to show up.

dc42

For comparison, my system has logged anything from 0 to 4 memory differences in 24 hours. I have two instances of DWC connected (from Firefox and Chrome running on the same PC) and I reduced the DWC polling interval to 100ms.

Exerqtor

@omtek said in Random Hard Fault resets on Duet 3 Mini WiFi:

@Exerqtor With both debug builds I logged a memory difference message within a couple of hours. The first debug build was under an hour, the second took maybe 1-2 hours to show up.

@dc42 said in Random Hard Fault resets on Duet 3 Mini WiFi:

For comparison, my system has logged anything from 0 to 4 memory differences in 24 hours. I have two instances of DWC connected (from Firefox and Chrome running on the same PC) and I reduced the DWC polling interval to 100ms.

Hmm ok, in that case my machine is either really good behaved or I've setup something wrong

Installed the 3.6-debug fw.
Hooked a computer with USB.
Installed the newest YAT from sourceforge.
Connected it to the printer (tested by sending G28 in the termnial and the printer homed).
After testing connection I sent M111 P8 S1 through YAT.
Opened three instances(not three tabs, if thats any difference) or DWC in chrome + one instance of OrcaSlicer on one pc, plus i check in on my phone every once i a while too.

It's now been sitting for roughly 22 hours without anything happening. No output in YAT, and no reboot/reset.

gloomyandy

@Exerqtor Did you...

Leave it running with at least one DWC session connected

I think DC42 said he had two DWC instances running (in different browsers).

Exerqtor

@gloomyandy Added that info to the previous post 🙈

dc42

@Exerqtor have you ever had one of these resets since you upgraded to 3.6 beta 1?

Exerqtor

@dc42 Not really, after I stopped using chrome (other than the occasionall check-in with chrome on my phone) I haven't really had any resets at all that I've noticed.

The laptop have restarted (forced Windows update 🤦‍♂️) within the last two hours. So i don't know if anything got output in that timespan (no resets though). I've got it back up logging now and plan to have it runnning until the weekend.

I see the debug log has reached 3.21gb by now though. So idk if I maybe should have disabled that.

Going on 22hours since the laptop rebooted, and it's still no output on my end.

omtek

Another memory difference message this afternoon. No resets to speak of, either.

*** Memory difference at line 2228 offset 12: original 0a0d392e copy 20032128, original changed, copy ok, fix=yes

and another

*** Memory difference at line 2228 offset 52: original 20036658 copy 0d0a0d6d, original ok, copy changed, fix=no

edit #3 - busy day...

*** Memory difference at line 2228 offset 60: original 2001882c copy 0a0d392e, original ok, copy changed, fix=no

omtek

Two more memory difference messages logged overnight. Printer was idle but has been printing nicely. Still no resets to speak of.

*** Memory difference at line 2228 offset 60: original 2001882c copy 0a0d392e, original ok, copy changed, fix=no

*** Memory difference at line 2228 offset 56: original 2002c5d8 copy 0a0d656e, original ok, copy changed, fix=no

Exerqtor

No output in my end what so ever. Has anyone else had outputs with the 3.6 build? 🤔

dc42

@Exerqtor I had no output or resets using 3.6 either. I reverted to 3.5.3 and got two memory difference reports within a few hours.

Exerqtor

~~@dc42~~ ~~Ok, should i do that too? At least until tomorrow, or do you have enough datapoints on 3.5?~~

Reverted to 3.5.3 debug and have it running now. Straight off the bat I noticed a difference, when I sent G28Z in YAT to comfirm connection i got ok<LF> in return, but I didn't get anything in return when I input something in the 3.6 debug.

oliof

@Exerqtor do your machines run in Marlin compat mode?

dc42

@Exerqtor bear in mind that recent versions of YAT do not reconnect automatically when the Duet restarts. I always have to disconnect and reconnect YAT manually using the icons in the menu bar.