M600 filament pause hangs on load, hotend drops to 0 °C on 0.4.0 / 1.4.46

[pdscomp]

Summary

M600-based filament pause is broken on the 0.4.0 patched release (built from upstream 1.4.46). The printer unloads filament, then hangs forever waiting for "load filament", never sets the hotend temp, and the hotend drops to 0 °C. Manually setting the hotend temp has no effect. User downgraded to 0.3.0 to restore working pause behavior.

Reported by

ornox. on Discord #patched-fw — see thread starting at https://discord.com/channels/1367538416539013122/1427058990759678103/1521437046516875344

Environment

• Firmware: OpenCentauri 0.4.0 (upstream app 1.4.46)
• Printer: Elegoo Centauri Carbon (CC1)
• Pause command: M600
• Slicer: not specified, but using stock M600 filament-change gcode

Observed behavior

1. Print reaches M600.
2. Printer performs unload successfully.
3. UI shows "load filament" and waits indefinitely.
4. The temp-setting step is skipped.
5. Hotend temperature falls to 0 °C.
6. Manually setting hotend temp via UI does nothing.
7. Print cannot resume.
8. Downgrade to 0.3.0 restores correct pause behavior.

Expected behavior

M600 should:

• Unload filament,
• Prompt/execute load filament,
• Maintain or restore target hotend temperature,
• Allow the user to resume the print.

Related reports from the last two weeks in #patched-fw

These may share root cause or help narrow whether the regression is upstream 1.4.46 vs. our patch layer.

Date	User	Relevant detail
2026-06-22	f0gh0rn	"issues unloading and then loading filament" on 1.4.46; rolled back to 1.4.40
2026-06-24	f0gh0rn	"filament not grabbing" on 1.4.46 is a "stock 1.4.46 issue"
2026-06-26	badpapabear1	Canvas refusing to load filament; suspected hardware, later recovered
2026-06-27	theshinyakuma	Canvas/hub filament sensor resistance issue
2026-06-30	ornox.	This report — M600 pause hangs on load, hotend drops to 0

Related issues

• Printer hang with pause in gcode generated by slicer #70 — broader pause/hang behavior with slicer-generated pause gcode
• Respect M0 gcode #82 — request to support M0 as a simple pause alternative

Hypothesis

pdscomp noted on Discord that this is likely an upstream 1.4.46 regression in the M600 handler / load-filament state machine, rather than something introduced by the OpenCentauri patch layer. The fact that 0.3.0 works and 0.4.0 (1.4.46 base) fails supports this.

Next steps / what would close this

1. Reproduce on a CC1 running 0.4.0 with a minimal M600 gcode snippet.
2. Compare the M600 handler between the 1.4.44 (0.3.0 base) and 1.4.46 app binaries.
3. Identify the state-machine or temp-management change that causes the load step to wait forever with hotend at 0 °C.
4. Either patch it in OpenCentauri or document a workaround (e.g. slicer-level M600 replacement).

Minimal reproduction gcode (suggested)

; Start with normal print warmup
M104 S200
M109 S200
G28
; ... a few print moves ...
M600
; ... resume print moves ...

Tested by

• No verified test reports on the #99 patch artifact yet.
• Related Discord reports (f0gh0rn, fleetofgreen, timnolte) confirm the M600 unload/no-load and temperature-drop behavior on 1.4.46; fleetofgreen reports the same multicolor loading issue on stock firmware.

[pdscomp]

Fix ready for test on paul/m600-load-temp-fix

Root cause: The 1.4.46 change_filament / M600 state machine is new and sensor-driven. After PauseResume::cmd_M600 queues PAUSE, CUT_OFF_FILAMENT, and EXTRUDE_FILAMENT E=-60, the unload sequence turns the hotend off with M104 S0. The handler then dispatches a generic resume/change-filament-completed event (0x1a) via sub_1c7908, but it never reissues a temperature command before the UI prompts the user to load filament. The hotend drops to 0 °C and the load step hangs.

This is not a side effect of fix-noncanvas-load (which only guards null sensor pointers); it's an upstream 1.4.46 regression in the new load/unload state machine. Subagent analysis of the 1.4.44 vs 1.4.46 decompilation and the 1.1.40 source confirmed this.

Fix: Added a new binary patch fix-m600-pause that hooks the bl sub_1c7908 at the end of the M600 handler (0x001b84c4) and injects M104 S200 before the event is dispatched. This keeps the hotend warm enough for manual filament loading; the original print temperature is restored by RESUME.

• Hook: 0x001b84c4 → 0x00450d00
• Code cave: 0x00450d00-0x00450d7f
• String cave: 0x00450c40-0x00450c7f (M104 S200)
• Works for both Canvas and non-Canvas toolheads (no sensor dependency)
• Final patched app SHA256: 3bde8bc2255e0d9c2b9878730f2ea4e64ed07047d16df4ae57b2476143cb76fa

Artifacts:

• oc-patches/cc1-app/fix-m600-pause/ — generator, bsdiff, patch integration, docs, and m600-test.gcode
• m600-test.gcode — minimal reproduction (heat to 220 °C, home, park, M600, park, shut down)

Verification:

• Local sudo ./build.sh 1.4.46 passes with the patch enabled.
• GitHub Actions build passes: https://github.com/OpenCentauri/cc-fw-tools/actions/runs/28448874216
• generate_patch.py output matches the committed bsdiff.
• Gilfoyle review was run against the branch; the only feedback was documentation/assertion improvements, which have been applied.

Next step: If anyone can flash the artifact from the Actions run above and confirm M600 no longer hangs with a cold hotend, we can merge the branch.

Branch: https://github.com/OpenCentauri/cc-fw-tools/tree/paul/m600-load-temp-fix

[timnolte]

Just hit this issue as well. Hoping the fix is working and can be pushed out soon. I was able to get things working by simply going to the main screen setting the temp and then pressing the the print Resume. One additional major issue is that the filament Unload/Load screen loses the ability to unload/load the filament, rebooting the printer is required to restore the functionality.

[pdscomp]

Consolidated Discord feedback from #patched-fw (as of 2026-07-05):

• No direct test reports on the #99 patch artifact yet.
• Related user reports confirm the M600 / filament-change symptom on 1.4.46:
  • f0gh0rn (Jul 4): M600 unloads then does not auto-load; temps drop until resume.
  • fleetofgreen (Jul 4): observed multicolor loading issue on stock 1.4.46, concluded it is not an OC issue.
  • timnolte (Jul 4): explained M600 should unload/load, and noted PAUSE gcode behavior has also regressed recently.

→ Still waiting for a verified test of the #99 patch artifact.