[BUG] G29 S-1 doesn't output any gcode after G29 P1 has completed

[Drachenkaetzchen]

Bug Description

When doing a G29 S-1 on factory defaults, lots of gcode commands are written to the console.

After a G29 P1 has completed, G29 S-1 does not output any gcode.

My Configurations

Configuration.h Configuration_adv.h

Steps to Reproduce

1. Run G29 P1
2. Run G29 S-1

Expected behavior:

Actual mesh should be outputted.

Actual behavior:

No output.

Additional info

This also happens on a RUMBA AVR-based machine with Marlin from several months ago. Seems like Marlin thinks the mesh is invalid for some reason.

[randellhodges]

I can confirm. (I can only probe 43 out of 49 points, those 7 are the cause of the problem)

If you toss in a G29 P3 before the G29 S-1, it'll work. A quick peek at the code reveals a call to mesh_is_valid, which if any point isnan, it won't export anything.

[Drachenkaetzchen]

I thought I had done a G29 P3 here as well and received also no output. I'll verify that shortly.

[tatusah]

+1 here, without filling the incomplete mesh with G29 P3 the G29 S-1 doesn't output the mesh in gcode. SKR mini e3 v1.2

[Drachenkaetzchen]

G29 P3 does not work after a G29 P1.

Result:

> G29 P1
Probing mesh point 1/100.
[...]
Probing mesh point 49/100.
X:150.00 Y:60.00 Z:20.00 E:0.00 Count A:16800 B:7200 Z:8000
ok
> G29 S-1
ok
> G29 P3
ok
> G29 S-1
ok

[Drachenkaetzchen]

A G29 T results in:

Send: G29 T
Recv: 
Recv: Bed Topography Report:
Recv: 
Recv:     (  1,279)                                                              (299,279)
Recv:         0       1       2       3       4       5       6       7       8       9
Recv:  9 | -0.000  -0.036  -0.072  -0.022  -0.030  -0.026  -0.025  +0.009  +0.012  +0.014
Recv:    |
Recv:  8 | -0.062  -0.067  -0.072  -0.034  -0.050  -0.039  -0.025  +0.000  -0.012  -0.012
Recv:    |
Recv:  7 | -0.098  -0.098  -0.064  -0.045  -0.071  -0.051  -0.007  -0.008  -0.036  -0.036
Recv:    |
Recv:  6 | -0.072  -0.078  -0.083  -0.023  -0.010  -0.023  -0.032  +0.002  -0.061  -0.061
Recv:    |
Recv:  5 | -0.069  -0.069  -0.053  -0.019  +0.002  +0.002  +0.002  -0.047  -0.063  -0.063
Recv:    |
Recv:  4 | -0.103  -0.103  -0.057  -0.079  -0.024 [-0.035] -0.040  -0.057  -0.062  -0.062
Recv:    |
Recv:  3 | -0.104  -0.104  -0.098  -0.047  -0.033  -0.074  -0.106  -0.110  -0.085  -0.059
Recv:    |
Recv:  2 | -0.104  -0.104  -0.098  -0.015  -0.033  -0.074  -0.106  -0.110  -0.085  -0.059
Recv:    |
Recv:  1 |   .       .       .       .       .       .       .       .       .       .
Recv:    |
Recv:  0 |   .       .       .       .       .       .       .       .       .       .
Recv:         0       1       2       3       4       5       6       7       8       9
Recv:     (  1,  1)                                                              (299,  1)
Recv: 

It seems that G29 P3 doesn't fill in some points, and the user is never notified about that.

[Drachenkaetzchen]

I tried to use the mesh editor to fill in the remaining values, but the bed crashed into the nozzle as Marlin didn't seem to use either BLTouch as an endstop or was confused about the mesh value being invalid.

[randellhodges]

Give it a try with a 7x7 grid. That is what I used on an SKR 1.3. I end up with only 1 set of points that cannot be probed due to my probe offset and the P3 works.

Edit: Someone more familiar with the code would need to chime in, but maybe smart_fill doesn't like so many unpopulated points and that is why it fails to fill them all in?

[tatusah]

I think that you have to use G29 P3 as many times as you have empty rows or columns in the mesh after the P1. Like in my case using 9x9 and the edge / offset limits I use there are 2 empty columns and I've to use G29 P3 twice to fill those both. This is the way I've seen the smart fill work for the time I've been using it.

[Drachenkaetzchen]

I think that G29 S-1 should report back the values, even if the mesh is invalid. This would be consistent with G29 Sx.

[randellhodges]

I would think that should be safe. Just need to remove one line.

https://github.com/MarlinFirmware/Marlin/blob/7e7cbecf9ae001d40f6d9398fc3105a89716ecaa/Marlin/src/feature/bedlevel/ubl/ubl.cpp#L47

Would there be any issues running the M421 it spits out but have some data "lingering" in the previous nan points? Or do you think it should also export the nan state and have M421 import that as well?

[Drachenkaetzchen]

I think it should behave like G29 Sn to have it consistent. The reason I'm asking for this is since flash is currently broken on the Fysetc S6 / STM32, and sdcard emulation isn't working either, I'm looking for a way to at least manually load/save UBL states via gcode.

[github-actions[bot]]

This issue is stale because it has been open 30 days with no activity. Remove stale label / comment or this will be closed in 5 days.

[phongshader]

has this been fixed, I've been unable to run any G29 command with the exception of G29 T

[github-actions[bot]]

This issue is stale because it has been open 30 days with no activity. Remove stale label / comment or this will be closed in 5 days.

[github-actions[bot]]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.