[BUG] Buffer overrun .. fixing G29 J

[bwspath]

Description

G29 J flushed the current mesh to 1 value .. didn't adjust
found the issue which is pretty much writing out of bounds. compiler complained about it too.

Values were stored at offset that are not located or used ( floats were stored at vector.x )

Requirements

bed probing and mesh tilting

Benefits

G29 J now works and adjusts the bed

Configurations

HAS_BED_PROBE

Related Issues

#22175

[DerAndere1]

This looks like a reasonable fix of a regression introduced in commit 7726af9. FYI: The original code before commit 7726af9 was:

  new_matrix.vectors[0] = row_0;
  new_matrix.vectors[1] = row_1;
  new_matrix.vectors[2] = row_2;

[bwspath]

@DerAndere1
well reverting back to before 7726af9 (didn't look at is as I am a git noob and still looking at how I can spot previous revisions) looks cleaner. Though I get the issues with adding the axes.. though that isn't an issue in the final transform ... or is it ? would be a stage before/after the first 3 axes transform ... which could be 1...

[DerAndere1]

@bwspath Sorry, I do not completely understand your final question. The only thing I can say is that your solution looks great. I just showed the original code for reference. Your code should be functionally equivalent. If you are asking about how to deal with axes beyond xyz (i.e. axes I, J, or K): Those are not relevant here. The vectors and 3x3 matrices so far do not deal with axes but rather with 3D coordinates, so they only have x, y and z elements).

[bwspath]

  new_matrix.vectors[0] = row_0;
  new_matrix.vectors[1] = row_1;
  new_matrix.vectors[2] = row_2;

would be a better solution that what I did .. I failed to get my git foo ok to see what it used to be.

[Roxy-3D]

@bwspath I agree with your thoughts here: #22398 (comment)

If you zap the Pull Request to reflect that, I'll press the 'Squash & Merge' button! Thank You for the fix!

[DerAndere1]

the proposed change was added to Marlin bugfix-2.0.x: with commit 154decf