To add, when I’m breaking my changes down into multiple parts for review, I tend to:

  * squash everything I’ve done into one commit
  * create a new branch off main/master that will be the “first commit”
  * cherry-pick changes (easy from some git guis) that represent a modular change.
  * push and make an MR from the new branch
  * rebase “the big commit” on top of the partial change.
  * wash, rinse and repeat for each change, building each MR off its requisite branch.

The squashing part is vital because otherwise you enter merge conflict hell with the rebase.

How about:

   * squash into one commit
   * git reset HEAD~1
   * git add -p
   * git commit -m commit1
   * repeat until no changes are left
   * add any file deletions/additions

I use this because you can have several commits marked e.g. "commit1". Then you make a final interactive rebase to squash them together.