The problem you ran into is one of the prime examples of why one should always get all the coauthors' approvals before submitting an article.
If you have in fact documented approval from your previous advisor for the paper in the form in which you submitted, the facts are on your side. And there's little your advisor can do about it: it is his own fault for not having caught something he disliked and to have given you the OK to submit. (Though pissing off one's advisor is generally not a good career move.)
If you have not obtained the approval, then you almost certainly must retract the paper. All journals I have submitted to and refereed for either requires the assent to publication be individually given by all the authors (in which case the journal will often ask for contact information of all authors to be keyed in when submitting), or that the corresponding author certify that he is in a position to speak on behalf of the other authors about the paper. The fact that your advisor is making noises means that either
- He will disagree, as a coauthor, for the paper to be published, or
- You have in fact lied when you certified that you can speak for him, in which case the journal will retract your paper for violating their rules
(assuming some form of this question was asked during the submission process). In other words, most likely you are already in a situation where you either voluntarily retract the submission (in which case the only people who will know about it are likely yourself, the two coauthors involved, and the handling editor at the journal), or, if you chose to publish anyway, be forced to retract the paper by the journal, since many journals take authorship problems rather seriously. In this latter case anyone with a subscription to the journal would know that your paper was retracted because of some sort of misconduct (yes, misrepresenting author information is scientific misconduct). Your call.
The above basically answers what you should do in your situation. But what about the case where the paper has not already been submitted, but there are disagreements over authorship and wording of the paper?
My only suggestion is that in that scenario, you should get together with your two "co-authors" and hammer out a compromise yourselves. While I find the objections by your advisor a bit strange, there may be ulterior reasons you are not telling us. A possible compromise in this case would be for you to publish one paper, containing only the contributions of your thesis work, with your advisor, and a second one containing the reanalysis and extension, with the other author.
And do not, in any case, go over the objections of your co-author(s) and submit a paper in a form they do not agree with. That's a recipe for academic misconduct right there.