OK so today was not a good day in thesisland. My advisor became suddenly aware of how the code framework I work in has fundamental flaws. I have been working in this environment for years now and am well aware of its limitations, and I have designed my thesis algorithm and justification to work within the confines of the system. He basically indicated that this was unsatisfactory and that there had to be a rigorous physical justification for all results. Given that the code I'm using does not match analytical test problems, this is not possible. He says that if the numerics aren't right, then there is no way to justify the thesis. He wants first-principles derivations of things that I'm not sure can be done; he wants to know information that I don't think is available. The bulk of the background information in the thesis is paraphrased right out of the literature by the heads of the field, but he indicates that he doesn't like the reasoning or refers to it as "controversial." I tell other people in the group what he's asking for and I get the response, "that's ridiculous."
He is basically having me re-derive everything and re-write every chapter as we go through. Nevermind that these writeups have been available on my website for the better part of two years, and that I've sent him the link to it at least a dozen times. My emails and phone messages went unanswered. He showed up three weeks ago and decreed that what I had was not acceptable and is demanding changes, some of which I'm not sure can be done at all, and the rest I'm pretty sure can't be done before the deadline. Many of the things he wants to know are well outside of my range of expertise and aren't really relevant to the thesis as far as I can tell.
I need to be done with the entire document in 6 weeks if I hope to defend in time for the deadline.
That's bullshit on the advisor's part. In case you weren't already certain of it. He's violating ethical norms of the academic community -- or, more specifically, he already has done so by accepting responsibility for you and for your work with your supervisor at the Lab, then failing to supervise.
In other words, the advisor has a responsibility to do one of the following:
a) Supervise the advisee, ensure that the advisee's work is progressing toward graduation, provide feedback at frequent intervals, and then support the work with his own prestige, or
b) Turn responsibility for the advisee's education and progress (as above) over to a trusted colleague, accept the colleague's judgment, and support that judgement with his own prestige.
In short, he is responsible. He does not have the ethical option of ignoring you for years, then bitching that he doesn't like what you've been doing. You have a nominal supervisor at the Lab; if that person has either guided you incorrectly or failed to guide you at all, then it's your advisor's fault for trusting that person, and he does not get a "get out of jail free card" by screwing you. By contesting your thesis plan at this late date, he's effectively saying, "I'm not competent to be a professor."
The November issue of Physics Today is devoted to ethics in the profession. In a nutshell, it says that the APS has rather abruptly realized that subordinate abuse is a huge problem. Grad student slavery, etc. This is an example. Talk to your department, and point out that (a) this is irresponsible treatment of a subordinate, and (b) that happens to be a big, hot issue right now.