If you are in school, and have no/little time constraints, switch your major to physics or declare a major in physics.
If you are in school, with severe time and financial constraints (that's what I was in) take as many physics classes as you can, take the PGRE and ace it, and apply for medium tier schools PHD programs and any MS programs only programs (get your MS then decide if you want to go to industry or try for a top tier PHD). Do well, then either work in industry, or get a PHD. Once in the MS program, they will usually ask you to take grad versions of whatever classes you've finished, while making up the classes you haven't finished at the BS level. This is useful: a MS is pretty employable, you can get publications in, there's little risk, you get paid for it at some universities or you can get your employer to pay for you, and you'll get to know your professors for them to recommend you to top tier PHD programs. Once in, you'll have an advantage over those who went straight from the BS in that you've already done everything once and can either test out and start research directly, or if you're forced to retake everything, at least you'd have gone through it before while others get "grad school shock" and burn out. This assumes you do well in your MS.
If you have already graduated and are in industry, self study the PGRE, apply for a MS, do well, get promoted in your industry job or go for your PHD.
After your PHD, have no illusions about academic work. You work in computer science - be prepared to go back to the same job you're working now, since theoretical physicists directly compete in your field for programming jobs.