This is a venerable fact. Exceptions in D. melanogaster strain Y-007 have been observed and D. ananassae has consistent male cross-overs, but this work dates back to the 1970s.
This Current Biology 2002 paper is not exactly new, but sheds some light on the issue in question.
In male Drosophila melanogaster, meiosis occurs in the absence of recombination or a recognizable synaptonemal complex (SC).
The authors used a GFP-lac fusion to observe a series of sections of the male fly in meiosis with integrated lac - specific binding regions. They see for the most part that the GFP bound segments of the DNA are not associated in cell division. That is, homologous chromosomes physically associate during meiosis (via the SC) and in male drosophila, they do not.
It seems from more recent work, that the details of how this work is still being worked out but several gene regions and factors have been identified. Suspected factors that help conspire to align the various homologous chromosomes are being looked into. Take a look at the P.I.s biblio for more.
This isn't my main field, but as I understand it, many aspects of meiosis are still being mapped out.
As to the question of how this evolved and why; its definitely strange.
It looks like there is a gender selected trait where the SC doesn't form in males. Its easy to imagine a mutation that caused males only to lose recombination - the process is so complex its hard to imagine such a trait not showing up.
Still it's surprising to think that the trait has not only survived but become dominant in Drosophila all over the world. To be a common traits to so many Drosophila species, it must have happened more than 10 million years ago. Explanations of the advantages of sexual reproduction often include recombination. Even the human Y chromosome has a self-recombination mechanism that reduces error in the coding regions. As such, its hard to believe that this is not a huge disadvantage for the flies. Still, if if it were such a big deal, you'd imagine Y-007 would have dominated and re-introduced the trait.
The only theory I can think of might be called 'lazy males'; that the males got the females to do one more thing for the offspring and it somehow outcompetes the sperm of the recombining males. Lots of sperm are non-motile; that's a fact for humans, I have no idea what its like for flies that recombine. Take a look at the variance in morphology as well - sperm are really crazy ugly sometimes and a good deal of the mating process is to help weed out the defective ones. Without recombination, might it be that males produce viable sperm at a greater rate, helping them in sperm competition, which I'm sure is the case for fly mating - Drosophila are not monogamous too I'd guess? Anyway that's not a researched idea, just a guess on my part.