In production, I haven't had boards which are programmed more at once, so can't really help you here, but I have done a lot of boards that have multiple IC's on them, all programmed through JTAG. As you said, the TDO from one is daisy chained with the TDI from another. It might be the same thing, but I can not be sure.
How this thing works is you must make a JTAG sequence in a program such as the JTAG ProVision (this is actually crap; hard to use and not user friendly, plus I don't know if it's free) or the AEX Sequencer (this is somewhat easyer to play with) In the sequence, you specify what kind of IC or ICs you are programming, then the path to the SW the JTAG program should load onto the IC, then maybe a few boundary check tests, (if needed, for high density pin ICs or BGAs) to check for any unsoldered or shorted pins.
These kinds of things are somewhat touchy so expect a lot of fiddling around, both with the hardware part and the software part (setting the right clocks, the right sequence order, etc.)
Hope this info is somewhat helpful to you.