Hmm. Yeah, the possibilities are getting a bit more limited here. If the led is flashing when a button is pressed, and stops flashing after a button stops being pressed, then there are no internal communication problems within Pierogi itself. Some sort of corruption within the Pierogi data would explain it, but wouldn't explain why the problem gets reset by turning the screen off and on again. A protocol bug would work too, but that wouldn't explain why the first keypress works. (And the "NEC" protocol is the best tested of all the protocol code in Pierogi.)

I hate to say it, but at this point, I'm thinking the issue probably lies outside of Pierogi itself...

If the led is lighting up, I can only think of two possibilities -- errors are being introduced into the led timing data as multiple requests come in, or the light is getting weaker as multiple requests come in. To rule out the second possibility, maybe you could hold the N900 right up to the sensor on the speaker set, and see if the buttons start working?

Odds are, though, that it's some sort of error in the timing data. Let me start digging into the CSSU changelogs again; maybe there's an IR-hardware-related change that went in somewhere. (Although, again, I would kind of expect that other users would have also experienced it by now. )