I've switched back to the rpi 4 as a sanity check. Recompiled the kernel & modules, and can confirm that the q2m overlay I mentioned above, which comes with the 6.6y kernel, works with fifioPi, up to 384/DSD128.
After looking over the output of dmesg more thoroughly, on the pi 5 the q2m/k2m "card" fails to be registered, and this is not the case on the pi 4. The pi 4 still shows the I/O -121 errors in dmesg when attempting to update bits, mute, etc, so it seems these can be ignored.
I will have to do some more learning about the new RP1 chip found on the pi 5, as it seems this is the crux of the problem. With the old broadcom soc found in the earlier pi's, i2c traffic can switch between master & slave mode on the fly, whereas that's not possible with RP1's i2c. However, there are multiple i2c busses available on the RP1, and from what I can gather from the docs, it should be possible to run one bus in master mode while the other runs in slave mode.