Been bouncing around ideas in my head about this. First of all, this might be the reason why qwerty12's R&D Mode Control (non-CLI) segfaults when ran in the /sbin/preinit launched shell. (Until I noticed this I blamed it on trying to load graphical elements from within a framebuffer console.)

Another thought - the open source getbootstate binary pali made uses libcal for reading the R&D mode flags. Which suggests somewhere I'm doing something wrong, because getbootstate is executed by /sbin/preinit before my shell runs, if I recall correctly. (Or maybe just after, but they're executed close together.) So it's not libcal by itself at fault, which is what I was thinking might be involved.

I will solve this eventually, when life gives me enough time again, the few users of my code have my word on that. My current plan is to start by looking at Pali's open source getbootstate code and see how he uses the libcal stuff.