The combination of gcc option -m32 is likely to conflict with -Duse64bitall or its implications

Yeah - I wasn't sure if -Duse64bitall would inevitably fail to provide the 64bit pointers on an -m32 build, and I also wondered whether Osprey actually wanted the -m32 arg.
In the end, rather than providing any detailed explanation, I just left it as a suggestion that might be tried - something that can be discussed further if Osprey is so inclined.

IIRC, if -Duse64bitint delivers 64bit integers then -Duse64bitall will do the same, even if it fails to deliver 64bit pointers.

Cheers,
Rob