🚀 go-pugleaf

I've been trying to get Matt Dillon's PIPE: device driver to run under
Aztec C 3.6a.  (I want to write my own device, and this seems like a
good first step.)  I've made these changes to the source (from Fish disk
55 or so):
    Put 'L' after all constant int args to lib functions
    Put in call to geta4() as first statement in _main (but after
	initializers -- is this a problem?  Do I need geta4() here?)
    replaced bmov() and bzero() calls with appropriate Aztec calls.

I haven't been compiling with -L (long int) switch.  It compiles
without complaint, and links ok with misc.o (which also compiles fine).
It also mounts without complaint, but when I try dir PIPE:, the guru
strikes.  (I know PIPE: doesn't support dir; that's why I tried it.  It
seemed like a good way to check that the driver is there without
checking any of the piping code.)

So my question is:  is there anything I'm missing?  Any compile or link
switches I need?  Any Aztec-specific restructuring (different main
procedure or something)?  Any suggestions would be appreciated.
Otherwise, it's time to SDB this sucker.  (BTW, the SDB manual implies
that it IS possible to write device drivers and libraries in Aztec.)

Thanks in advance.
-Peter Schachte
pds@quintus.uucp
..!sun!quintus!pds