🚀 go-pugleaf

I built a version of glibc using gcc2.2.2 in a cross
development environment, target=sun3 host=sun4,
intended to run a standalone 68020 based platform.

it seems with this configuration of gcc2.2.2 that size_t is an int,
to be consistent (urgh) with sun-lossage; clearly, it should be
an unsigned.  This is a tremendous rathole, I realize.

in vsprintf is the line
  f.__bufsize = (size_t) (f.__put_limit - f.__bufp);
in which __put_limit is initialized to 0xffffffff;
this ends up giving f.__bufsize a large negative number (since size_t
is an int), which then causes these lines in fwrite
	if (n > buffer_space)
	  n = buffer_space;
to be executed (n is now < 0), which is later used as an argument to
	    memcpy ((PTR) stream->__bufp, (PTR) p, n);
and since n < 0 memcpy fortunately copies nothing.