>     I'm trying to set up a cross compiler for a vax-netbsd target
>     on a i586-linux host but have run into a few problems.

Don't forget to define REAL_ARITHMETIC in the gcc/Makefile, e.g.:

CFLAGS = -DREAL_ARITHMETIC

This forces GCC to software-simulate VAX floats so it can output the correct
bit patterns for VAX assembly.

Normally GCC has a grody assumption that all targets use the same float
format as the host, which these days is almost always IEEE, so it gets
away with it.

But cross-compiling to VAX breaks this immediately...

Todd Whitesel
toddpw @ best.com