RE: Bug#116772: [hpoj-devel] Re: Bug#116772: hpoj: failed to buil d on m68k.

["PASCHAL,DAVID \(HP-Roseville,ex1\)" <david_paschal@hp.com>]

Mark Purcell wrote:
> I have isolated the problem. I was pretty sure you were 
> already checking for
> the existence of sys/io.h and you were in the config script.
Not completely.  The intention was to check for the ability to include
sys/io.h from C++ code, but if that fails, it (incorrectly) assumes that
sys/io.h can be included from C code, enables ioWrapper.c to be compiled and
linked in, and adds the BROKEN_IO_H define.  I'm working on correcting this
in the configure script.

> The problem was a typo in ioWrapper.c
> 
> --- hpoj-0.8.orig/mlcd/ioWrapper.c
> +++ hpoj-0.8/mlcd/ioWrapper.c
> @@ -24,8 +24,9 @@
>  #error Must be compiled as C, not C++
>  #endif
> 
> -#undef BROKEN_IO_H
> +#ifdef BROKEN_IO_H
>  #include "ioWrapper.h"
> +#endif
> 
>  int ioplWrapper(int level) {
>         return iopl(level);
That wasn't a typo.  Although your change makes it compile, the existence of
the BROKEN_IO_H macro, which is always present when ioWrapper.c is compiled
anyway, causes ioWrapper.h to redefine iopl() to ioplWrapper(), etc.,
thereby causing infinite recursion where ioplWrapper inadvertently calls
itself.

David