[Patch 2] Native QNX Neutrino support

[Kris Warkentin]

> I finally found the time to review your "take 2".

Thanks Mark.

> * Use xstrdup() instead of strdup().

Done.

> * The function calls in the definitions of the QNX_NODE macro need
>   spaces before the '(' just like in normal functions.  Could you turn
>   this macro into a function?

Turned the macro and function into one nto_node() function.

> * While what Andrew says about not having an XM_FILE is desirable, I
>   think we still need xm-i386.h to have the best possible handling of
>   floating-point numbers on a native GDB.  If you don't set it,
>   unnecessary conversions of floating-point numbers might occur,
>   leading to loss of precision.  I'll leave the decision to you.

While we don't need an xm-nto.h, it wouldn't seem to hurt to specify XM_FILE
as xm-i386.h.  I'll do that.

> With those changes, please go ahead and check it in.

Done.

cheers,

Kris