Re: building OSKit for GCC 4.8.1
Posted: Tue Sep 24, 2013 6:58 am
brighteningeyes: The thing the assembler is failing at is a C preprocessor macro, as heavily used by NetBSD assembly. The assembly files #include some special headers that #define a bunch of things to help writing assembly. The _C_LABEL(cpu_class) expression expands somehow to (OSKIT_NETBSD_UVM_cpu_class). This isn't legal assembler in AT&T (but that's irrelevant, actually). Your problem is that the macro doesn't expand to what it is supposed to and you generate the wrong code. This is a C preprocessor problem - You should have known that since you claim expertise of C and C++.
The solution is to find out where _C_LABEL is declared and what part of the expansion process fails. My bet is that you failed to #define OSKIT_NETBSD_UVM_cpu_class yourself, or perhaps that particular expression is glued from OSKIT_NETBSD_UVM and cpu_class and you need to #define OSKIT_NETBSD_UVM. I don't know the answer to this, you'll need to read the headers yourself. This has nothing to do with AT&T syntax, so don't blame it on that.
The solution is to find out where _C_LABEL is declared and what part of the expansion process fails. My bet is that you failed to #define OSKIT_NETBSD_UVM_cpu_class yourself, or perhaps that particular expression is glued from OSKIT_NETBSD_UVM and cpu_class and you need to #define OSKIT_NETBSD_UVM. I don't know the answer to this, you'll need to read the headers yourself. This has nothing to do with AT&T syntax, so don't blame it on that.