On Sun, 8 Apr 2007, Adrian Bunk wrote: > On Sun, Apr 08, 2007 at 04:56:07PM -0400, Robert P. J. Day wrote: > > (in fact, if you take a quick look at all of the > > include/asm-*/auxvec.h headers, most of them are empty. what's > > the point?) > The point is that an "#include <linuc/auxvec.h>" compiles. i've been thinking about this, and this approach just grates on me, but i don't have an immediate solution for it. as it stands, there's a generic, arch-independent header file linux/auxvec.h. ok, that's reasonable. so anyone who needs auxvec.h content does: #include <linux/auxvec.h> but there are *some* architectures (as i read them, only 5 out of 23) which need additional content in their auxvec.h files. and to solve that, linux/auxvec.h is augmented to include asm/auxvec.h, which means that every *other* architecture (the remaining 18 out of 23) is now required to implement an empty auxvec.h header file, for the sake of a correct compilation. that strikes me as an absurd approach for two reasons: 1) it doesn't scale well. in the extreme case, you might someday have a single architecture that requires something in the way of extra content in a header file, at which point you'll have to add n-1 dummy header files to all of the other architectures. 2) more importantly, the definition of an architecture's header files is being dictated by a *totally different architecture*. that's just grotesque. there *has* to be a cleaner solution. what the above is doing is just ugly. rday -- ======================================================================== Robert P. J. Day Linux Consulting, Training and Annoying Kernel Pedantry Waterloo, Ontario, CANADA http://fsdev.net/wiki/index.php?title=Main_Page ======================================================================== -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
