On Sun, Jun 11, 2017 at 1:33 PM, Xi Ruoyao <ryxi@xxxxxxxxxxxxxxxxx> wrote: > On 2017-06-11 09:29 +0200, Toby Douglass wrote: >> However, when I look in /usr/lib, I find this... >> >> /usr/lib/i386-linux-gnu > > This is just Debian multiarch hierarchy specification. > Irrelevant with your issue. Ah, so. Yes. I can install other archs, and they will turn up there with their names Stll, it indicates what the current system thinks it is, and I thought that might relate in some way to what the GCC driver is doing, when it guesses what the native system is. >> But when I use "-march=native", like so; "gcc -march=native -dM -E - < >> /dev/null | grep 86" >> >> I only get this! >> >> #define __i386 1 >> #define __i386__ 1 >> #define i386 1 > > In the first case, the architecture is decided by "--with-arch=" when > GCC was configured. Yes. When I make my own compiler, targetting i586, I get __586__. When I do the same are target i686, I get __686__. > In the second case, the architecture is guessed by GCC driver. Right. > Use "gcc [-march=native] -E -x c /dev/null -v". And see the flag > "-march=xxx" passed to `cc1` by `gcc` to figure out the real > architecture GCC used. Using the OS GCC (4.9.2, native build for i586-linux-gnu); WIthout, I get arch i586, tune generic. With, I get - ha! - "-march=native", and no tune. >> Now, I may be completely wrong, but it looks to me where the OS is >> pretending to be i386, it's actually misleading the mechanism behind >> "-march=native". There *is* no actual i686 Debian build, so I can't >> just use that instead of the i386 Debian. > > No. I don't think so. GCC is guessing the architecture for > "-march=native" using "cpuid" instruction on x86. GCC doesn't > look at OS. Hmm, that is very interesting. I have an E640. The result conflicts with the Debian installer, which selected i686. I wonder how the Debian installer guesses. > See the source code of > gcc_source/gcc/config/i386/driver-i386.c: host_detect_local_cpu. Huh. I didn't know there was an instrinsic for "cpuid". There's one on Windows, too, actually. Doesn't help me at compile time, unless I run something to generate the makefile. I've read through the code. I'm thinking about compiling a debug build and running it in the debugger to see what it does in host_detect_local_cpu. The processor here is a Atom E640, which may just be confusing the compiler. The processor was released in 2010. GCC 4.8.5 and 4.9.2 both came out over five years later, though. >> I want to use "-march=native" because it means the makefile selects >> the correct target for whomever is building. > > I doubt it. '__i386__' is defined for all 32-bit x86 architectures, > but '__i686__' is only for "-march=i686", not its 'subsets'. But that seems right, and it is what I expect. > For example, on my laptop (x86-64, multilib), I can use > "gcc -m32 -march=native". But then GCC guessed out > "- > march=ivybridge" and the macro "__corei7_avx__" was defined, > instead of "__i686__". Ooof. That does not sound like what I'd want to have happen, certainly from a compiler define point of view. >> Any advice, insight, corrections, words of wisdom, etc? > > I think it's better to use configuration script (maybe generated > by GNU autoconf) to get the architecture from several tests. I was hoping never to use autconf. Needing to learn a new language (m4) to generate makefiles is not appealling! :-P :-)
