I don't see why - architecture support should be orthogonal to the core itself. I think if you have tests that can't pass with an a57+v8.1a, it smells like dodgy tests IMHO. On Tue, 31 Mar 2015 at 16:50, Vladimir Sukharev <[email protected]> wrote:
> It has been added already in backend - http://reviews.llvm.org/D8505 , > where we got no other decent possibility. > Why not to use the same in frontend? > Also, see comments below > > > ================ > Comment at: lib/Basic/Targets.cpp:4197 > @@ -4184,2 +4196,3 @@ > .Cases("cortex-a53", "cortex-a57", "cortex-a72", "8A") > + .Case("generic-armv8.1-a","8_1A") > .Default(nullptr); > ---------------- > for a57+v8.1a it would be weird to try to pass the test, commented below. > a57 would normally normally yield macro > ``` > __ARM_ARCH_8A__ 1 > ``` > see https://github.com/llvm-mirror/clang/blob/ > 1d941335617c6199e8340084c8a6c37c1b8014ec/lib/Basic/Targets.cpp#L4244 > > But I suspect we'd need to have > ``` > __ARM_ARCH_8_1A__ 1 > ``` > , as tested in very last test here. > Unfortunately, without ACLE issued we could not yet realize, whether we'd > need both of them. > > ================ > Comment at: test/Preprocessor/arm-target-features.c:322 > @@ +321,3 @@ > +// CHECK-V81A: __ARM_ARCH 8 > +// CHECK-V81A: __ARM_ARCH_8_1A__ 1 > +// CHECK-V81A: #define __ARM_ARCH_PROFILE 'A' > ---------------- > for a57+v8.1a it would be weird to pass this test > > http://reviews.llvm.org/D8493 > > EMAIL PREFERENCES > http://reviews.llvm.org/settings/panel/emailpreferences/ > > > > _______________________________________________ > cfe-commits mailing list > [email protected] > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits >
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
