[Solaris] __float128 is supported on Solaris/x86

[ro]

Summary

When rebasing https://reviews.llvm.org/D40898 with GCC 5.4 on Solaris 11.4, I ran
into a few instances of

In file included from /vol/llvm/src/compiler-rt/local/test/asan/TestCases/Posix/asan-symbolize-sanity-test.cc:19:
In file included from /usr/gcc/5/lib/gcc/x86_64-pc-solaris2.11/5.4.0/../../../../include/c++/5.4.0/string:40:
In file included from /usr/gcc/5/lib/gcc/x86_64-pc-solaris2.11/5.4.0/../../../../include/c++/5.4.0/bits/char_traits.h:39:
In file included from /usr/gcc/5/lib/gcc/x86_64-pc-solaris2.11/5.4.0/../../../../include/c++/5.4.0/bits/stl_algobase.h:64:
In file included from /usr/gcc/5/lib/gcc/x86_64-pc-solaris2.11/5.4.0/../../../../include/c++/5.4.0/bits/stl_pair.h:59:
In file included from /usr/gcc/5/lib/gcc/x86_64-pc-solaris2.11/5.4.0/../../../../include/c++/5.4.0/bits/move.h:57:
/usr/gcc/5/lib/gcc/x86_64-pc-solaris2.11/5.4.0/../../../../include/c++/5.4.0/type_traits:311:39: error: __float128 is not supported on this target

struct __is_floating_point_helper<__float128>
                                  ^

during make check-all. The line above is inside

#if !defined(STRICT_ANSI) && defined(_GLIBCXX_USE_FLOAT128)

template<>
  struct __is_floating_point_helper<__float128>
  : public true_type { };

#endif

While the libstdc++ header indicates support for __float128, clang does not, but
should. The following patch implements this and fixed those errors.

[ro]

I forgot: this patch is on top of https://reviews.llvm.org/D35755.

[rsmith]

Mechanically, the code change looks fine, but I can't comment on whether this is a correct change for Solaris, or whether the type provided by __float128 would use the right floating-point representation. You will also need to provide a test for this change.

[ro]

In D41240#955391, @rsmith wrote:

Mechanically, the code change looks fine, but I can't comment on whether this is a correct change for Solaris, or whether the type provided by __float128 would use the right floating-point representation. You will also need to provide a test for this change.

The representation of __float128 is the same on all x86 targets AFAICT (that's certainly how this is done in GCC).

I've looked at the tests currently testing __float128 and except for one (to be enabled in a moment) already pass with this patch.

[ro]

Enables test/CodeGenCXX/float128-declarations.cpp on Solaris/x86, too.

[ro]

I believe there's nothing more left for me here. Is the patch ok now?

Thanks.

Rainer

[ro]

ping

[ro]

ping^2

[ro]

ping^3

[rsmith]

@fedor.sergeev Can you review this for correctness? Thanks!

[fedor.sergeev]

LGTM.

[krytarowski]

Does it work for you for 32-bit? @joerg opposed adoption of 32-bit version as it seemed to be unimplemented..

[joerg]

Things are different for a libgcc-based toolchain and a compiler-rt based toolchain.

lib/Basic/Targets/OSTargets.h

void getOSDefines(const LangOptions &Opts, const llvm::Triple &Triple,
     else
       Builder.defineMacro("_XOPEN_SOURCE", "500");
     if (Opts.CPlusPlus)
       Builder.defineMacro("__C99FEATURES__");
     Builder.defineMacro("_LARGEFILE_SOURCE");
     Builder.defineMacro("_LARGEFILE64_SOURCE");
     Builder.defineMacro("__EXTENSIONS__");
     Builder.defineMacro("_REENTRANT");
+    if (this->HasFloat128)
+      Builder.defineMacro("__FLOAT128__");
   }
 
 public:
   SolarisTargetInfo(const llvm::Triple &Triple, const TargetOptions &Opts)
       : OSTargetInfo<Target>(Triple, Opts) {
     // FIXME: WIntType should be SignedLong
+    switch (Triple.getArch()) {
+    default:
+      break;
+    case llvm::Triple::x86:
+    case llvm::Triple::x86_64:
+      this->HasFloat128 = true;
+      break;
+    }
   }
 };
 
 // Windows target
 template <typename Target>
 class LLVM_LIBRARY_VISIBILITY WindowsTargetInfo : public OSTargetInfo<Target> {
 protected:
   void getOSDefines(const LangOptions &Opts, const llvm::Triple &Triple,