[clang] [clang] Implement constexpr bit_cast for vectors (PR #66894)

[Richard Smith via cfe-commits]

================
@@ -140,11 +140,12 @@ void g28(void) {
   typedef short v12i16 __attribute((vector_size(24)));
   typedef long double v2f80 __attribute((vector_size(24)));
   // CHECK: @g28.a = internal global <1 x i64> <i64 10>
-  // CHECK: @g28.b = internal global <12 x i16> <i16 0, i16 0, i16 0, i16 -32768, i16 16383, i16 0, i16 0, i16 0, i16 0, i16 -32768, i16 16384, i16 0>
-  // CHECK: @g28.c = internal global <2 x x86_fp80> <x86_fp80 0xK3FFF8000000000000000, x86_fp80 0xK40008000000000000000>, align 32
+  // @g28.b = internal global <12 x i16> <i16 0, i16 0, i16 0, i16 -32768, i16 16383, i16 0, i16 0, i16 0, i16 0, i16 -32768, i16 16384, i16 0>
+  // @g28.c = internal global <2 x x86_fp80> <x86_fp80 0xK3FFF8000000000000000, x86_fp80 0xK40008000000000000000>, align 32
   static v1i64 a = (v1i64)10LL;
-  static v12i16 b = (v12i16)(v2f80){1,2};
-  static v2f80 c = (v2f80)(v12i16){0,0,0,-32768,16383,0,0,0,0,-32768,16384,0};
+  //FIXME: support constant bitcast between vectors of x86_fp80
+  //static v12i16 b = (v12i16)(v2f80){1,2};
+  //static v2f80 c = (v2f80)(v12i16){0,0,0,-32768,16383,0,0,0,0,-32768,16384,0};
----------------
zygoloid wrote:

Please add some test coverage that we properly diagnose this situation.

I think we'll produce a diagnostic mentioning `bit_cast`, which is probably not what we want here. Maybe the simplest thing to do to fix that would be to replace `bit_cast` with `bit cast` in all the diagnostics. That seems at least as good for the `std::bit_cast` / `__builtin_bit_cast` case, and better for the vector cast case.

https://github.com/llvm/llvm-project/pull/66894