[PATCH] D103112: Reimplement __builtin_unique_stable_name as __builtin_sycl_unique_stable_name

[Aaron Ballman via Phabricator via cfe-commits]

aaron.ballman added inline comments.


================
Comment at: clang/include/clang/Basic/LangOptions.h:448
+
+  bool isSYCL() const { return SYCLIsDevice || SYCLIsHost; }
 };
----------------
FWIW, we also have `SYCLVersion != SYCL_None` as a possible way to express this. Perhaps we should just use that anywhere we're using `SYCLIsDevice || SYCLIsHost` currently? (I don't have a strong opinion but am bringing it up in case someone else does.)


================
Comment at: clang/test/CodeGenSYCL/unique_stable_name.cpp:86-89
+  // FIXME: Ensure that j is incremented because VLAs are terrible.
+  int j = 55;
+  puts(__builtin_sycl_unique_stable_name(int[++j]));
+  // CHECK: call spir_func void @puts(i8 addrspace(4)* addrspacecast (i8* getelementptr inbounds ([[STRING_SIZE]], [[STRING_SIZE]]* @[[STRING]], i32 0, i32 0) to i8 addrspace(4)*))
----------------
@rjmccall  -- any opinions or ideas on this? I think VLAs should likely behave the same as they do in `sizeof`, etc.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D103112/new/

https://reviews.llvm.org/D103112