[PATCH] D138868: AMDGPU/clang: Remove target features from address space test builtins

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: clang/test/CodeGenOpenCL/builtins-amdgcn-flat-address-space.cl:8
+// be initialized to something useful. The proper way to diagnose invalid flat
+// usage is to forbid flat pointers on unsupported targets.
+
----------------
Joe_Nash wrote:
> arsenm wrote:
> > Joe_Nash wrote:
> > > What part of the toolchain is responsible for forbiding flat pointers on unsupported targets?
> > The frontend should just error on languages with flat pointers. Really we should just implement software flat pointers, it wouldn’t be difficult and might only require minimal driver cooperation if any 
> Ok, I just want to ask if that is in place or should be in place before this patch lands. So we don't accidentally convert a compile time error into a runtime error. Otherwise LGTM.
In practice you’ll be failing to select all over on regular code long before you see this predicate 


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D138868/new/

https://reviews.llvm.org/D138868