[PATCH] D123471: [CUDA] Create offloading entries when using the new driver

[Joseph Huber via Phabricator via cfe-commits]

jhuber6 added inline comments.


================
Comment at: clang/lib/CodeGen/CGCUDARuntime.h:56-70
+    enum OffloadRegionEntryKindFlag : uint32_t {
+      /// Mark the region entry as a kernel.
+      OffloadRegionKernelEntry = 0x0,
+    };
+
+    /// The kind flag of the global variable entry.
+    enum OffloadVarEntryKindFlag : uint32_t {
----------------
tra wrote:
> jhuber6 wrote:
> > tra wrote:
> > > We can also fold both enums into one, as we still have the ambiguity of what `flags=0` means. 
> > They're selected based on the size, if the size is zero it uses the kernel flags, otherwise it uses the variable flags. That's how it's done for OpenMP. I figured keeping the enums separate makes that more clear.
> > They're selected based on the size, if the size is zero it uses the kernel flags, otherwise it uses the variable flags. 
> 
> Why use two different enums, when one would do? It does not buy us anything other than unnecessary additional complexity.
> 
I mostly copied this from OpenMP, I can merge it into one.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D123471/new/

https://reviews.llvm.org/D123471