[llvm] [AMDGPU] Support preloading hidden kernel arguments (PR #98861)

[Matt Arsenault via llvm-commits]

================
@@ -31,9 +33,93 @@ class PreloadKernelArgInfo {
   const GCNSubtarget &ST;
   unsigned NumFreeUserSGPRs;
 
-public:
-  SmallVector<llvm::Metadata *, 8> KernelArgMetadata;
+  enum HiddenArg : unsigned {
+    HIDDEN_BLOCK_COUNT_X,
+    HIDDEN_BLOCK_COUNT_Y,
+    HIDDEN_BLOCK_COUNT_Z,
+    HIDDEN_GROUP_SIZE_X,
+    HIDDEN_GROUP_SIZE_Y,
+    HIDDEN_GROUP_SIZE_Z,
+    HIDDEN_REMAINDER_X,
+    HIDDEN_REMAINDER_Y,
+    HIDDEN_REMAINDER_Z,
+    END_HIDDEN_ARGS
+  };
+
+  struct HiddenArgInfo {
+    unsigned Offset;
+    unsigned Size;
+    const char *Name;
+  };
+
+  static constexpr HiddenArgInfo HiddenArgs[END_HIDDEN_ARGS] = {
----------------
arsenm wrote:

Document what this is and is for 

https://github.com/llvm/llvm-project/pull/98861