[clang] [llvm] [AMDGPU] Implement AsyncMark Stages (PR #220442)

[Sameer Sahasrabuddhe via cfe-commits]

================
@@ -65,6 +115,11 @@ If `A` is *completed-at* a `wait.asyncmark()` operation `Y`, then `A`
 
 ## Examples
 
+The mask argument is omitted in the examples below; they all use the empty mask,
+so every asyncmark tracks every async operation and there is only one sequence
+to reason about. The {ref}`interleaved stages <amdgpu-async-interleaved-stages>`
+example shows what narrower masks add.
----------------
ssahasra wrote:

Examples need to follow the syntax in the spec. If there is no intrinsic that takes an empty mask, this example should be changed to take a zero mask.

https://github.com/llvm/llvm-project/pull/220442