[llvm-branch-commits] [clang] [llvm] [Clang][OMPX] Add the code generation for multi-dim `thread_limit` clause (PR #102717)

[Alexey Bataev via llvm-branch-commits]

================
@@ -505,11 +505,14 @@ void OpenMPIRBuilder::getKernelArgsVector(TargetKernelArgs &KernelArgs,
 
   Value *NumTeams3D =
       Builder.CreateInsertValue(ZeroArray, KernelArgs.NumTeams[0], {0});
+  Value *NumThreads3D =
+      Builder.CreateInsertValue(ZeroArray, KernelArgs.NumThreads[0], {0});
   for (unsigned I = 1; I < std::min(KernelArgs.NumTeams.size(), MaxDim); ++I)
     NumTeams3D =
         Builder.CreateInsertValue(NumTeams3D, KernelArgs.NumTeams[I], {I});
-  Value *NumThreads3D =
-      Builder.CreateInsertValue(ZeroArray, KernelArgs.NumThreads, {0});
+  for (unsigned I = 1; I < std::min(KernelArgs.NumThreads.size(), MaxDim); ++I)
----------------
alexey-bataev wrote:

`for (unsigned I : seq<unsigned>(1, std::min(KernelArgs.NumThreads.size(), MaxDim)))`

https://github.com/llvm/llvm-project/pull/102717