[OpenMPOpt][FIX] Properly track changes to NestedParallelism

If we update the state, or indicate a pessimistic fixpoint, we need to
consider NestedParallelism too.

Fixes part of #66708

That said, the reproducer still needs malloc which we don't support on
AMD GPU. Will be added later.

Author: jdoerfert

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

@@ -733,6 +733,7 @@ struct KernelInfoState : AbstractState {
     SPMDCompatibilityTracker.indicatePessimisticFixpoint();
     ReachedKnownParallelRegions.indicatePessimisticFixpoint();
     ReachedUnknownParallelRegions.indicatePessimisticFixpoint();
+    NestedParallelism = true;
     return ChangeStatus::CHANGED;
   }
 
@@ -762,6 +763,8 @@ struct KernelInfoState : AbstractState {
       return false;
     if (ParallelLevels != RHS.ParallelLevels)
       return false;
+    if (NestedParallelism != RHS.NestedParallelism)
+      return false;
     return true;
   }
 
@@ -3580,7 +3583,8 @@ struct AAKernelInfo : public StateWrapper<KernelInfoState, AbstractAttribute> {
            ", #ParLevels: " +
            (ParallelLevels.isValidState()
                 ? std::to_string(ParallelLevels.size())
-                : "<invalid>");
+                : "<invalid>") +
+           ", NestedPar: " + (NestedParallelism ? "yes" : "no");
   }
 
   /// Create an abstract attribute biew for the position \p IRP.

llvm/test/Transforms/OpenMP/parallel_level_fold.ll

@@ -14,7 +14,7 @@ target triple = "nvptx64"
 ; CHECK: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i16
 ; CHECK: @[[NONE_SPMD_KERNEL_ENVIRONMENT:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr constant [[STRUCT_KERNELENVIRONMENTTY:%.*]] { [[STRUCT_CONFIGURATIONENVIRONMENTTY:%.*]] { i8 0, i8 0, i8 1 }, ptr null, ptr null }
 ; CHECK: @[[SPMD_KERNEL_ENVIRONMENT:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr constant [[STRUCT_KERNELENVIRONMENTTY:%.*]] { [[STRUCT_CONFIGURATIONENVIRONMENTTY:%.*]] { i8 0, i8 0, i8 2 }, ptr null, ptr null }
-; CHECK: @[[PARALLEL_KERNEL_ENVIRONMENT:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr constant [[STRUCT_KERNELENVIRONMENTTY:%.*]] { [[STRUCT_CONFIGURATIONENVIRONMENTTY:%.*]] { i8 0, i8 0, i8 2 }, ptr null, ptr null }
+; CHECK: @[[PARALLEL_KERNEL_ENVIRONMENT:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr constant [[STRUCT_KERNELENVIRONMENTTY:%.*]] { [[STRUCT_CONFIGURATIONENVIRONMENTTY:%.*]] { i8 0, i8 1, i8 2 }, ptr null, ptr null }
 ;.
 define weak void @none_spmd() "kernel" {
 ; CHECK-LABEL: define {{[^@]+}}@none_spmd