[AMDGPU] Fix 160181. Be less optimistic when allocating module scope lds (llvm#4338)

From: Jon Chesterfield <[email protected]>

Author: hjagasiaAMD

llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp

@@ -649,7 +649,8 @@ class AMDGPULowerModuleLDS {
           ModuleScopeVariables.insert(GV);
         } else if (K.second.size() == 1) {
           KernelAccessVariables.insert(GV);
-        } else if (K.second == HybridModuleRootKernels) {
+        } else if (K.second == HybridModuleRootKernels &&
+                   set_is_subset(K.second, HybridModuleRootKernels)) {
           ModuleScopeVariables.insert(GV);
         } else {
           TableLookupVariables.insert(GV);

llvm/test/CodeGen/AMDGPU/lower-module-lds-precise-allocate-to-module-struct.ll

@@ -138,4 +138,3 @@ define amdgpu_kernel void @kern_block_direct_allocation() {
 ; CHECK: attributes #[[ATTR1]] = { "amdgpu-lds-size"="16" }
 ; CHECK: attributes #[[ATTR2:[0-9]+]] = { nocallback nofree nosync nounwind willreturn memory(none) }
 ; CHECK: attributes #[[ATTR3:[0-9]+]] = { nocallback nofree nosync nounwind speculatable willreturn memory(none) }
-;.