[clang][SME] Emit error for OpemMP captured regions in streaming functions

Currently, these generate incorrect code, as streaming attributes are
not propagated to the outlined function. As we've yet to work on mixing
OpenMP and streaming functions (and determine how they should interact
with OpenMP's runtime), we think it is best to disallow this for now.

Author: MacDue

clang/include/clang/Basic/DiagnosticSemaKinds.td

@@ -3864,6 +3864,8 @@ def err_sme_definition_using_za_in_non_sme_target : Error<
   "function using ZA state requires 'sme'">;
 def err_sme_definition_using_zt0_in_non_sme2_target : Error<
   "function using ZT0 state requires 'sme2'">;
+def err_sme_openmp_captured_region : Error<
+  "OpenMP captured regions are not yet supported in streaming functions">;
 def warn_sme_streaming_pass_return_vl_to_non_streaming : Warning<
   "%select{returning|passing}0 a VL-dependent argument %select{from|to}0 a function with a different"
   " streaming-mode is undefined behaviour when the streaming and non-streaming vector lengths are different at runtime">,

clang/lib/Sema/SemaStmt.cpp

@@ -4568,9 +4568,23 @@ buildCapturedStmtCaptureList(Sema &S, CapturedRegionScopeInfo *RSI,
   return false;
 }
 
+static bool
+isOpenMPCapturedRegionInArmStreamingFunction(Sema const &S,
+                                             CapturedRegionKind Kind) {
+  if (!S.getLangOpts().OpenMP || Kind != CR_OpenMP)
+    return false;
+  FunctionDecl *FD = S.getCurFunctionDecl(/*AllowLambda=*/true);
+  if (!FD)
+    return false;
+  return IsArmStreamingFunction(FD, /*IncludeLocallyStreaming=*/true);
+}
+
 void Sema::ActOnCapturedRegionStart(SourceLocation Loc, Scope *CurScope,
                                     CapturedRegionKind Kind,
                                     unsigned NumParams) {
+  if (isOpenMPCapturedRegionInArmStreamingFunction(*this, Kind))
+    Diag(Loc, diag::err_sme_openmp_captured_region);
+
   CapturedDecl *CD = nullptr;
   RecordDecl *RD = CreateCapturedStmtRecordDecl(CD, Loc, NumParams);
 
@@ -4602,6 +4616,9 @@ void Sema::ActOnCapturedRegionStart(SourceLocation Loc, Scope *CurScope,
                                     CapturedRegionKind Kind,
                                     ArrayRef<CapturedParamNameType> Params,
                                     unsigned OpenMPCaptureLevel) {
+  if (isOpenMPCapturedRegionInArmStreamingFunction(*this, Kind))
+    Diag(Loc, diag::err_sme_openmp_captured_region);
+
   CapturedDecl *CD = nullptr;
   RecordDecl *RD = CreateCapturedStmtRecordDecl(CD, Loc, Params.size());
 

clang/test/Sema/aarch64-sme-streaming-openmp-captured-region.c

@@ -0,0 +1,46 @@
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme -fopenmp -fsyntax-only -verify %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme -fopenmp -fsyntax-only -verify=expected-cpp -x c++ %s
+
+int compute(int);
+
+void streaming_openmp_captured_region(int* out) __arm_streaming
+{
+  // expected-error@+2 {{OpenMP captured regions are not yet supported in streaming functions}}
+  // expected-cpp-error@+1 {{OpenMP captured regions are not yet supported in streaming functions}}
+  #pragma omp parallel for num_threads(32)
+  for(int ci =0;ci< 8;ci++)
+  {
+    out[ci] =compute(ci);
+  }
+}
+
+__arm_locally_streaming void locally_streaming_openmp_captured_region(int* out)
+{
+  // expected-error@+2 {{OpenMP captured regions are not yet supported in streaming functions}}
+  // expected-cpp-error@+1 {{OpenMP captured regions are not yet supported in streaming functions}}
+  #pragma omp parallel for num_threads(32)
+  for(int ci =0;ci< 8;ci++)
+  {
+    out[ci] = compute(ci);
+  }
+}
+
+/// OpenMP directives that don't create a captured region are okay:
+
+void streaming_function_openmp(int* out) __arm_streaming
+{
+  #pragma omp unroll full
+  for(int ci =0;ci< 8;ci++)
+  {
+    out[ci] =compute(ci);
+  }
+}
+
+__arm_locally_streaming void locally_streaming_openmp(int* out)
+{
+  #pragma omp unroll full
+  for(int ci =0;ci< 8;ci++)
+  {
+    out[ci] = compute(ci);
+  }
+}