commit 7557afa000527fa3d8d3da58015b35028b6c370a
author Michael Liao <michael.hliao@gmail.com> Tue Feb 26 18:49:36 2019 +0000
committer Michael Liao <michael.hliao@gmail.com> Tue Feb 26 18:49:36 2019 +0000
tree ae088c03d7bccad344311f75075354becbc0ac39
parent c0ffe705c5bf908e9eaeef4fc4d0ed3ad85811f1

[AMDGPU] Allow using integral non-type template parameters

Summary:
- Allow using integral non-type template parameters in the following
  attributes

  __attribute__((amdgpu_flat_work_group_size(<min>, <max>)))
  __attribute__((amdgpu_waves_per_eu(<min>[, <max>])))

Reviewers: kzhuravl, yaxunl

Subscribers: jvesely, wdng, nhaehnle, dstuttard, tpr, t-tye, jdoerfert, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58623

llvm-svn: 354909

clang/lib/Sema/SemaTemplateInstantiateDecl.cpp

diff --git a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
index 91a402d..db63529 100644
--- a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
+++ b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
@@ -344,6 +344,51 @@
       Attr.getRange());
 }
 
+static void instantiateDependentAMDGPUFlatWorkGroupSizeAttr(
+    Sema &S, const MultiLevelTemplateArgumentList &TemplateArgs,
+    const AMDGPUFlatWorkGroupSizeAttr &Attr, Decl *New) {
+  // Both min and max expression are constant expressions.
+  EnterExpressionEvaluationContext Unevaluated(
+      S, Sema::ExpressionEvaluationContext::ConstantEvaluated);
+
+  ExprResult Result = S.SubstExpr(Attr.getMin(), TemplateArgs);
+  if (Result.isInvalid())
+    return;
+  Expr *MinExpr = Result.getAs<Expr>();
+
+  Result = S.SubstExpr(Attr.getMax(), TemplateArgs);
+  if (Result.isInvalid())
+    return;
+  Expr *MaxExpr = Result.getAs<Expr>();
+
+  S.addAMDGPUFlatWorkGroupSizeAttr(Attr.getLocation(), New, MinExpr, MaxExpr,
+                                   Attr.getSpellingListIndex());
+}
+
+static void instantiateDependentAMDGPUWavesPerEUAttr(
+    Sema &S, const MultiLevelTemplateArgumentList &TemplateArgs,
+    const AMDGPUWavesPerEUAttr &Attr, Decl *New) {
+  // Both min and max expression are constant expressions.
+  EnterExpressionEvaluationContext Unevaluated(
+      S, Sema::ExpressionEvaluationContext::ConstantEvaluated);
+
+  ExprResult Result = S.SubstExpr(Attr.getMin(), TemplateArgs);
+  if (Result.isInvalid())
+    return;
+  Expr *MinExpr = Result.getAs<Expr>();
+
+  Expr *MaxExpr = nullptr;
+  if (auto Max = Attr.getMax()) {
+    Result = S.SubstExpr(Max, TemplateArgs);
+    if (Result.isInvalid())
+      return;
+    MaxExpr = Result.getAs<Expr>();
+  }
+
+  S.addAMDGPUWavesPerEUAttr(Attr.getLocation(), New, MinExpr, MaxExpr,
+                            Attr.getSpellingListIndex());
+}
+
 void Sema::InstantiateAttrsForDecl(
     const MultiLevelTemplateArgumentList &TemplateArgs, const Decl *Tmpl,
     Decl *New, LateInstantiatedAttrVec *LateAttrs,
@@ -437,6 +482,18 @@
       continue;
     }
 
+    if (const AMDGPUFlatWorkGroupSizeAttr *AMDGPUFlatWorkGroupSize =
+            dyn_cast<AMDGPUFlatWorkGroupSizeAttr>(TmplAttr)) {
+      instantiateDependentAMDGPUFlatWorkGroupSizeAttr(
+          *this, TemplateArgs, *AMDGPUFlatWorkGroupSize, New);
+    }
+
+    if (const AMDGPUWavesPerEUAttr *AMDGPUFlatWorkGroupSize =
+            dyn_cast<AMDGPUWavesPerEUAttr>(TmplAttr)) {
+      instantiateDependentAMDGPUWavesPerEUAttr(*this, TemplateArgs,
+                                               *AMDGPUFlatWorkGroupSize, New);
+    }
+
     // Existing DLL attribute on the instantiation takes precedence.
     if (TmplAttr->getKind() == attr::DLLExport ||
         TmplAttr->getKind() == attr::DLLImport) {