[PATCH] D39364: [Polly][acc] Do not statically dispatch into IslNodeBuilder's createFor

[Philip Pfaffe via Phabricator via llvm-commits]

philip.pfaffe created this revision.
Herald added subscribers: kbarton, nemanjai.

When GPUNodeBuilder creates loops inside the kernel, it dispatches to
IslNodeBuilder. This however is surprisingly dangerous, since it accesses the
AST Node's user through the wrong type. This patch fixes this problem by
overriding createFor correctly.

This fixes PR35010.


https://reviews.llvm.org/D39364

Files:
  include/polly/CodeGen/IslNodeBuilder.h
  lib/CodeGen/IslNodeBuilder.cpp
  lib/CodeGen/PPCGCodeGeneration.cpp


Index: lib/CodeGen/PPCGCodeGeneration.cpp
===================================================================
--- lib/CodeGen/PPCGCodeGeneration.cpp
+++ lib/CodeGen/PPCGCodeGeneration.cpp
@@ -415,6 +415,8 @@
   /// @param UserStmt The ast node to generate code for.
   virtual void createUser(__isl_take isl_ast_node *UserStmt);
 
+  virtual void createFor(__isl_take isl_ast_node *Node);
+
   enum DataDirection { HOST_TO_DEVICE, DEVICE_TO_HOST };
 
   /// Create code for a data transfer statement
@@ -1286,6 +1288,11 @@
   isl_ast_node_free(UserStmt);
   return;
 }
+
+void GPUNodeBuilder::createFor(__isl_take isl_ast_node *Node) {
+  createForSequential(Node, false);
+}
+
 void GPUNodeBuilder::createKernelCopy(ppcg_kernel_stmt *KernelStmt) {
   isl_ast_expr *LocalIndex = isl_ast_expr_copy(KernelStmt->u.c.local_index);
   LocalIndex = isl_ast_expr_address_of(LocalIndex);
Index: lib/CodeGen/IslNodeBuilder.cpp
===================================================================
--- lib/CodeGen/IslNodeBuilder.cpp
+++ lib/CodeGen/IslNodeBuilder.cpp
@@ -530,19 +530,15 @@
 }
 
 void IslNodeBuilder::createForSequential(__isl_take isl_ast_node *For,
-                                         bool KnownParallel) {
+                                         bool MarkParallel) {
   isl_ast_node *Body;
   isl_ast_expr *Init, *Inc, *Iterator, *UB;
   isl_id *IteratorID;
   Value *ValueLB, *ValueUB, *ValueInc;
   Type *MaxType;
   BasicBlock *ExitBlock;
   Value *IV;
   CmpInst::Predicate Predicate;
-  bool Parallel;
-
-  Parallel = KnownParallel || (IslAstInfo::isParallel(For) &&
-                               !IslAstInfo::isReductionParallel(For));
 
   bool LoopVectorizerDisabled =
       IsLoopVectorizerDisabled(isl::manage(isl_ast_node_copy(For)));
@@ -582,13 +578,13 @@
   bool UseGuardBB =
       !SE.isKnownPredicate(Predicate, SE.getSCEV(ValueLB), SE.getSCEV(ValueUB));
   IV = createLoop(ValueLB, ValueUB, ValueInc, Builder, LI, DT, ExitBlock,
-                  Predicate, &Annotator, Parallel, UseGuardBB,
+                  Predicate, &Annotator, MarkParallel, UseGuardBB,
                   LoopVectorizerDisabled);
   IDToValue[IteratorID] = IV;
 
   create(Body);
 
-  Annotator.popLoop(Parallel);
+  Annotator.popLoop(MarkParallel);
 
   IDToValue.erase(IDToValue.find(IteratorID));
 
@@ -795,7 +791,9 @@
     createForParallel(For);
     return;
   }
-  createForSequential(For, false);
+  bool Parallel =
+      (IslAstInfo::isParallel(For) && !IslAstInfo::isReductionParallel(For));
+  createForSequential(For, Parallel);
 }
 
 void IslNodeBuilder::createIf(__isl_take isl_ast_node *If) {
Index: include/polly/CodeGen/IslNodeBuilder.h
===================================================================
--- include/polly/CodeGen/IslNodeBuilder.h
+++ include/polly/CodeGen/IslNodeBuilder.h
@@ -351,7 +351,7 @@
   bool preloadInvariantEquivClass(InvariantEquivClassTy &IAClass);
 
   void createForVector(__isl_take isl_ast_node *For, int VectorWidth);
-  void createForSequential(__isl_take isl_ast_node *For, bool KnownParallel);
+  void createForSequential(__isl_take isl_ast_node *For, bool MarkParallel);
 
   /// Create LLVM-IR that executes a for node thread parallel.
   ///


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D39364.120557.patch
Type: text/x-patch
Size: 3211 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20171027/7716e73b/attachment.bin>