[OpenACC][NFC] AST changes for Reduction combiner #162573
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This is the first patch of a handful to get the reduction combiner recipe lowering properly. THIS patch is NFC as it doesn't actually change anything except the structure of the AST. For each 'combiner' recipe we need a 'LHS' 'RHS' and expression to represent the operation. Each var-reference can have 1 or more combiners. IF it is a plain scalar, or a struct with the proper operator, or an array of either of those, there will be 1. HOWEVER, aggregates without the proper operator are supposed to be broken down and done from their elements (which can only be scalars). In this case, we will represent 1 'combiner' recipe per field-decl. This patch only puts the infrastructure in place to do so, future patches wll do the work to fill this in.
llvmbot
added
clang
|
@llvm/pr-subscribers-clang @llvm/pr-subscribers-clang-modules Author: Erich Keane (erichkeane) ChangesThis is the first patch of a handful to get the reduction combiner recipe lowering properly. THIS patch is NFC as it doesn't actually change anything except the structure of the AST. For each 'combiner' recipe we need a 'LHS' 'RHS' and expression to represent the operation. Each var-reference can have 1 or more combiners. IF it is a plain scalar, or a struct with the proper operator, or an array of either of those, there will be 1. HOWEVER, aggregates without the proper operator are supposed to be broken down and done from their elements (which can only be scalars). In this case, we will represent 1 'combiner' recipe per field-decl. This patch only puts the infrastructure in place to do so, future patches wll do the work to fill this in. Full diff: https://github.com/llvm/llvm-project/pull/162573.diff 5 Files Affected:
diff --git a/clang/include/clang/AST/OpenACCClause.h b/clang/include/clang/AST/OpenACCClause.h
index 58ba8d91f1277..79cffeb4dade9 100644
--- a/clang/include/clang/AST/OpenACCClause.h
+++ b/clang/include/clang/AST/OpenACCClause.h
@@ -1280,13 +1280,31 @@ class OpenACCCreateClause final
// 'main' declaration used for initializaiton, which is fixed.
struct OpenACCReductionRecipe {
VarDecl *AllocaDecl;
- // TODO: OpenACC: this should eventually have the operations here too.
- OpenACCReductionRecipe(VarDecl *A) : AllocaDecl(A) {}
+ // A combiner recipe is represented by an operation expression. However, in
+ // order to generate these properly, we have to make up a LHS and a RHS
+ // expression for the purposes of generation.
+ struct CombinerRecipe {
+ VarDecl *LHS;
+ VarDecl *RHS;
+ Expr *Op;
+ };
+
+ // Contains a collection of the recipe elements we need for the combiner:
+ // -For Scalars, there will be 1 element, just the combiner for that scalar.
+ // -For a struct with a valid operator, this will be 1 element, just that
+ // call.
+ // -For a struct without the operator, this will be 1 element per field, which
+ // should be the combiner for that element.
+ // -For an array of any of the above, it will be the above for the element.
+ llvm::SmallVector<CombinerRecipe, 1> CombinerRecipes;
+
+ OpenACCReductionRecipe(VarDecl *A, llvm::ArrayRef<CombinerRecipe> Combiners)
+ : AllocaDecl(A), CombinerRecipes(Combiners) {}
bool isSet() const { return AllocaDecl; }
static OpenACCReductionRecipe Empty() {
- return OpenACCReductionRecipe(/*AllocaDecl=*/nullptr);
+ return OpenACCReductionRecipe(/*AllocaDecl=*/nullptr, {});
}
};
diff --git a/clang/lib/AST/StmtProfile.cpp b/clang/lib/AST/StmtProfile.cpp
index f3b5478222488..3cd033e73800f 100644
--- a/clang/lib/AST/StmtProfile.cpp
+++ b/clang/lib/AST/StmtProfile.cpp
@@ -2769,10 +2769,19 @@ void OpenACCClauseProfiler::VisitReductionClause(
for (auto &Recipe : Clause.getRecipes()) {
Profiler.VisitDecl(Recipe.AllocaDecl);
+
// TODO: OpenACC: Make sure we remember to update this when we figure out
// what we're adding for the operation recipe, in the meantime, a static
// assert will make sure we don't add something.
- static_assert(sizeof(OpenACCReductionRecipe) == sizeof(int *));
+ static_assert(sizeof(OpenACCReductionRecipe::CombinerRecipe) ==
+ 3 * sizeof(int *));
+ for (auto &CombinerRecipe : Recipe.CombinerRecipes) {
+ if (CombinerRecipe.Op) {
+ Profiler.VisitDecl(CombinerRecipe.LHS);
+ Profiler.VisitDecl(CombinerRecipe.RHS);
+ Profiler.VisitStmt(CombinerRecipe.Op);
+ }
+ }
}
}
diff --git a/clang/lib/Sema/SemaOpenACC.cpp b/clang/lib/Sema/SemaOpenACC.cpp
index 8471f02f323d6..4824b5a3082a4 100644
--- a/clang/lib/Sema/SemaOpenACC.cpp
+++ b/clang/lib/Sema/SemaOpenACC.cpp
@@ -2946,5 +2946,5 @@ OpenACCReductionRecipe SemaOpenACC::CreateReductionInitRecipe(
AllocaDecl->setInit(Init.get());
AllocaDecl->setInitStyle(VarDecl::CallInit);
}
- return OpenACCReductionRecipe(AllocaDecl);
+ return OpenACCReductionRecipe(AllocaDecl, {});
}
diff --git a/clang/lib/Serialization/ASTReader.cpp b/clang/lib/Serialization/ASTReader.cpp
index 6acf79acea111..868f0cc8b1da7 100644
--- a/clang/lib/Serialization/ASTReader.cpp
+++ b/clang/lib/Serialization/ASTReader.cpp
@@ -13009,9 +13009,22 @@ OpenACCClause *ASTRecordReader::readOpenACCClause() {
llvm::SmallVector<OpenACCReductionRecipe> RecipeList;
for (unsigned I = 0; I < VarList.size(); ++I) {
- static_assert(sizeof(OpenACCReductionRecipe) == sizeof(int *));
VarDecl *Recipe = readDeclAs<VarDecl>();
- RecipeList.push_back({Recipe});
+
+ static_assert(sizeof(OpenACCReductionRecipe::CombinerRecipe) ==
+ 3 * sizeof(int *));
+
+ llvm::SmallVector<OpenACCReductionRecipe::CombinerRecipe> Combiners;
+ unsigned NumCombiners = readInt();
+ for (unsigned I = 0; I < NumCombiners; ++I) {
+ VarDecl *LHS = readDeclAs<VarDecl>();
+ VarDecl *RHS = readDeclAs<VarDecl>();
+ Expr *Op = readExpr();
+
+ Combiners.push_back({LHS, RHS, Op});
+ }
+
+ RecipeList.push_back({Recipe, Combiners});
}
return OpenACCReductionClause::Create(getContext(), BeginLoc, LParenLoc, Op,
diff --git a/clang/lib/Serialization/ASTWriter.cpp b/clang/lib/Serialization/ASTWriter.cpp
index 09b1e58ef220c..82ccde84de37d 100644
--- a/clang/lib/Serialization/ASTWriter.cpp
+++ b/clang/lib/Serialization/ASTWriter.cpp
@@ -8925,8 +8925,17 @@ void ASTRecordWriter::writeOpenACCClause(const OpenACCClause *C) {
writeOpenACCVarList(RC);
for (const OpenACCReductionRecipe &R : RC->getRecipes()) {
- static_assert(sizeof(OpenACCReductionRecipe) == 1 * sizeof(int *));
AddDeclRef(R.AllocaDecl);
+
+ static_assert(sizeof(OpenACCReductionRecipe::CombinerRecipe) ==
+ 3 * sizeof(int *));
+ writeUInt32(R.CombinerRecipes.size());
+
+ for (auto &CombinerRecipe : R.CombinerRecipes) {
+ AddDeclRef(CombinerRecipe.LHS);
+ AddDeclRef(CombinerRecipe.RHS);
+ AddStmt(CombinerRecipe.Op);
+ }
}
return;
}
|
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/27/builds/17287 Here is the relevant piece of the build log for the reference |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/25/builds/12191 Here is the relevant piece of the build log for the reference |
DharuniRAcharya
pushed a commit
to DharuniRAcharya/llvm-project
that referenced
this pull request
Oct 13, 2025
This is the first patch of a handful to get the reduction combiner recipe lowering properly. THIS patch is NFC as it doesn't actually change anything except the structure of the AST. For each 'combiner' recipe we need a 'LHS' 'RHS' and expression to represent the operation. Each var-reference can have 1 or more combiners. IF it is a plain scalar, or a struct with the proper operator, or an array of either of those, there will be 1. HOWEVER, aggregates without the proper operator are supposed to be broken down and done from their elements (which can only be scalars). In this case, we will represent 1 'combiner' recipe per field-decl. This patch only puts the infrastructure in place to do so, future patches wll do the work to fill this in.
akadutta
pushed a commit
to akadutta/llvm-project
that referenced
this pull request
Oct 14, 2025
This is the first patch of a handful to get the reduction combiner recipe lowering properly. THIS patch is NFC as it doesn't actually change anything except the structure of the AST. For each 'combiner' recipe we need a 'LHS' 'RHS' and expression to represent the operation. Each var-reference can have 1 or more combiners. IF it is a plain scalar, or a struct with the proper operator, or an array of either of those, there will be 1. HOWEVER, aggregates without the proper operator are supposed to be broken down and done from their elements (which can only be scalars). In this case, we will represent 1 'combiner' recipe per field-decl. This patch only puts the infrastructure in place to do so, future patches wll do the work to fill this in.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is the first patch of a handful to get the reduction combiner recipe lowering properly. THIS patch is NFC as it doesn't actually change anything except the structure of the AST.
For each 'combiner' recipe we need a 'LHS' 'RHS' and expression to represent the operation.
Each var-reference can have 1 or more combiners.
IF it is a plain scalar, or a struct with the proper operator, or an array of either of those, there will be 1.
HOWEVER, aggregates without the proper operator are supposed to be broken down and done from their elements (which can only be scalars). In this case, we will represent 1 'combiner' recipe per field-decl.
This patch only puts the infrastructure in place to do so, future patches wll do the work to fill this in.