[MergeFunctions] Add support to run the pass over a set of function pointers #111045
+310
−15
Conversation
This file contains 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
remove templates unit tests added format updated data structures format
|
Thank you for submitting a Pull Request (PR) to the LLVM Project! This PR will be automatically labeled and the relevant teams will be notified. If you wish to, you can add reviewers by using the "Reviewers" section on this page. If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers. If you have further questions, they may be answered by the LLVM GitHub User Guide. You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums. |
|
@llvm/pr-subscribers-llvm-transforms Author: Rafael Eckstein (Casperento) ChangesThis modification will enable the usage of Summary of Changes:
Full diff: https://github.com/llvm/llvm-project/pull/111045.diff 5 Files Affected:
diff --git a/llvm/include/llvm/Transforms/IPO/MergeFunctions.h b/llvm/include/llvm/Transforms/IPO/MergeFunctions.h
index 822f0fd99188d0..1b3b1d22f11e28 100644
--- a/llvm/include/llvm/Transforms/IPO/MergeFunctions.h
+++ b/llvm/include/llvm/Transforms/IPO/MergeFunctions.h
@@ -15,7 +15,10 @@
#ifndef LLVM_TRANSFORMS_IPO_MERGEFUNCTIONS_H
#define LLVM_TRANSFORMS_IPO_MERGEFUNCTIONS_H
+#include "llvm/IR/Function.h"
#include "llvm/IR/PassManager.h"
+#include <map>
+#include <set>
namespace llvm {
@@ -25,6 +28,10 @@ class Module;
class MergeFunctionsPass : public PassInfoMixin<MergeFunctionsPass> {
public:
PreservedAnalyses run(Module &M, ModuleAnalysisManager &AM);
+
+ static bool runOnModule(Module &M);
+ static std::pair<bool, std::map<Function *, Function *>>
+ runOnFunctions(std::set<Function *> &F);
};
} // end namespace llvm
diff --git a/llvm/lib/Transforms/IPO/MergeFunctions.cpp b/llvm/lib/Transforms/IPO/MergeFunctions.cpp
index b50a700e09038f..a434d7920b6ccf 100644
--- a/llvm/lib/Transforms/IPO/MergeFunctions.cpp
+++ b/llvm/lib/Transforms/IPO/MergeFunctions.cpp
@@ -123,6 +123,7 @@
#include <algorithm>
#include <cassert>
#include <iterator>
+#include <map>
#include <set>
#include <utility>
#include <vector>
@@ -198,6 +199,8 @@ class MergeFunctions {
}
bool runOnModule(Module &M);
+ bool runOnFunctions(std::set<Function *> &F);
+ std::map<Function *, Function *> &getDelToNewMap();
private:
// The function comparison operator is provided here so that FunctionNodes do
@@ -298,17 +301,31 @@ class MergeFunctions {
// dangling iterators into FnTree. The invariant that preserves this is that
// there is exactly one mapping F -> FN for each FunctionNode FN in FnTree.
DenseMap<AssertingVH<Function>, FnTreeType::iterator> FNodesInTree;
+
+ /// Deleted-New functions mapping
+ std::map<Function *, Function *> DelToNewMap;
};
} // end anonymous namespace
PreservedAnalyses MergeFunctionsPass::run(Module &M,
ModuleAnalysisManager &AM) {
- MergeFunctions MF;
- if (!MF.runOnModule(M))
+ if (!MergeFunctionsPass::runOnModule(M))
return PreservedAnalyses::all();
return PreservedAnalyses::none();
}
+bool MergeFunctionsPass::runOnModule(Module &M) {
+ MergeFunctions MF;
+ return MF.runOnModule(M);
+}
+
+std::pair<bool, std::map<Function *, Function *>>
+MergeFunctionsPass::runOnFunctions(std::set<Function *> &F) {
+ MergeFunctions MF;
+ bool MergeResult = MF.runOnFunctions(F);
+ return {MergeResult, MF.getDelToNewMap()};
+}
+
#ifndef NDEBUG
bool MergeFunctions::doFunctionalCheck(std::vector<WeakTrackingVH> &Worklist) {
if (const unsigned Max = NumFunctionsForVerificationCheck) {
@@ -468,6 +485,47 @@ bool MergeFunctions::runOnModule(Module &M) {
return Changed;
}
+bool MergeFunctions::runOnFunctions(std::set<Function *> &F) {
+ bool Changed = false;
+ std::vector<std::pair<IRHash, Function *>> HashedFuncs;
+ for (Function *Func : F) {
+ if (isEligibleForMerging(*Func)) {
+ HashedFuncs.push_back({StructuralHash(*Func), Func});
+ }
+ }
+ llvm::stable_sort(HashedFuncs, less_first());
+ auto S = HashedFuncs.begin();
+ for (auto I = HashedFuncs.begin(), IE = HashedFuncs.end(); I != IE; ++I) {
+ if ((I != S && std::prev(I)->first == I->first) ||
+ (std::next(I) != IE && std::next(I)->first == I->first)) {
+ Deferred.push_back(WeakTrackingVH(I->second));
+ }
+ }
+ do {
+ std::vector<WeakTrackingVH> Worklist;
+ Deferred.swap(Worklist);
+ LLVM_DEBUG(dbgs() << "size of function: " << F.size() << '\n');
+ LLVM_DEBUG(dbgs() << "size of worklist: " << Worklist.size() << '\n');
+ for (WeakTrackingVH &I : Worklist) {
+ if (!I)
+ continue;
+ Function *F = cast<Function>(I);
+ if (!F->isDeclaration() && !F->hasAvailableExternallyLinkage()) {
+ Changed |= insert(F);
+ }
+ }
+ LLVM_DEBUG(dbgs() << "size of FnTree: " << FnTree.size() << '\n');
+ } while (!Deferred.empty());
+ FnTree.clear();
+ FNodesInTree.clear();
+ GlobalNumbers.clear();
+ return Changed;
+}
+
+std::map<Function *, Function *> &MergeFunctions::getDelToNewMap() {
+ return this->DelToNewMap;
+}
+
// Replace direct callers of Old with New.
void MergeFunctions::replaceDirectCallers(Function *Old, Function *New) {
for (Use &U : llvm::make_early_inc_range(Old->uses())) {
@@ -1004,6 +1062,7 @@ bool MergeFunctions::insert(Function *NewFunction) {
Function *DeleteF = NewFunction;
mergeTwoFunctions(OldF.getFunc(), DeleteF);
+ this->DelToNewMap.emplace(DeleteF, OldF.getFunc());
return true;
}
diff --git a/llvm/unittests/Transforms/Utils/CMakeLists.txt b/llvm/unittests/Transforms/Utils/CMakeLists.txt
index 5c7ec28709c169..7effa5d8e7d6d2 100644
--- a/llvm/unittests/Transforms/Utils/CMakeLists.txt
+++ b/llvm/unittests/Transforms/Utils/CMakeLists.txt
@@ -26,6 +26,7 @@ add_llvm_unittest(UtilsTests
LoopUtilsTest.cpp
MemTransferLowering.cpp
ModuleUtilsTest.cpp
+ MergeFunctionsTest.cpp
ScalarEvolutionExpanderTest.cpp
SizeOptsTest.cpp
SSAUpdaterBulkTest.cpp
diff --git a/llvm/unittests/Transforms/Utils/MergeFunctionsTest.cpp b/llvm/unittests/Transforms/Utils/MergeFunctionsTest.cpp
new file mode 100644
index 00000000000000..696c5391ef4f68
--- /dev/null
+++ b/llvm/unittests/Transforms/Utils/MergeFunctionsTest.cpp
@@ -0,0 +1,271 @@
+//===- MergeFunctionsTest.cpp - Unit tests for
+//MergeFunctionsPass-----------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Transforms/IPO/MergeFunctions.h"
+#include "llvm/AsmParser/Parser.h"
+#include "llvm/IR/LLVMContext.h"
+#include "llvm/IR/Module.h"
+#include "llvm/Support/SourceMgr.h"
+#include "gtest/gtest.h"
+#include <memory>
+
+using namespace llvm;
+
+namespace {
+
+TEST(MergeFunctions, TrueOutputModuleTest) {
+ LLVMContext Ctx;
+ SMDiagnostic Err;
+ std::unique_ptr<Module> M(parseAssemblyString(R"invalid(
+ @.str = private unnamed_addr constant [10 x i8] c"On f: %d\0A\00", align 1
+ @.str.1 = private unnamed_addr constant [13 x i8] c"On main: %d\0A\00", align 1
+
+ define dso_local i32 @f(i32 noundef %arg) #0 {
+ entry:
+ %add109 = call i32 @_slice_add10(i32 %arg)
+ %call = call i32 (ptr, ...) @printf(ptr noundef @.str, i32 noundef %add109)
+ ret i32 %add109
+ }
+
+ declare i32 @printf(ptr noundef, ...) #1
+
+ define dso_local i32 @main(i32 noundef %argc, ptr noundef %argv) #0 {
+ entry:
+ %add99 = call i32 @_slice_add10(i32 %argc)
+ %call = call i32 @f(i32 noundef 2)
+ %sub = sub nsw i32 %call, 6
+ %call10 = call i32 (ptr, ...) @printf(ptr noundef @.str.1, i32 noundef %add99)
+ ret i32 %add99
+ }
+
+ define internal i32 @_slice_add10(i32 %arg) #2 {
+ sliceclone_entry:
+ %0 = mul nsw i32 %arg, %arg
+ %1 = mul nsw i32 %0, 2
+ %2 = mul nsw i32 %1, 2
+ %3 = mul nsw i32 %2, 2
+ %4 = add nsw i32 %3, 2
+ ret i32 %4
+ }
+
+ define internal i32 @_slice_add10_alt(i32 %arg) #2 {
+ sliceclone_entry:
+ %0 = mul nsw i32 %arg, %arg
+ %1 = mul nsw i32 %0, 2
+ %2 = mul nsw i32 %1, 2
+ %3 = mul nsw i32 %2, 2
+ %4 = add nsw i32 %3, 2
+ ret i32 %4
+ }
+
+ attributes #0 = { noinline nounwind uwtable "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
+ attributes #1 = { "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
+ attributes #2 = { nounwind willreturn }
+ )invalid",
+ Err, Ctx));
+
+ // Expects true after merging _slice_add10 and _slice_add10_alt
+ EXPECT_TRUE(MergeFunctionsPass::runOnModule(*M));
+}
+
+TEST(MergeFunctions, TrueOutputFunctionsTest) {
+ LLVMContext Ctx;
+ SMDiagnostic Err;
+ std::unique_ptr<Module> M(parseAssemblyString(R"invalid(
+ @.str = private unnamed_addr constant [10 x i8] c"On f: %d\0A\00", align 1
+ @.str.1 = private unnamed_addr constant [13 x i8] c"On main: %d\0A\00", align 1
+
+ define dso_local i32 @f(i32 noundef %arg) #0 {
+ entry:
+ %add109 = call i32 @_slice_add10(i32 %arg)
+ %call = call i32 (ptr, ...) @printf(ptr noundef @.str, i32 noundef %add109)
+ ret i32 %add109
+ }
+
+ declare i32 @printf(ptr noundef, ...) #1
+
+ define dso_local i32 @main(i32 noundef %argc, ptr noundef %argv) #0 {
+ entry:
+ %add99 = call i32 @_slice_add10(i32 %argc)
+ %call = call i32 @f(i32 noundef 2)
+ %sub = sub nsw i32 %call, 6
+ %call10 = call i32 (ptr, ...) @printf(ptr noundef @.str.1, i32 noundef %add99)
+ ret i32 %add99
+ }
+
+ define internal i32 @_slice_add10(i32 %arg) #2 {
+ sliceclone_entry:
+ %0 = mul nsw i32 %arg, %arg
+ %1 = mul nsw i32 %0, 2
+ %2 = mul nsw i32 %1, 2
+ %3 = mul nsw i32 %2, 2
+ %4 = add nsw i32 %3, 2
+ ret i32 %4
+ }
+
+ define internal i32 @_slice_add10_alt(i32 %arg) #2 {
+ sliceclone_entry:
+ %0 = mul nsw i32 %arg, %arg
+ %1 = mul nsw i32 %0, 2
+ %2 = mul nsw i32 %1, 2
+ %3 = mul nsw i32 %2, 2
+ %4 = add nsw i32 %3, 2
+ ret i32 %4
+ }
+
+ attributes #0 = { noinline nounwind uwtable "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
+ attributes #1 = { "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
+ attributes #2 = { nounwind willreturn }
+ )invalid",
+ Err, Ctx));
+
+ std::set<Function *> FunctionsSet;
+ for (Function &F : *M)
+ FunctionsSet.insert(&F);
+
+ std::pair<bool, std::map<Function *, Function *>> MergeResult =
+ MergeFunctionsPass::runOnFunctions(FunctionsSet);
+
+ // Expects true after merging _slice_add10 and _slice_add10_alt
+ EXPECT_TRUE(MergeResult.first);
+
+ // Expects that both functions (_slice_add10 and _slice_add10_alt)
+ // be mapped to the same new function
+ EXPECT_TRUE(MergeResult.second.size() > 0);
+ std::map<Function *, Function *> DelToNew = MergeResult.second;
+ Function *NewFunction = M->getFunction("_slice_add10");
+ for (auto P : DelToNew)
+ if (P.second)
+ EXPECT_EQ(P.second, NewFunction);
+}
+
+TEST(MergeFunctions, FalseOutputModuleTest) {
+ LLVMContext Ctx;
+ SMDiagnostic Err;
+ std::unique_ptr<Module> M(parseAssemblyString(R"invalid(
+ @.str = private unnamed_addr constant [10 x i8] c"On f: %d\0A\00", align 1
+ @.str.1 = private unnamed_addr constant [13 x i8] c"On main: %d\0A\00", align 1
+
+ define dso_local i32 @f(i32 noundef %arg) #0 {
+ entry:
+ %add109 = call i32 @_slice_add10(i32 %arg)
+ %call = call i32 (ptr, ...) @printf(ptr noundef @.str, i32 noundef %add109)
+ ret i32 %add109
+ }
+
+ declare i32 @printf(ptr noundef, ...) #1
+
+ define dso_local i32 @main(i32 noundef %argc, ptr noundef %argv) #0 {
+ entry:
+ %add99 = call i32 @_slice_add10(i32 %argc)
+ %call = call i32 @f(i32 noundef 2)
+ %sub = sub nsw i32 %call, 6
+ %call10 = call i32 (ptr, ...) @printf(ptr noundef @.str.1, i32 noundef %add99)
+ ret i32 %add99
+ }
+
+ define internal i32 @_slice_add10(i32 %arg) #2 {
+ sliceclone_entry:
+ %0 = mul nsw i32 %arg, %arg
+ %1 = mul nsw i32 %0, 2
+ %2 = mul nsw i32 %1, 2
+ %3 = mul nsw i32 %2, 2
+ %4 = add nsw i32 %3, 2
+ ret i32 %4
+ }
+
+ define internal i32 @_slice_add10_alt(i32 %arg) #2 {
+ sliceclone_entry:
+ %0 = mul nsw i32 %arg, %arg
+ %1 = mul nsw i32 %0, 2
+ %2 = mul nsw i32 %1, 2
+ %3 = mul nsw i32 %2, 2
+ %4 = add nsw i32 %3, 2
+ ret i32 %0
+ }
+
+ attributes #0 = { noinline nounwind uwtable "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
+ attributes #1 = { "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
+ attributes #2 = { nounwind willreturn }
+ )invalid",
+ Err, Ctx));
+
+ // Expects false after trying to merge _slice_add10 and _slice_add10_alt
+ EXPECT_FALSE(MergeFunctionsPass::runOnModule(*M));
+}
+
+TEST(MergeFunctions, FalseOutputFunctionsTest) {
+ LLVMContext Ctx;
+ SMDiagnostic Err;
+ std::unique_ptr<Module> M(parseAssemblyString(R"invalid(
+ @.str = private unnamed_addr constant [10 x i8] c"On f: %d\0A\00", align 1
+ @.str.1 = private unnamed_addr constant [13 x i8] c"On main: %d\0A\00", align 1
+
+ define dso_local i32 @f(i32 noundef %arg) #0 {
+ entry:
+ %add109 = call i32 @_slice_add10(i32 %arg)
+ %call = call i32 (ptr, ...) @printf(ptr noundef @.str, i32 noundef %add109)
+ ret i32 %add109
+ }
+
+ declare i32 @printf(ptr noundef, ...) #1
+
+ define dso_local i32 @main(i32 noundef %argc, ptr noundef %argv) #0 {
+ entry:
+ %add99 = call i32 @_slice_add10(i32 %argc)
+ %call = call i32 @f(i32 noundef 2)
+ %sub = sub nsw i32 %call, 6
+ %call10 = call i32 (ptr, ...) @printf(ptr noundef @.str.1, i32 noundef %add99)
+ ret i32 %add99
+ }
+
+ define internal i32 @_slice_add10(i32 %arg) #2 {
+ sliceclone_entry:
+ %0 = mul nsw i32 %arg, %arg
+ %1 = mul nsw i32 %0, 2
+ %2 = mul nsw i32 %1, 2
+ %3 = mul nsw i32 %2, 2
+ %4 = add nsw i32 %3, 2
+ ret i32 %4
+ }
+
+ define internal i32 @_slice_add10_alt(i32 %arg) #2 {
+ sliceclone_entry:
+ %0 = mul nsw i32 %arg, %arg
+ %1 = mul nsw i32 %0, 2
+ %2 = mul nsw i32 %1, 2
+ %3 = mul nsw i32 %2, 2
+ %4 = add nsw i32 %3, 2
+ ret i32 %0
+ }
+
+ attributes #0 = { noinline nounwind uwtable "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
+ attributes #1 = { "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
+ attributes #2 = { nounwind willreturn }
+ )invalid",
+ Err, Ctx));
+
+ std::set<Function *> FunctionsSet;
+ for (Function &F : *M)
+ FunctionsSet.insert(&F);
+
+ std::pair<bool, std::map<Function *, Function *>> MergeResult =
+ MergeFunctionsPass::runOnFunctions(FunctionsSet);
+
+ for (auto P : MergeResult.second)
+ std::cout << P.first << " " << P.second << "\n";
+
+ // Expects false after trying to merge _slice_add10 and _slice_add10_alt
+ EXPECT_FALSE(MergeResult.first);
+
+ // Expects empty map
+ EXPECT_EQ(MergeResult.second.size(), 0u);
+}
+
+} // namespace
\ No newline at end of file
diff --git a/llvm/utils/gn/secondary/llvm/unittests/Transforms/Utils/BUILD.gn b/llvm/utils/gn/secondary/llvm/unittests/Transforms/Utils/BUILD.gn
index 380ed71a2bc010..fcea55c91f083c 100644
--- a/llvm/utils/gn/secondary/llvm/unittests/Transforms/Utils/BUILD.gn
+++ b/llvm/utils/gn/secondary/llvm/unittests/Transforms/Utils/BUILD.gn
@@ -27,6 +27,7 @@ unittest("UtilsTests") {
"LoopUtilsTest.cpp",
"MemTransferLowering.cpp",
"ModuleUtilsTest.cpp",
+ "MergeFunctionsTest.cpp",
"ProfDataUtilTest.cpp",
"SSAUpdaterBulkTest.cpp",
"ScalarEvolutionExpanderTest.cpp",
|
|
Ping |
You can test this locally with the following command:git-clang-format --diff 9043bdbce4ab1c6f07e72ddfedf7165bdf2b3e40 ce17acc603d32b1033c650083f00421fea4637c1 --extensions cpp,h -- llvm/unittests/Transforms/Utils/MergeFunctionsTest.cpp llvm/include/llvm/Transforms/IPO/MergeFunctions.h llvm/lib/Transforms/IPO/MergeFunctions.cppView the diff from clang-format here.diff --git a/llvm/unittests/Transforms/Utils/MergeFunctionsTest.cpp b/llvm/unittests/Transforms/Utils/MergeFunctionsTest.cpp
index 696c5391ef..2f86ec6ed7 100644
--- a/llvm/unittests/Transforms/Utils/MergeFunctionsTest.cpp
+++ b/llvm/unittests/Transforms/Utils/MergeFunctionsTest.cpp
@@ -1,5 +1,5 @@
//===- MergeFunctionsTest.cpp - Unit tests for
-//MergeFunctionsPass-----------===//
+// MergeFunctionsPass-----------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
|
Casperento
force-pushed
the
merge-functions
branch
from
7285b40
to
69e1cc7
Compare
|
Requesting for reviewers |
SLTozer
reviewed
Nov 5, 2024
| @@ -467,6 +484,47 @@ bool MergeFunctions::runOnModule(Module &M) { | |||
| return Changed; | |||
| } | |||
|
|
|||
| bool MergeFunctions::runOnFunctions(std::set<Function *> &F) { | |||
There was a problem hiding this comment.
It'd be good if this could be de-duplicated with MergeFunctions::runOnModule - the implementation looks to be identical, apart from iterating over a std::set<Function *> rather than a Module, and not having the collectUsedGlobalVariables step at the start (it's not immediately clear to me if this step would be meaningful in the runOnFunctions case.
SLTozer
reviewed
Nov 5, 2024
| @@ -297,17 +300,31 @@ class MergeFunctions { | |||
| // dangling iterators into FnTree. The invariant that preserves this is that | |||
| // there is exactly one mapping F -> FN for each FunctionNode FN in FnTree. | |||
| DenseMap<AssertingVH<Function>, FnTreeType::iterator> FNodesInTree; | |||
|
|
|||
| /// Deleted-New functions mapping | |||
| std::map<Function *, Function *> DelToNewMap; | |||
There was a problem hiding this comment.
This (and probably other maps here) could be a DenseMap instead, unless there's a reason this needs std::map. I think it could also be a returned value from runOnFunctions rather than a member, but that's mostly subjective.
SLTozer
reviewed
Nov 15, 2024
Comment on lines
1
to
2
| //===- MergeFunctionsTest.cpp - Unit tests for | ||
| // MergeFunctionsPass-----------===// |
There was a problem hiding this comment.
Nit, this should all be on one line.
Suggested change
| //===- MergeFunctionsTest.cpp - Unit tests for | |
| // MergeFunctionsPass-----------===// | |
| //===- MergeFunctionsTest.cpp - Unit tests for MergeFunctionsPass ---------===// |
There was a problem hiding this comment.
clang format keeps changing that to two lines: 1fdf94e.
There was a problem hiding this comment.
Clang-format will format to a max column with of 80 - for the file header comments, you should change the padding hyphens so that it's exactly 80 characters wide (example in the github suggestion).
There was a problem hiding this comment.
Oosp, my bad. Fixed it! d193a63
Comment on lines
435
to
440
| if constexpr (std::is_same<FuncContainer, Module>::value) { | ||
| SmallVector<GlobalValue *, 4> UsedV; | ||
| collectUsedGlobalVariables(M, UsedV, /*CompilerUsed=*/false); | ||
| collectUsedGlobalVariables(M, UsedV, /*CompilerUsed=*/true); | ||
| Used.insert(UsedV.begin(), UsedV.end()); | ||
| } |
There was a problem hiding this comment.
Since this runs at the top of the function and is only being used for the module case, which is only called from run, this could be moved into run before the call to runOnModule.
Comment on lines
445
to
456
| if constexpr (std::is_same<FuncContainer, std::set<Function *>>::value) { | ||
| for (Function *Func : M) { | ||
| if (isEligibleForMerging(*Func)) { | ||
| HashedFuncs.push_back({StructuralHash(*Func), Func}); | ||
| } | ||
| } | ||
| } | ||
| if constexpr (std::is_same<FuncContainer, Module>::value) { | ||
| for (Function &Func : M) { | ||
| if (isEligibleForMerging(Func)) { | ||
| HashedFuncs.push_back({StructuralHash(Func), &Func}); | ||
| } |
There was a problem hiding this comment.
Small suggestion, but this could be further simplified (and made less dependent on the exact types passed into it) by using one loop with an auto type and extracting a Function* from that. In particular, right now if the user changed the variable type being passed in from std::set to llvm::DenseSet or some similar change, this function would successfully compile but neither of these blocks would be entered. I think the simplest way would be to create a pair of functions above this one:
inline Function *asPtr(Function *Fn) { return Fn; }
inline Function *asPtr(Function &Fn) { return &Fn; }
...
for (auto Func : M) {
Function *FuncPtr = asPtr(Func);
Or, you could do it inline with if constexpr:
Function *FuncPtr;
if constexpr (std::is_reference_type_v<decltype(Func)>)
FuncPtr = &Func;
else
FuncPtr = Func;
| @@ -409,24 +428,36 @@ static bool isEligibleForMerging(Function &F) { | |||
| !hasDistinctMetadataIntrinsic(F); | |||
| } | |||
|
|
|||
| bool MergeFunctions::runOnModule(Module &M) { | |||
| template <typename FuncContainer> | |||
| bool MergeFunctions::runOnModule(FuncContainer &M) { | |||
There was a problem hiding this comment.
The name should probably change since it now takes different types.
Suggested change
| bool MergeFunctions::runOnModule(FuncContainer &M) { | |
| bool MergeFunctions::run(FuncContainer &M) { |
| } | ||
| } | ||
|
|
||
| llvm::stable_sort(HashedFuncs, less_first()); | ||
| stable_sort(HashedFuncs, less_first()); |
There was a problem hiding this comment.
Is there a reason for removing the llvm:: namespace qualifiers here and elsewhere in this file?
Comment on lines
122
to
124
| attributes #0 = { noinline nounwind uwtable "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" } | ||
| attributes #1 = { "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" } | ||
| attributes #2 = { nounwind willreturn } |
There was a problem hiding this comment.
Any of these attributes that aren't needed for the test could be deleted (and the same for all subsequent tests).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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 modification will enable the usage of
MergeFunctionsas a standalone library. Currently,MergeFunctionscan only be applied to an entire module. By adopting this change, developers will gain the flexibility to reuse theMergeFunctionscode within their own projects, choosing which functions to merge; hence, promoting code reusability. Notice that this modification will not break backward compatibility, becauseMergeFunctionswill still work as a pass after the modification.Summary of Changes:
MergeFunctionsPassto allow running the pass over a set of function pointers.Module.