[clang] [llvm] Optimize Module Dependency Handling for Efficient Memory Usage (PR #132294)

[Ayush Pareek via llvm-commits]

https://github.com/ayushpareek2003 updated https://github.com/llvm/llvm-project/pull/132294

>From 53a31ec6ed3d9ace8e7d822c2b2de4f6f58ece8d Mon Sep 17 00:00:00 2001
From: Ayush Pareek <AYUSHPAREEK1980 at GMAIL.COM>
Date: Fri, 21 Mar 2025 03:44:05 +0530
Subject: [PATCH 1/3] Optimize Module Dependency Handling for Efficient Memory
 Usage

-Optimized addModuleFiles functions for both CompilerInvocation and CowCompilerInvocation to reduce redundant function calls and improve efficiency

-Introduced memory preallocation using reserve() when eager load is enabled to reduce reallocation overhead

-Used try_emplace() instead of insert() for PrebuiltModuleFiles to avoid unnecessary overwrites
---
 .../DependencyScanning/ModuleDepCollector.cpp | 61 ++++++++++++-------
 1 file changed, 39 insertions(+), 22 deletions(-)

diff --git a/clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp b/clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp
index d715ef874e002..e172285c4b32d 100644
--- a/clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp
+++ b/clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp
@@ -393,41 +393,58 @@ void ModuleDepCollector::addModuleMapFiles(
     CompilerInvocation &CI, ArrayRef<ModuleID> ClangModuleDeps) const {
   if (Service.shouldEagerLoadModules())
     return; // Only pcm is needed for eager load.
-
+  
+  // Preallocate memory to avoid multiple allocations
+  CI.getFrontendOpts().ModuleMapFiles.reserve(
+      CI.getFrontendOpts().ModuleMapFiles.size() + ClangModuleDeps.size());
   for (const ModuleID &MID : ClangModuleDeps) {
-    ModuleDeps *MD = ModuleDepsByID.lookup(MID);
-    assert(MD && "Inconsistent dependency info");
-    CI.getFrontendOpts().ModuleMapFiles.push_back(MD->ClangModuleMapFile);
+    if (ModuleDeps *MD = ModuleDepsByID.lookup(MID)) { // Single lookup
+      assert(MD && "Inconsistent dependency info");
+      CI.getFrontendOpts().ModuleMapFiles.emplace_back(MD->ClangModuleMapFile);
+    }
   }
 }
 
 void ModuleDepCollector::addModuleFiles(
     CompilerInvocation &CI, ArrayRef<ModuleID> ClangModuleDeps) const {
+  
+    // Preallocate memory if eager load is enabled
+  if (Service.shouldEagerLoadModules()) {
+    CI.getFrontendOpts().ModuleFiles.reserve(
+        CI.getFrontendOpts().ModuleFiles.size() + ClangModuleDeps.size());
+  }
   for (const ModuleID &MID : ClangModuleDeps) {
-    ModuleDeps *MD = ModuleDepsByID.lookup(MID);
-    std::string PCMPath =
-        Controller.lookupModuleOutput(*MD, ModuleOutputKind::ModuleFile);
-
-    if (Service.shouldEagerLoadModules())
-      CI.getFrontendOpts().ModuleFiles.push_back(std::move(PCMPath));
-    else
-      CI.getHeaderSearchOpts().PrebuiltModuleFiles.insert(
-          {MID.ModuleName, std::move(PCMPath)});
+    if (ModuleDeps *MD = ModuleDepsByID.lookup(MID)) {  
+      std::string PCMPath =
+          Controller.lookupModuleOutput(*MD, ModuleOutputKind::ModuleFile);
+      if (Service.shouldEagerLoadModules()) {
+        CI.getFrontendOpts().ModuleFiles.emplace_back(std::move(PCMPath));
+      } else {
+        CI.getHeaderSearchOpts().PrebuiltModuleFiles.try_emplace(
+            MID.ModuleName, std::move(PCMPath));
+      }
+    }
   }
 }
 
 void ModuleDepCollector::addModuleFiles(
     CowCompilerInvocation &CI, ArrayRef<ModuleID> ClangModuleDeps) const {
+    // Preallocation
+  if (Service.shouldEagerLoadModules()) {
+    CI.getMutFrontendOpts().ModuleFiles.reserve(
+        CI.getMutFrontendOpts().ModuleFiles.size() + ClangModuleDeps.size());
+  }
   for (const ModuleID &MID : ClangModuleDeps) {
-    ModuleDeps *MD = ModuleDepsByID.lookup(MID);
-    std::string PCMPath =
-        Controller.lookupModuleOutput(*MD, ModuleOutputKind::ModuleFile);
-
-    if (Service.shouldEagerLoadModules())
-      CI.getMutFrontendOpts().ModuleFiles.push_back(std::move(PCMPath));
-    else
-      CI.getMutHeaderSearchOpts().PrebuiltModuleFiles.insert(
-          {MID.ModuleName, std::move(PCMPath)});
+    if (ModuleDeps *MD = ModuleDepsByID.lookup(MID)) {  
+      std::string PCMPath =
+          Controller.lookupModuleOutput(*MD, ModuleOutputKind::ModuleFile);
+      if (Service.shouldEagerLoadModules()) {
+        CI.getMutFrontendOpts().ModuleFiles.emplace_back(std::move(PCMPath));
+      } else {
+        CI.getMutHeaderSearchOpts().PrebuiltModuleFiles.try_emplace(
+            MID.ModuleName, std::move(PCMPath));
+      }
+    }
   }
 }
 

>From 153cc53186c60454154e205e48249177b1695eff Mon Sep 17 00:00:00 2001
From: Ayush Pareek <AYUSHPAREEK1980 at GMAIL.COM>
Date: Mon, 24 Mar 2025 03:02:03 +0530
Subject: [PATCH 2/3] Update README.md

---
 bolt/README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/bolt/README.md b/bolt/README.md
index fe54bd82a356a..44ef33bcebeb0 100644
--- a/bolt/README.md
+++ b/bolt/README.md
@@ -7,6 +7,7 @@ An overview of the ideas implemented in BOLT along with a discussion of its
 potential and current results is available in
 [CGO'19 paper](https://research.fb.com/publications/bolt-a-practical-binary-optimizer-for-data-centers-and-beyond/).
 
+
 ## Input Binary Requirements
 
 BOLT operates on X86-64 and AArch64 ELF binaries. At the minimum, the binaries

>From 1b853396809777fd9e7c4849a3a9fc8ac56133a0 Mon Sep 17 00:00:00 2001
From: Ayush Pareek <AYUSHPAREEK1980 at GMAIL.COM>
Date: Mon, 24 Mar 2025 03:04:55 +0530
Subject: [PATCH 3/3] Update README.md

---
 bolt/README.md | 1 -
 1 file changed, 1 deletion(-)

diff --git a/bolt/README.md b/bolt/README.md
index 44ef33bcebeb0..fe54bd82a356a 100644
--- a/bolt/README.md
+++ b/bolt/README.md
@@ -7,7 +7,6 @@ An overview of the ideas implemented in BOLT along with a discussion of its
 potential and current results is available in
 [CGO'19 paper](https://research.fb.com/publications/bolt-a-practical-binary-optimizer-for-data-centers-and-beyond/).
 
-
 ## Input Binary Requirements
 
 BOLT operates on X86-64 and AArch64 ELF binaries. At the minimum, the binaries