[llvm] [nfc] Clarify when the various PGO instrumentation passes run (PR #92330)

[Mircea Trofin via llvm-commits]

https://github.com/mtrofin updated https://github.com/llvm/llvm-project/pull/92330

>From b719481c3c22015c69af0d949457ceb6ea47c0c0 Mon Sep 17 00:00:00 2001
From: Mircea Trofin <mtrofin at google.com>
Date: Wed, 15 May 2024 17:11:56 -0700
Subject: [PATCH] [nfc] Clarify when the various PGO instrumentation passes run

The code seems easier to read if it's centered on what the user
wants rather than combinations of whatever internal variables.
---
 llvm/include/llvm/Passes/PassBuilder.h   |  1 +
 llvm/lib/Passes/PassBuilderPipelines.cpp | 68 ++++++++++++++----------
 2 files changed, 41 insertions(+), 28 deletions(-)

diff --git a/llvm/include/llvm/Passes/PassBuilder.h b/llvm/include/llvm/Passes/PassBuilder.h
index 9b4da1c466f57..8de0f024a4b11 100644
--- a/llvm/include/llvm/Passes/PassBuilder.h
+++ b/llvm/include/llvm/Passes/PassBuilder.h
@@ -723,6 +723,7 @@ class PassBuilder {
                          bool AtomicCounterUpdate, std::string ProfileFile,
                          std::string ProfileRemappingFile,
                          IntrusiveRefCntPtr<vfs::FileSystem> FS);
+  void addPostPGOLoopRotation(ModulePassManager &MPM, OptimizationLevel Level);
 
   // Extension Point callbacks
   SmallVector<std::function<void(FunctionPassManager &, OptimizationLevel)>, 2>
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp b/llvm/lib/Passes/PassBuilderPipelines.cpp
index 72f273972f2f7..dfde877744aa6 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -803,6 +803,20 @@ void PassBuilder::addPreInlinerPasses(ModulePassManager &MPM,
   MPM.addPass(GlobalDCEPass());
 }
 
+void PassBuilder::addPostPGOLoopRotation(ModulePassManager &MPM,
+                                         OptimizationLevel Level) {
+  if (EnablePostPGOLoopRotation) {
+    // Disable header duplication in loop rotation at -Oz.
+    MPM.addPass(createModuleToFunctionPassAdaptor(
+        createFunctionToLoopPassAdaptor(
+            LoopRotatePass(EnableLoopHeaderDuplication ||
+                           Level != OptimizationLevel::Oz),
+            /*UseMemorySSA=*/false,
+            /*UseBlockFrequencyInfo=*/false),
+        PTO.EagerlyInvalidateAnalyses));
+  }
+}
+
 void PassBuilder::addPGOInstrPasses(ModulePassManager &MPM,
                                     OptimizationLevel Level, bool RunProfileGen,
                                     bool IsCS, bool AtomicCounterUpdate,
@@ -824,21 +838,7 @@ void PassBuilder::addPGOInstrPasses(ModulePassManager &MPM,
   // Perform PGO instrumentation.
   MPM.addPass(PGOInstrumentationGen(IsCS));
 
-  if (EnablePostPGOLoopRotation) {
-    // Disable header duplication in loop rotation at -Oz.
-    MPM.addPass(createModuleToFunctionPassAdaptor(
-        createFunctionToLoopPassAdaptor(
-            LoopRotatePass(EnableLoopHeaderDuplication ||
-                           Level != OptimizationLevel::Oz),
-            /*UseMemorySSA=*/false,
-            /*UseBlockFrequencyInfo=*/false),
-        PTO.EagerlyInvalidateAnalyses));
-  }
-
-  if (PGOCtxProfLoweringPass::isContextualIRPGOEnabled()) {
-    MPM.addPass(PGOCtxProfLoweringPass());
-    return;
-  }
+  addPostPGOLoopRotation(MPM, Level);
   // Add the profile lowering pass.
   InstrProfOptions Options;
   if (!ProfileFile.empty())
@@ -1145,29 +1145,41 @@ PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level,
   MPM.addPass(createModuleToFunctionPassAdaptor(std::move(GlobalCleanupPM),
                                                 PTO.EagerlyInvalidateAnalyses));
 
-  // Invoke the pre-inliner passes for instrumentation PGO or MemProf.
-  if (PGOOpt && Phase != ThinOrFullLTOPhase::ThinLTOPostLink &&
-      (PGOOpt->Action == PGOOptions::IRInstr ||
-       PGOOpt->Action == PGOOptions::IRUse || !PGOOpt->MemoryProfile.empty()))
+  const bool IsPreLink = Phase != ThinOrFullLTOPhase::ThinLTOPostLink;
+  const bool IsPGOPreLink = PGOOpt && IsPreLink;
+  const bool IsPGOInstrGen =
+      IsPGOPreLink && PGOOpt->Action == PGOOptions::IRInstr;
+  const bool IsPGOInstrUse =
+      IsPGOPreLink && PGOOpt->Action == PGOOptions::IRUse;
+  const bool IsMemprof = IsPGOPreLink && !PGOOpt->MemoryProfile.empty();
+  // We don't want to mix pgo ctx gen and pgo gen; we also don't currently
+  // enable ctx profiling from the frontend.
+  const bool IsCtxProfGen = !IsPGOInstrGen && IsPreLink &&
+                            PGOCtxProfLoweringPass::isContextualIRPGOEnabled();
+
+  if (IsPGOInstrGen || IsPGOInstrUse || IsMemprof || IsCtxProfGen)
     addPreInlinerPasses(MPM, Level, Phase);
 
   // Add all the requested passes for instrumentation PGO, if requested.
-  if (PGOOpt && Phase != ThinOrFullLTOPhase::ThinLTOPostLink &&
-      (PGOOpt->Action == PGOOptions::IRInstr ||
-       PGOOpt->Action == PGOOptions::IRUse)) {
+  if (IsPGOInstrGen || IsPGOInstrUse) {
     addPGOInstrPasses(MPM, Level,
-                      /*RunProfileGen=*/PGOOpt->Action == PGOOptions::IRInstr,
+                      /*RunProfileGen=*/IsPGOInstrGen,
                       /*IsCS=*/false, PGOOpt->AtomicCounterUpdate,
                       PGOOpt->ProfileFile, PGOOpt->ProfileRemappingFile,
                       PGOOpt->FS);
-    MPM.addPass(PGOIndirectCallPromotion(false, false));
+  } else if (IsCtxProfGen) {
+    MPM.addPass(PGOInstrumentationGen(false));
+    addPostPGOLoopRotation(MPM, Level);
+    MPM.addPass(PGOCtxProfLoweringPass());
   }
-  if (PGOOpt && Phase != ThinOrFullLTOPhase::ThinLTOPostLink &&
-      PGOOpt->CSAction == PGOOptions::CSIRInstr)
+
+  if (IsPGOInstrGen || IsPGOInstrUse || IsCtxProfGen)
+    MPM.addPass(PGOIndirectCallPromotion(false, false));
+
+  if (IsPGOPreLink && PGOOpt->CSAction == PGOOptions::CSIRInstr)
     MPM.addPass(PGOInstrumentationGenCreateVar(PGOOpt->CSProfileGenFile));
 
-  if (PGOOpt && Phase != ThinOrFullLTOPhase::ThinLTOPostLink &&
-      !PGOOpt->MemoryProfile.empty())
+  if (IsMemprof)
     MPM.addPass(MemProfUsePass(PGOOpt->MemoryProfile, PGOOpt->FS));
 
   // Synthesize function entry counts for non-PGO compilation.