[llvm] [BOLT] Report adjusted program stats from perf2bolt in BAT mode (PR #91683)

[Amir Ayupov via llvm-commits]

https://github.com/aaupov created https://github.com/llvm/llvm-project/pull/91683

In BOLTed binaries split functions are non-simple (due to non-relocation
mode), but the original function is known to be simple and we have a
valid profile for it. Take that into account to accurately report the
number of functions in the profile.

Test Plan: updated bolt-address-translation-yaml.test


>From 1171fdde14bdb4904893594669df1ca8c67041ec Mon Sep 17 00:00:00 2001
From: Amir Ayupov <aaupov at fb.com>
Date: Wed, 8 May 2024 11:57:18 -0700
Subject: [PATCH] [BOLT] Report adjusted program stats from perf2bolt in BAT
 mode

In BOLTed binaries split functions are non-simple (due to non-relocation
mode), but the original function is known to be simple and we have a
valid profile for it. Take that into account to accurately report the
number of functions in the profile.

Test Plan: updated bolt-address-translation-yaml.test
---
 bolt/include/bolt/Passes/BinaryPasses.h          |  5 ++++-
 bolt/lib/Passes/BinaryPasses.cpp                 | 12 +++++++++++-
 bolt/lib/Profile/DataAggregator.cpp              |  3 ++-
 bolt/test/X86/bolt-address-translation-yaml.test |  1 +
 4 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/bolt/include/bolt/Passes/BinaryPasses.h b/bolt/include/bolt/Passes/BinaryPasses.h
index 5d7692559eda8..9b4c690018171 100644
--- a/bolt/include/bolt/Passes/BinaryPasses.h
+++ b/bolt/include/bolt/Passes/BinaryPasses.h
@@ -16,6 +16,7 @@
 #include "bolt/Core/BinaryContext.h"
 #include "bolt/Core/BinaryFunction.h"
 #include "bolt/Core/DynoStats.h"
+#include "bolt/Profile/BoltAddressTranslation.h"
 #include "llvm/Support/CommandLine.h"
 #include <atomic>
 #include <set>
@@ -399,8 +400,10 @@ class PrintProfileStats : public BinaryFunctionPass {
 /// Prints a list of the top 100 functions sorted by a set of
 /// dyno stats categories.
 class PrintProgramStats : public BinaryFunctionPass {
+  BoltAddressTranslation *BAT = nullptr;
 public:
-  explicit PrintProgramStats() : BinaryFunctionPass(false) {}
+  explicit PrintProgramStats(BoltAddressTranslation *BAT = nullptr)
+      : BinaryFunctionPass(false), BAT(BAT) {}
 
   const char *getName() const override { return "print-stats"; }
   bool shouldPrint(const BinaryFunction &) const override { return false; }
diff --git a/bolt/lib/Passes/BinaryPasses.cpp b/bolt/lib/Passes/BinaryPasses.cpp
index c0ba73108f577..666020bc86ba2 100644
--- a/bolt/lib/Passes/BinaryPasses.cpp
+++ b/bolt/lib/Passes/BinaryPasses.cpp
@@ -1378,9 +1378,19 @@ Error PrintProgramStats::runOnFunctions(BinaryContext &BC) {
     if (Function.isPLTFunction())
       continue;
 
+    // Adjustment for BAT mode: the profile for BOLT split fragments is combined
+    // so only count the hot fragment.
+    const uint64_t Address = Function.getAddress();
+    bool IsHotParentOfBOLTSplitFunction = !Function.getFragments().empty() &&
+                                          BAT && BAT->isBATFunction(Address) &&
+                                          !BAT->fetchParentAddress(Address);
+
     ++NumRegularFunctions;
 
-    if (!Function.isSimple()) {
+    // In BOLTed binaries split functions are non-simple (due to non-relocation
+    // mode), but the original function is known to be simple and we have a
+    // valid profile for it.
+    if (!Function.isSimple() && !IsHotParentOfBOLTSplitFunction) {
       if (Function.hasProfile())
         ++NumNonSimpleProfiledFunctions;
       continue;
diff --git a/bolt/lib/Profile/DataAggregator.cpp b/bolt/lib/Profile/DataAggregator.cpp
index 5108392c824c1..4d98d35210b64 100644
--- a/bolt/lib/Profile/DataAggregator.cpp
+++ b/bolt/lib/Profile/DataAggregator.cpp
@@ -612,7 +612,8 @@ Error DataAggregator::readProfile(BinaryContext &BC) {
         if (std::error_code EC = writeBATYAML(BC, opts::SaveProfile))
           report_error("cannot create output data file", EC);
     }
-    BC.logBOLTErrorsAndQuitOnFatal(PrintProgramStats().runOnFunctions(BC));
+    PrintProgramStats PPS(BAT);
+    BC.logBOLTErrorsAndQuitOnFatal(PPS.runOnFunctions(BC));
   }
 
   return Error::success();
diff --git a/bolt/test/X86/bolt-address-translation-yaml.test b/bolt/test/X86/bolt-address-translation-yaml.test
index af24c3d84a0f1..61b65852b2097 100644
--- a/bolt/test/X86/bolt-address-translation-yaml.test
+++ b/bolt/test/X86/bolt-address-translation-yaml.test
@@ -23,6 +23,7 @@ WRITE-BAT-CHECK: BOLT-INFO: BAT section size (bytes): 384
 READ-BAT-CHECK-NOT: BOLT-ERROR: unable to save profile in YAML format for input file processed by BOLT
 READ-BAT-CHECK: BOLT-INFO: Parsed 5 BAT entries
 READ-BAT-CHECK: PERF2BOLT: read 79 aggregated LBR entries
+READ-BAT-CHECK: BOLT-INFO: 5 out of 21 functions in the binary (23.8%) have non-empty execution profile
 
 YAML-BAT-CHECK:      functions:
 # Function not covered by BAT - has insns in basic block