Reland "clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM"

This patch contains the basic functionality for reporting potentially
incorrect usage of __builtin_expect() by comparing the developer's
annotation against a collected PGO profile. A more detailed proposal and
discussion appears on the CFE-dev mailing list
(http://lists.llvm.org/pipermail/cfe-dev/2019-July/062971.html) and a
prototype of the initial frontend changes appear here in D65300

We revised the work in D65300 by moving the misexpect check into the
LLVM backend, and adding support for IR and sampling based profiles, in
addition to frontend instrumentation.

We add new misexpect metadata tags to those instructions directly
influenced by the llvm.expect intrinsic (branch, switch, and select)
when lowering the intrinsics. The misexpect metadata contains
information about the expected target of the intrinsic so that we can
check against the correct PGO counter when emitting diagnostics, and the
compiler's values for the LikelyBranchWeight and UnlikelyBranchWeight.
We use these branch weight values to determine when to emit the
diagnostic to the user.

A future patch should address the comment at the top of
LowerExpectIntrisic.cpp to hoist the LikelyBranchWeight and
UnlikelyBranchWeight values into a shared space that can be accessed
outside of the LowerExpectIntrinsic pass. Once that is done, the
misexpect metadata can be updated to be smaller.

In the long term, it is possible to reconstruct portions of the
misexpect metadata from the existing profile data. However, we have
avoided this to keep the code simple, and because some kind of metadata
tag will be required to identify which branch/switch/select instructions
are influenced by the use of llvm.expect

Patch By: paulkirth
Differential Revision: https://reviews.llvm.org/D66324

llvm-svn: 371635

Author: petrhosek

clang/include/clang/Basic/DiagnosticFrontendKinds.td

@@ -275,7 +275,12 @@ def warn_profile_data_missing : Warning<
 def warn_profile_data_unprofiled : Warning<
   "no profile data available for file \"%0\"">,
   InGroup<ProfileInstrUnprofiled>;
-
+def warn_profile_data_misexpect : Warning<
+  "Potential performance regression from use of __builtin_expect(): "
+  "Annotation was correct on %0 of profiled executions.">,
+  BackendInfo,
+  InGroup<MisExpect>,
+  DefaultIgnore;
 } // end of instrumentation issue category
 
 }

clang/include/clang/Basic/DiagnosticGroups.td

@@ -1042,6 +1042,7 @@ def BackendOptimizationFailure : DiagGroup<"pass-failed">;
 def ProfileInstrMissing : DiagGroup<"profile-instr-missing">;
 def ProfileInstrOutOfDate : DiagGroup<"profile-instr-out-of-date">;
 def ProfileInstrUnprofiled : DiagGroup<"profile-instr-unprofiled">;
+def MisExpect : DiagGroup<"misexpect">;
 
 // AddressSanitizer frontend instrumentation remarks.
 def SanitizeAddressRemarks : DiagGroup<"sanitize-address">;

clang/lib/CodeGen/CodeGenAction.cpp

@@ -14,6 +14,7 @@
 #include "clang/AST/ASTContext.h"
 #include "clang/AST/DeclCXX.h"
 #include "clang/AST/DeclGroup.h"
+#include "clang/Basic/DiagnosticFrontend.h"
 #include "clang/Basic/FileManager.h"
 #include "clang/Basic/LangStandard.h"
 #include "clang/Basic/SourceManager.h"
@@ -365,6 +366,9 @@ namespace clang {
     bool StackSizeDiagHandler(const llvm::DiagnosticInfoStackSize &D);
     /// Specialized handler for unsupported backend feature diagnostic.
     void UnsupportedDiagHandler(const llvm::DiagnosticInfoUnsupported &D);
+    /// Specialized handler for misexpect warnings.
+    /// Note that misexpect remarks are emitted through ORE
+    void MisExpectDiagHandler(const llvm::DiagnosticInfoMisExpect &D);
     /// Specialized handlers for optimization remarks.
     /// Note that these handlers only accept remarks and they always handle
     /// them.
@@ -617,6 +621,25 @@ void BackendConsumer::UnsupportedDiagHandler(
         << Filename << Line << Column;
 }
 
+void BackendConsumer::MisExpectDiagHandler(
+    const llvm::DiagnosticInfoMisExpect &D) {
+  StringRef Filename;
+  unsigned Line, Column;
+  bool BadDebugInfo = false;
+  FullSourceLoc Loc =
+      getBestLocationFromDebugLoc(D, BadDebugInfo, Filename, Line, Column);
+
+  Diags.Report(Loc, diag::warn_profile_data_misexpect) << D.getMsg().str();
+
+  if (BadDebugInfo)
+    // If we were not able to translate the file:line:col information
+    // back to a SourceLocation, at least emit a note stating that
+    // we could not translate this location. This can happen in the
+    // case of #line directives.
+    Diags.Report(Loc, diag::note_fe_backend_invalid_loc)
+        << Filename << Line << Column;
+}
+
 void BackendConsumer::EmitOptimizationMessage(
     const llvm::DiagnosticInfoOptimizationBase &D, unsigned DiagID) {
   // We only support warnings and remarks.
@@ -787,6 +810,9 @@ void BackendConsumer::DiagnosticHandlerImpl(const DiagnosticInfo &DI) {
   case llvm::DK_Unsupported:
     UnsupportedDiagHandler(cast<DiagnosticInfoUnsupported>(DI));
     return;
+  case llvm::DK_MisExpect:
+    MisExpectDiagHandler(cast<DiagnosticInfoMisExpect>(DI));
+    return;
   default:
     // Plugin IDs are not bound to any value as they are set dynamically.
     ComputeDiagRemarkID(Severity, backend_plugin, DiagID);

clang/lib/Frontend/CompilerInvocation.cpp

@@ -3453,6 +3453,9 @@ bool CompilerInvocation::CreateFromArgs(CompilerInvocation &Res,
     }
   }
 
+  if (Diags.isIgnored(diag::warn_profile_data_misexpect, SourceLocation()))
+    Res.FrontendOpts.LLVMArgs.push_back("-pgo-warn-misexpect");
+
   LangOpts.FunctionAlignment =
       getLastArgIntValue(Args, OPT_function_alignment, 0, Diags);
 

clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext

@@ -0,0 +1,9 @@
+bar
+# Func Hash:
+11262309464
+# Num Counters:
+2
+# Counter Values:
+200000
+2
+

clang/test/Profile/Inputs/misexpect-branch.proftext

@@ -0,0 +1,9 @@
+bar
+# Func Hash:
+45795613684824
+# Num Counters:
+2
+# Counter Values:
+200000
+0
+

clang/test/Profile/Inputs/misexpect-switch-default-only.proftext

@@ -0,0 +1,12 @@
+main
+# Func Hash:
+79676873694057560
+# Num Counters:
+5
+# Counter Values:
+1
+20
+20000
+20000
+20000
+

clang/test/Profile/Inputs/misexpect-switch-default.proftext

@@ -0,0 +1,16 @@
+main
+# Func Hash:
+8712453512413296413
+# Num Counters:
+9
+# Counter Values:
+1
+20000
+20000
+4066
+11889
+0
+0
+4045
+0
+

clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext

@@ -0,0 +1,17 @@
+main
+# Func Hash:
+1965403898329309329
+# Num Counters:
+10
+# Counter Values:
+1
+20
+20000
+20000
+1
+0
+0
+0
+19999
+0
+

clang/test/Profile/Inputs/misexpect-switch.proftext

@@ -0,0 +1,16 @@
+main
+# Func Hash:
+1965403898329309329
+# Num Counters:
+9
+# Counter Values:
+1
+20
+20000
+20000
+12
+26
+0
+0
+19962
+

clang/test/Profile/misexpect-branch-cold.c

@@ -0,0 +1,26 @@
+// Test that misexpect emits no warning when prediction is correct
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-branch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect
+
+// expected-no-diagnostics
+#define likely(x) __builtin_expect(!!(x), 1)
+#define unlikely(x) __builtin_expect(!!(x), 0)
+
+int foo(int);
+int baz(int);
+int buzz();
+
+const int inner_loop = 100;
+const int outer_loop = 2000;
+
+int bar() {
+  int rando = buzz();
+  int x = 0;
+  if (unlikely(rando % (outer_loop * inner_loop) == 0)) {
+    x = baz(rando);
+  } else {
+    x = foo(50);
+  }
+  return x;
+}

clang/test/Profile/misexpect-branch-nonconst-expected-val.c

@@ -0,0 +1,23 @@
+// Test that misexpect emits no warning when condition is not a compile-time constant
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-branch-nonconst-expect-arg.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect
+
+// expected-no-diagnostics
+int foo(int);
+int baz(int);
+int buzz();
+
+const int inner_loop = 100;
+const int outer_loop = 2000;
+
+int bar() {
+  int rando = buzz();
+  int x = 0;
+  if (__builtin_expect(rando % (outer_loop * inner_loop) == 0, buzz())) {
+    x = baz(rando);
+  } else {
+    x = foo(50);
+  }
+  return x;
+}

clang/test/Profile/misexpect-branch-unpredictable.c

@@ -0,0 +1,25 @@
+// Test that misexpect emits no warning when prediction is correct
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-branch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect
+
+// expected-no-diagnostics
+#define unpredictable(x) __builtin_unpredictable(!!(x))
+
+int foo(int);
+int baz(int);
+int buzz();
+
+const int inner_loop = 100;
+const int outer_loop = 2000;
+
+int bar() {
+  int rando = buzz();
+  int x = 0;
+  if (unpredictable(rando % (outer_loop * inner_loop) == 0)) {
+    x = baz(rando);
+  } else {
+    x = foo(50);
+  }
+  return x;
+}

clang/test/Profile/misexpect-branch.c

@@ -0,0 +1,28 @@
+// Test that misexpect detects mis-annotated branches
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-branch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify=imprecise -Wmisexpect
+// RUN: %clang_cc1 %s -O2 -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify=exact -Wmisexpect -debug-info-kind=line-tables-only
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify=foo
+
+// foo-no-diagnostics
+#define likely(x) __builtin_expect(!!(x), 1)
+#define unlikely(x) __builtin_expect(!!(x), 0)
+
+int foo(int);
+int baz(int);
+int buzz();
+
+const int inner_loop = 100;
+const int outer_loop = 2000;
+
+int bar() { // imprecise-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% ({{[0-9]+ / [0-9]+}}) of profiled executions.}}
+  int rando = buzz();
+  int x = 0;
+  if (likely(rando % (outer_loop * inner_loop) == 0)) { // exact-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% ({{[0-9]+ / [0-9]+}}) of profiled executions.}}
+    x = baz(rando);
+  } else {
+    x = foo(50);
+  }
+  return x;
+}

clang/test/Profile/misexpect-switch-default.c

@@ -0,0 +1,40 @@
+// Test that misexpect detects mis-annotated switch statements for default case
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect -debug-info-kind=line-tables-only
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+void init_arry();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+int main() {
+  init_arry();
+  int val = 0;
+  int j;
+  for (j = 0; j < outer_loop * inner_loop; ++j) {
+    unsigned condition = rand() % 5;
+    switch (__builtin_expect(condition, 6)) { // expected-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% ({{[0-9]+ / [0-9]+}}) of profiled executions.}}
+    case 0:
+      val += sum(arry, arry_size);
+      break;
+    case 1:
+    case 2:
+    case 3:
+      break;
+    case 4:
+      val += random_sample(arry, arry_size);
+      break;
+    default:
+      __builtin_unreachable();
+    } // end switch
+  }   // end outer_loop
+
+  return 0;
+}

clang/test/Profile/misexpect-switch-nonconst.c

@@ -0,0 +1,43 @@
+// Test that misexpect emits no warning when switch condition is non-const
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-nonconst.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+void init_arry();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+    for (k = 0; k < inner_loop; ++k) {
+      unsigned condition = rand() % 10000;
+      switch (__builtin_expect(condition, rand())) {
+      case 0:
+        val += sum(arry, arry_size);
+        break;
+      case 1:
+      case 2:
+      case 3:
+      case 4:
+        val += random_sample(arry, arry_size);
+        break;
+      default:
+        __builtin_unreachable();
+      } // end switch
+    }   // end inner_loop
+  }     // end outer_loop
+
+  return 0;
+}

clang/test/Profile/misexpect-switch-only-default-case.c

@@ -0,0 +1,35 @@
+// Test that misexpect emits no warning when there is only one switch case
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default-only.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect -debug-info-kind=line-tables-only
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+void init_arry();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+    for (k = 0; k < inner_loop; ++k) {
+      unsigned condition = rand() % 10000;
+      switch (__builtin_expect(condition, 0)) {
+      default:
+        val += random_sample(arry, arry_size);
+        break;
+      }; // end switch
+    }    // end inner_loop
+  }      // end outer_loop
+
+  return 0;
+}