Supporting -f(no)-reorder-functions flag, llvm side change

[twoh]

Summary

When profile data is given, .hot/.unlikely section prefix is
added to hot/cold functions for linker to improve code locality. GCC
controls this behavior with '-f(no)-reorder-functions' flag, while LLVM
uses opt tool option '-profile-guided-section-prefix=true/false'. This
patch is for LLVM to support the same driver flag with GCC. Clang side
patch with test cases will follow.

[twoh]

https://reviews.llvm.org/D34796 is clang side change.

[twoh]

Ping. Thanks!

[davidxl]

lib/CodeGen/CodeGenPrepare.cpp

306

Can TM be null?

[twoh]

@davidxl I think it is theoretically possible, if the if branch is not taken on line 294. Did I miss something? Thanks!

[xbolva00]

Ping, plesse rebase

[lebedev.ri]

Test?

[twoh]

Rebase. Tests are provided in the clang counterpart (D34796).

include/llvm/Target/TargetMachine.h

public:
   void setFastISel(bool Enable) { Options.EnableFastISel = Enable; }
   bool getO0WantsFastISel() { return O0WantsFastISel; }
   void setO0WantsFastISel(bool Enable) { O0WantsFastISel = Enable; }
 
   bool shouldPrintMachineCode() const { return Options.PrintMachineCode; }
 
   bool getUniqueSectionNames() const { return Options.UniqueSectionNames; }
 
+  /// Return true if profile-guided section prefix that indicates hotness of
+  /// the function should be added, corresponds to -freorder-functions.
+  bool getReorderFunctions() const { return Options.ReorderFunctions; }
+
   /// Return true if data objects should be emitted into their own section,
   /// corresponds to -fdata-sections.
   bool getDataSections() const {
     return Options.DataSections;
   }
 
   /// Return true if functions should be emitted into their own section,
   /// corresponding to -ffunction-sections.

include/llvm/Target/TargetOptions.h

TargetOptions()
         : PrintMachineCode(false), UnsafeFPMath(false), NoInfsFPMath(false),
           NoNaNsFPMath(false), NoTrappingFPMath(false),
           NoSignedZerosFPMath(false),
           HonorSignDependentRoundingFPMathOption(false), NoZerosInBSS(false),
           GuaranteedTailCallOpt(false), StackSymbolOrdering(true),
           EnableFastISel(false), UseInitArray(false),
           DisableIntegratedAS(false), RelaxELFRelocations(false),
           FunctionSections(false), DataSections(false),
-          UniqueSectionNames(true), TrapUnreachable(false), EmulatedTLS(false),
-          EnableIPRA(false) {}
+          UniqueSectionNames(true), ReorderFunctions(true),
+          TrapUnreachable(false), EmulatedTLS(false), EnableIPRA(false) {}
 
     /// PrintMachineCode - This flag is enabled when the -print-machineinstrs
     /// option is specified on the command line, and should enable debugging
     /// output from the code generator.
     unsigned PrintMachineCode : 1;
 
     /// DisableFramePointerElim - This returns true if frame pointer elimination
     /// optimization should be disabled for the given machine function.
public:
     /// Emit functions into separate sections.
     unsigned FunctionSections : 1;
 
     /// Emit data into separate sections.
     unsigned DataSections : 1;
 
     unsigned UniqueSectionNames : 1;
 
+    /// Add section prefix for hot/cold functions.
+    unsigned ReorderFunctions : 1;
+
     /// Emit target-specific trap instruction for 'unreachable' IR instructions.
     unsigned TrapUnreachable : 1;
 
     /// EmulatedTLS - This flag enables emulated TLS model, using emutls
     /// function in the runtime library..
     unsigned EmulatedTLS : 1;
 
     /// This flag enables InterProcedural Register Allocation (IPRA).

lib/CodeGen/CodeGenPrepare.cpp

if (auto *TPC = getAnalysisIfAvailable<TargetPassConfig>()) {
     TLI = SubtargetInfo->getTargetLowering();
     TRI = SubtargetInfo->getRegisterInfo();
   }
   TLInfo = &getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
   TTI = &getAnalysis<TargetTransformInfoWrapperPass>().getTTI(F);
   LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
   OptSize = F.optForSize();
 
-  if (ProfileGuidedSectionPrefix) {
-    ProfileSummaryInfo *PSI =
-        getAnalysis<ProfileSummaryInfoWrapperPass>().getPSI();
+  bool AddProfileGuidedSectionPrefix =
+      (!TM || ProfileGuidedSectionPrefix.getNumOccurrences())
+          ? ProfileGuidedSectionPrefix
+          : TM->getReorderFunctions();
+  if (AddProfileGuidedSectionPrefix) {
+    if (auto *PSI = getAnalysis<ProfileSummaryInfoWrapperPass>().getPSI()) {
       if (PSI->isFunctionHotInCallGraph(&F))
         F.setSectionPrefix(".hot");
       else if (PSI->isFunctionColdInCallGraph(&F))
         F.setSectionPrefix(".unlikely");
     }
+  }
 
   /// This optimization identifies DIV instructions that can be
   /// profitably bypassed and carried out with a shorter, faster divide.
   if (!OptSize && TLI && TLI->isSlowDivBypassed()) {
     const DenseMap<unsigned int, unsigned int> &BypassWidths =
        TLI->getBypassSlowDivWidths();
     BasicBlock* BB = &*F.begin();
     while (BB != nullptr) {