Reapply "[llvm] Add CalleeTypeIds field to CallSiteInfo" (#150335) (#150990)

This reverts commit 05e08cd.
Adding the missing -mtriple flags in MIR/X86 test files which caused
these tests to fail which was the reason for reverting the patch.

Author: Prabhuk

llvm/include/llvm/CodeGen/CommandFlags.h

@@ -133,6 +133,8 @@ LLVM_ABI bool getEnableStackSizeSection();
 
 LLVM_ABI bool getEnableAddrsig();
 
+LLVM_ABI bool getEnableCallGraphSection();
+
 LLVM_ABI bool getEmitCallSiteInfo();
 
 LLVM_ABI bool getEnableMachineFunctionSplitter();

llvm/include/llvm/CodeGen/MIRYamlMapping.h

@@ -482,6 +482,8 @@ struct CallSiteInfo {
 
   MachineInstrLoc CallLocation;
   std::vector<ArgRegPair> ArgForwardingRegs;
+  /// Numeric callee type identifiers for the callgraph section.
+  std::vector<uint64_t> CalleeTypeIds;
 
   bool operator==(const CallSiteInfo &Other) const {
     return CallLocation.BlockNum == Other.CallLocation.BlockNum &&
@@ -511,6 +513,7 @@ template <> struct MappingTraits<CallSiteInfo> {
     YamlIO.mapRequired("offset", CSInfo.CallLocation.Offset);
     YamlIO.mapOptional("fwdArgRegs", CSInfo.ArgForwardingRegs,
                        std::vector<CallSiteInfo::ArgRegPair>());
+    YamlIO.mapOptional("calleeTypeIds", CSInfo.CalleeTypeIds);
   }
 
   static const bool flow = true;

llvm/include/llvm/CodeGen/MachineFunction.h

@@ -515,6 +515,8 @@ class LLVM_ABI MachineFunction {
   struct CallSiteInfo {
     /// Vector of call argument and its forwarding register.
     SmallVector<ArgRegPair, 1> ArgRegPairs;
+    /// Callee type ids.
+    SmallVector<ConstantInt *, 4> CalleeTypeIds;
   };
 
   struct CalledGlobalInfo {

llvm/include/llvm/Target/TargetOptions.h

@@ -133,10 +133,11 @@ class TargetOptions {
         EmitStackSizeSection(false), EnableMachineOutliner(false),
         EnableMachineFunctionSplitter(false),
         EnableStaticDataPartitioning(false), SupportsDefaultOutlining(false),
-        EmitAddrsig(false), BBAddrMap(false), EmitCallSiteInfo(false),
-        SupportsDebugEntryValues(false), EnableDebugEntryValues(false),
-        ValueTrackingVariableLocations(false), ForceDwarfFrameSection(false),
-        XRayFunctionIndex(true), DebugStrictDwarf(false), Hotpatch(false),
+        EmitAddrsig(false), BBAddrMap(false), EmitCallGraphSection(false),
+        EmitCallSiteInfo(false), SupportsDebugEntryValues(false),
+        EnableDebugEntryValues(false), ValueTrackingVariableLocations(false),
+        ForceDwarfFrameSection(false), XRayFunctionIndex(true),
+        DebugStrictDwarf(false), Hotpatch(false),
         PPCGenScalarMASSEntries(false), JMCInstrument(false),
         EnableCFIFixup(false), MisExpect(false), XCOFFReadOnlyPointers(false),
         VerifyArgABICompliance(true),
@@ -319,6 +320,9 @@ class TargetOptions {
   /// to selectively generate basic block sections.
   std::shared_ptr<MemoryBuffer> BBSectionsFuncListBuf;
 
+  /// Emit section containing call graph metadata.
+  unsigned EmitCallGraphSection : 1;
+
   /// The flag enables call site info production. It is used only for debug
   /// info, and it is restricted only to optimized code. This can be used for
   /// something else, so that should be controlled in the frontend.

llvm/lib/CodeGen/CommandFlags.cpp

@@ -101,6 +101,7 @@ CGOPT(EABI, EABIVersion)
 CGOPT(DebuggerKind, DebuggerTuningOpt)
 CGOPT(bool, EnableStackSizeSection)
 CGOPT(bool, EnableAddrsig)
+CGOPT(bool, EnableCallGraphSection)
 CGOPT(bool, EmitCallSiteInfo)
 CGOPT(bool, EnableMachineFunctionSplitter)
 CGOPT(bool, EnableStaticDataPartitioning)
@@ -461,6 +462,11 @@ codegen::RegisterCodeGenFlags::RegisterCodeGenFlags() {
       cl::init(false));
   CGBINDOPT(EnableAddrsig);
 
+  static cl::opt<bool> EnableCallGraphSection(
+      "call-graph-section", cl::desc("Emit a call graph section"),
+      cl::init(false));
+  CGBINDOPT(EnableCallGraphSection);
+
   static cl::opt<bool> EmitCallSiteInfo(
       "emit-call-site-info",
       cl::desc(
@@ -595,6 +601,7 @@ codegen::InitTargetOptionsFromCodeGenFlags(const Triple &TheTriple) {
   Options.EnableMachineFunctionSplitter = getEnableMachineFunctionSplitter();
   Options.EnableStaticDataPartitioning = getEnableStaticDataPartitioning();
   Options.EmitAddrsig = getEnableAddrsig();
+  Options.EmitCallGraphSection = getEnableCallGraphSection();
   Options.EmitCallSiteInfo = getEmitCallSiteInfo();
   Options.EnableDebugEntryValues = getEnableDebugEntryValues();
   Options.ForceDwarfFrameSection = getForceDwarfFrameSection();

llvm/lib/CodeGen/MIRParser/MIRParser.cpp

@@ -504,13 +504,21 @@ bool MIRParserImpl::initializeCallSiteInfo(
         return error(Error, ArgRegPair.Reg.SourceRange);
       CSInfo.ArgRegPairs.emplace_back(Reg, ArgRegPair.ArgNo);
     }
+    if (!YamlCSInfo.CalleeTypeIds.empty()) {
+      for (auto CalleeTypeId : YamlCSInfo.CalleeTypeIds) {
+        IntegerType *Int64Ty = Type::getInt64Ty(Context);
+        CSInfo.CalleeTypeIds.push_back(ConstantInt::get(Int64Ty, CalleeTypeId,
+                                                        /*isSigned=*/false));
+      }
+    }
 
-    if (TM.Options.EmitCallSiteInfo)
+    if (TM.Options.EmitCallSiteInfo || TM.Options.EmitCallGraphSection)
       MF.addCallSiteInfo(&*CallI, std::move(CSInfo));
   }
 
-  if (YamlMF.CallSitesInfo.size() && !TM.Options.EmitCallSiteInfo)
-    return error(Twine("Call site info provided but not used"));
+  if (!YamlMF.CallSitesInfo.empty() &&
+      !(TM.Options.EmitCallSiteInfo || TM.Options.EmitCallGraphSection))
+    return error("call site info provided but not used");
   return false;
 }
 

llvm/lib/CodeGen/MIRPrinter.cpp

@@ -525,24 +525,30 @@ static void convertCallSiteObjects(yaml::MachineFunction &YMF,
                                    const MachineFunction &MF,
                                    ModuleSlotTracker &MST) {
   const auto *TRI = MF.getSubtarget().getRegisterInfo();
-  for (auto CSInfo : MF.getCallSitesInfo()) {
+  for (auto [MI, CallSiteInfo] : MF.getCallSitesInfo()) {
     yaml::CallSiteInfo YmlCS;
     yaml::MachineInstrLoc CallLocation;
 
     // Prepare instruction position.
-    MachineBasicBlock::const_instr_iterator CallI = CSInfo.first->getIterator();
+    MachineBasicBlock::const_instr_iterator CallI = MI->getIterator();
     CallLocation.BlockNum = CallI->getParent()->getNumber();
     // Get call instruction offset from the beginning of block.
     CallLocation.Offset =
         std::distance(CallI->getParent()->instr_begin(), CallI);
     YmlCS.CallLocation = CallLocation;
+
+    auto [ArgRegPairs, CalleeTypeIds] = CallSiteInfo;
     // Construct call arguments and theirs forwarding register info.
-    for (auto ArgReg : CSInfo.second.ArgRegPairs) {
+    for (auto ArgReg : ArgRegPairs) {
       yaml::CallSiteInfo::ArgRegPair YmlArgReg;
       YmlArgReg.ArgNo = ArgReg.ArgNo;
       printRegMIR(ArgReg.Reg, YmlArgReg.Reg, TRI);
       YmlCS.ArgForwardingRegs.emplace_back(YmlArgReg);
     }
+    // Get type ids.
+    for (auto *CalleeTypeId : CalleeTypeIds) {
+      YmlCS.CalleeTypeIds.push_back(CalleeTypeId->getZExtValue());
+    }
     YMF.CallSitesInfo.push_back(std::move(YmlCS));
   }
 

llvm/lib/CodeGen/MachineFunction.cpp

@@ -919,7 +919,7 @@ MachineFunction::getCallSiteInfo(const MachineInstr *MI) {
   assert(MI->isCandidateForAdditionalCallInfo() &&
          "Call site info refers only to call (MI) candidates");
 
-  if (!Target.Options.EmitCallSiteInfo)
+  if (!Target.Options.EmitCallSiteInfo && !Target.Options.EmitCallGraphSection)
     return CallSitesInfo.end();
   return CallSitesInfo.find(MI);
 }

llvm/test/CodeGen/MIR/X86/call-site-info-ambiguous-indirect-call-typeid.mir

@@ -0,0 +1,31 @@
+# Test MIR printer and parser to check if a call instruction with multiple
+# callee types are handled correctly.
+
+# RUN: llc -mtriple=x86_64 --call-graph-section %s -run-pass=none -o - | FileCheck --match-full-lines %s
+# CHECK: name: ambiguous_caller
+# CHECK: callSites:
+# CHECK-NEXT: - { bb: {{.*}}, offset: {{.*}}, fwdArgRegs: {{.*}}, calleeTypeIds:
+# CHECK-NEXT: [ 1234, 5678 ] }
+
+--- |
+  define ptr @ambiguous_caller() {
+  entry:
+    %fn = alloca ptr, align 8
+    %call1 = call ptr %fn(i64 4), !callee_type !0
+    ret ptr %call1
+  }
+  
+  !0 = !{!1, !2}
+  !1 = !{i64 0, !"callee_type0.generalized"}
+  !2 = !{i64 0, !"callee_type2.generalized"}
+...
+---
+name:            ambiguous_caller
+callSites:
+  - { bb: 0, offset: 1, fwdArgRegs: [], calleeTypeIds: [ 1234, 5678 ] }
+body:             |
+  bb.0.entry:
+    %0:gr64 = MOV32ri64 4
+    CALL64r killed %0, csr_64, implicit $rsp, implicit $ssp, implicit $rdi, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+    RET 0, $rax
+...

llvm/test/CodeGen/MIR/X86/call-site-info-direct-calls-typeid.mir

@@ -0,0 +1,54 @@
+# Test MIR printer and parser to NOT have `CalleeTypeIds` field in callSites.
+# `CalleeTypeId` is used for propagating call site type identifiers for
+# indirect targets only. This test does not contain any indirect targets.
+
+# RUN: llc -mtriple=x86_64 --call-graph-section %s -run-pass=none -o - | FileCheck --match-full-lines %s
+# CHECK-NOT: calleeTypeIds
+# CHECK: name: bar
+# CHECK: callSites:
+# CHECK-NEXT: - { bb: {{.*}}, offset: {{.*}}, fwdArgRegs: [] }
+# CHECK-NEXT: - { bb: {{.*}}, offset: {{.*}}, fwdArgRegs: [] }
+# CHECK: name: foo
+# CHECK: callSites:
+# CHECK-NEXT: - { bb: {{.*}}, offset: {{.*}}, fwdArgRegs: [] }
+
+--- |
+  declare i32 @fizz(i32, i32)
+  
+  declare i32 @buzz(i32, i32)
+  
+  define i32 @bar(i32 %x, i32 %y) !type !0 {
+  entry:
+    %call = call i32 @buzz(i32 %x, i32 %x)
+    %call1 = call i32 @fizz(i32 %x, i32 %x)
+    ret i32 0
+  }
+  
+  define i32 @foo(i32 %x, i32 %y) !type !0 {
+  entry:
+    %call1 = call i32 @bar(i32 %x, i32 %x)
+    ret i32 0
+  }
+  
+  !0 = !{i64 0, !"_ZTSFiiiE.generalized"}
+...
+---
+name:            bar
+callSites:
+  - { bb: 0, offset: 0, fwdArgRegs: [] }
+  - { bb: 0, offset: 1, fwdArgRegs: [] }
+body:             |
+  bb.0.entry:
+    CALL64pcrel32 target-flags(x86-plt) @buzz, csr_64, implicit $rsp, implicit $ssp, implicit $edi, implicit $esi, implicit-def $rsp, implicit-def $ssp, implicit-def $eax
+    CALL64pcrel32 target-flags(x86-plt) @fizz, csr_64, implicit $rsp, implicit $ssp, implicit $edi, implicit $esi, implicit-def $rsp, implicit-def $ssp, implicit-def $eax
+
+...
+---
+name:            foo
+callSites:
+  - { bb: 0, offset: 0, fwdArgRegs: [] }
+body:             |
+  bb.0.entry:
+    CALL64pcrel32 target-flags(x86-plt) @bar, csr_64, implicit $rsp, implicit $ssp, implicit $edi, implicit $esi, implicit-def $rsp, implicit-def $ssp, implicit-def $eax
+
+...