handle invokes like calls for sample profile data

[david2050]

Summary

Dropping the names of invokes will lead to loss
of information possibly leading to inlining of
cold calls unnecessarily.

[vsk]

lib/Transforms/IPO/SampleProfile.cpp

664

CallSite(&Inst).getCalledFunction() might be easier to read here.

[david2050]

apply suggestion from @vsk

[vsk]

(Looks reasonable to me. I haven't really touched this code, so it'd help if another review could +1 the change.)

lib/Transforms/IPO/SampleProfile.cpp

 /// location.
 ///
 /// \param Inst Call/Invoke instruction to query.
 ///
 /// \returns The FunctionSamples pointer to the inlined instance.
 const FunctionSamples *
 SampleProfileLoader::findCalleeFunctionSamples(const Instruction &Inst) const {
   const DILocation *DIL = Inst.getDebugLoc();
-  if (!DIL) {
+  if (!DIL)
     return nullptr;
-  }
-
-  StringRef CalleeName;
-  if (const CallInst *CI = dyn_cast<CallInst>(&Inst))
-    if (Function *Callee = CI->getCalledFunction())
-      CalleeName = Callee->getName();
 
   const FunctionSamples *FS = findFunctionSamples(Inst);
   if (FS == nullptr)
     return nullptr;
 
+  StringRef CalleeName;
+  if (Function *Callee = CallSite(const_cast<Instruction*>(&Inst)).getCalledFunction())
+    CalleeName = Callee->getName();
   return FS->findFunctionSamplesAt(LineLocation(FunctionSamples::getOffset(DIL),
                                                 DIL->getBaseDiscriminator()),
                                    CalleeName);
 }
 
 /// Returns a vector of FunctionSamples that are the indirect call targets
 /// of \p Inst. The vector is sorted by the total number of samples. Stores
 /// the total call count of the indirect call in \p Sum.
 std::vector<const FunctionSamples *>
 SampleProfileLoader::findIndirectCallFunctionSamples(
     const Instruction &Inst, uint64_t &Sum) const {
   const DILocation *DIL = Inst.getDebugLoc();

test/Transforms/SampleProfile/Inputs/invoke.prof

+_Z3foov:120:20
+ 2: 20
+ 2.1: 21
+ 2.2: 22
+ 2.1: _Z12bar_noexceptv:0
+    1: 34
+ 2.2: _Z3barv:0
+    1: 0
+ 2.2: Other:35
+    1: 35

test/Transforms/SampleProfile/invoke.ll

+; RUN: opt < %s -S -add-discriminators -sample-profile -sample-profile-file=%S/Inputs/invoke.prof | FileCheck %s
+; ModuleID = 'invoke.bc'
+source_filename = "invoke.cpp"
+target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-apple-macosx10.14.0"
+
+; Function Attrs: nounwind ssp uwtable
+define void @_Z3barv() #0 !dbg !8 {
+entry:
+  ret void, !dbg !11
+}
+
+; Function Attrs: ssp uwtable
+define void @_Z3foov() #1 personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) !dbg !12 {
+entry:
+  %exn.slot = alloca i8*
+  %ehselector.slot = alloca i32
+  call void @_Z12bar_noexceptv() #4, !dbg !13
+  call void @_Z12bar_noexceptv() #4, !dbg !15
+  ; we should not inline this...
+  ; CHECK: invoke void @_Z3barv
+  invoke void @_Z3barv()
+          to label %invoke.cont unwind label %lpad, !dbg !16
+
+invoke.cont:                                      ; preds = %entry
+  br label %try.cont, !dbg !17
+
+lpad:                                             ; preds = %entry
+  %0 = landingpad { i8*, i32 }
+          catch i8* null, !dbg !18
+  %1 = extractvalue { i8*, i32 } %0, 0, !dbg !18
+  store i8* %1, i8** %exn.slot, align 8, !dbg !18
+  %2 = extractvalue { i8*, i32 } %0, 1, !dbg !18
+  store i32 %2, i32* %ehselector.slot, align 4, !dbg !18
+  br label %catch, !dbg !18
+
+catch:                                            ; preds = %lpad
+  %exn = load i8*, i8** %exn.slot, align 8, !dbg !17
+  %3 = call i8* @__cxa_begin_catch(i8* %exn) #4, !dbg !17
+  invoke void @__cxa_rethrow() #5
+          to label %unreachable unwind label %lpad1, !dbg !19
+
+lpad1:                                            ; preds = %catch
+  %4 = landingpad { i8*, i32 }
+          cleanup, !dbg !21
+  %5 = extractvalue { i8*, i32 } %4, 0, !dbg !21
+  store i8* %5, i8** %exn.slot, align 8, !dbg !21
+  %6 = extractvalue { i8*, i32 } %4, 1, !dbg !21
+  store i32 %6, i32* %ehselector.slot, align 4, !dbg !21
+  invoke void @__cxa_end_catch()
+          to label %invoke.cont2 unwind label %terminate.lpad, !dbg !22
+
+invoke.cont2:                                     ; preds = %lpad1
+  br label %eh.resume, !dbg !22
+
+try.cont:                                         ; preds = %invoke.cont
+  ret void, !dbg !23
+
+eh.resume:                                        ; preds = %invoke.cont2
+  %exn3 = load i8*, i8** %exn.slot, align 8, !dbg !22
+  %sel = load i32, i32* %ehselector.slot, align 4, !dbg !22
+  %lpad.val = insertvalue { i8*, i32 } undef, i8* %exn3, 0, !dbg !22
+  %lpad.val4 = insertvalue { i8*, i32 } %lpad.val, i32 %sel, 1, !dbg !22
+  resume { i8*, i32 } %lpad.val4, !dbg !22
+
+terminate.lpad:                                   ; preds = %lpad1
+  %7 = landingpad { i8*, i32 }
+          catch i8* null, !dbg !22
+  %8 = extractvalue { i8*, i32 } %7, 0, !dbg !22
+  call void @__clang_call_terminate(i8* %8) #6, !dbg !22
+  unreachable, !dbg !22
+
+unreachable:                                      ; preds = %catch
+  unreachable
+}
+
+; Function Attrs: nounwind
+declare void @_Z12bar_noexceptv() #2
+
+declare i32 @__gxx_personality_v0(...)
+
+declare i8* @__cxa_begin_catch(i8*)
+
+declare void @__cxa_rethrow()
+
+declare void @__cxa_end_catch()
+
+; Function Attrs: noinline noreturn nounwind
+define linkonce_odr hidden void @__clang_call_terminate(i8*) #3 {
+  %2 = call i8* @__cxa_begin_catch(i8* %0) #4
+  call void @_ZSt9terminatev() #6
+  unreachable
+}
+
+declare void @_ZSt9terminatev()
+
+attributes #0 = { nounwind ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="penryn" "target-features"="+cx16,+fxsr,+mmx,+sahf,+sse,+sse2,+sse3,+sse4.1,+ssse3,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #1 = { ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="penryn" "target-features"="+cx16,+fxsr,+mmx,+sahf,+sse,+sse2,+sse3,+sse4.1,+ssse3,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #2 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="penryn" "target-features"="+cx16,+fxsr,+mmx,+sahf,+sse,+sse2,+sse3,+sse4.1,+ssse3,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #3 = { noinline noreturn nounwind }
+attributes #4 = { nounwind }
+attributes #5 = { noreturn }
+attributes #6 = { noreturn nounwind }
+
+!llvm.dbg.cu = !{!0}
+!llvm.module.flags = !{!3, !4, !5, !6}
+!llvm.ident = !{!7}
+
+!0 = distinct !DICompileUnit(language: DW_LANG_C_plus_plus, file: !1, producer: "clang version 8.0.0", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, nameTableKind: GNU)
+!1 = !DIFile(filename: "invoke.cpp", directory: "examples")
+!2 = !{}
+!3 = !{i32 2, !"Dwarf Version", i32 4}
+!4 = !{i32 2, !"Debug Info Version", i32 3}
+!5 = !{i32 1, !"wchar_size", i32 4}
+!6 = !{i32 7, !"PIC Level", i32 2}
+!7 = !{!"clang version 8.0.0"}
+!8 = distinct !DISubprogram(name: "bar", linkageName: "_Z3barv", scope: !1, file: !1, line: 2, type: !9, scopeLine: 2, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2)
+!9 = !DISubroutineType(types: !10)
+!10 = !{null}
+!11 = !DILocation(line: 3, column: 1, scope: !8)
+!12 = distinct !DISubprogram(name: "foo", linkageName: "_Z3foov", scope: !1, file: !1, line: 6, type: !9, scopeLine: 6, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2)
+!13 = !DILocation(line: 8, column: 5, scope: !14)
+!14 = distinct !DILexicalBlock(scope: !12, file: !1, line: 7, column: 7)
+!15 = !DILocation(line: 8, column: 21, scope: !14)
+!16 = !DILocation(line: 8, column: 37, scope: !14)
+!17 = !DILocation(line: 9, column: 3, scope: !14)
+!18 = !DILocation(line: 13, column: 1, scope: !14)
+!19 = !DILocation(line: 11, column: 5, scope: !20)
+!20 = distinct !DILexicalBlock(scope: !12, file: !1, line: 10, column: 15)
+!21 = !DILocation(line: 13, column: 1, scope: !20)
+!22 = !DILocation(line: 12, column: 3, scope: !20)
+!23 = !DILocation(line: 13, column: 1, scope: !12)