Add -finstrument-function-entry-bare flag

[hans]

Summary

This is an instrumentation flag, similar to -finstrument-functions, but it only inserts calls on function entry, the calls are inserted post-inlining, and they don't take any arugments.

This is intended for users who want to instrument function entry with minimal overhead.

(-pg would be another alternative, but foces frame pointer emission and affects linking, so is probably best left alone to be used for generating gcov data.)

[pasko]

Instrumenting the function entry post-inlining, without function exit, and with no parameters is exactly what we need. The __cyg_profile_func_enter_bare sounds good to me as a name. Thank you!

Unnecessary thoughts just to get a feeling we are on the same page: this could theoretically be made more orthogonal where -finstrument-functions-after-inlining could regulate whether the call is pre- or post-inlining, but I don't see how pre-inlining without parameters would be usable without too much DWARF digging, which is not too practical.

[mattcary]

It looks like there also has to be a change to llvm/lib/Transforms/Utils/EntryExitInstrumenter.cpp?

(I realize that seems to be in a slightly different part of the tree than clang, but have no idea how to best coordinate the change)

Index: lib/Transforms/Utils/EntryExitInstrumenter.cpp

• lib/Transforms/Utils/EntryExitInstrumenter.cpp (revision 318760)

+++ lib/Transforms/Utils/EntryExitInstrumenter.cpp (working copy)
@@ -51,6 +51,10 @@

CallInst::Create(Fn, ArrayRef<Value *>(Args), "", InsertionPt);
return;

+ } else if (Func == "__cyg_profile_func_enter_bare") {
+ Constant *Fn = M.getOrInsertFunction(Func, Type::getVoidTy(C));
+ CallInst::Create(Fn, "", InsertionPt);
+ return;

}
 
// We only know how to call a fixed set of instrumentation functions, because

[hans]

Reid, are you happy with this too?

In D40276#931502, @pasko wrote:

Instrumenting the function entry post-inlining, without function exit, and with no parameters is exactly what we need. The __cyg_profile_func_enter_bare sounds good to me as a name. Thank you!

Great!

Unnecessary thoughts just to get a feeling we are on the same page: this could theoretically be made more orthogonal where -finstrument-functions-after-inlining could regulate whether the call is pre- or post-inlining, but I don't see how pre-inlining without parameters would be usable without too much DWARF digging, which is not too practical.

The way I think about them, these flags all enable separate instrumentations (though you can only enable one at a time), they don't modify eachother as I think that might end up messier for the user.

In D40276#931706, @mattcary wrote:

It looks like there also has to be a change to llvm/lib/Transforms/Utils/EntryExitInstrumenter.cpp?

Yes, I'll land that in llvm once we're happy with this. I just wanted to make sure we get the flag and function names right, then the rest is trivial.

[rnk]

Yep, looks good. :)

cfe/trunk/include/clang/Driver/Options.td

def fexperimental_new_pass_manager : Flag<["-"], "fexperimental-new-pass-manager">,
   Group<f_clang_Group>, Flags<[CC1Option]>,
   HelpText<"Enables an experimental new pass manager in LLVM.">;
 def finput_charset_EQ : Joined<["-"], "finput-charset=">, Group<f_Group>;
 def fexec_charset_EQ : Joined<["-"], "fexec-charset=">, Group<f_Group>;
 def finstrument_functions : Flag<["-"], "finstrument-functions">, Group<f_Group>, Flags<[CC1Option]>,
   HelpText<"Generate calls to instrument function entry and exit">;
 def finstrument_functions_after_inlining : Flag<["-"], "finstrument-functions-after-inlining">, Group<f_Group>, Flags<[CC1Option]>,
   HelpText<"Like -finstrument-functions, but insert the calls after inlining">;
+def finstrument_function_entry_bare : Flag<["-"], "finstrument-function-entry-bare">, Group<f_Group>, Flags<[CC1Option]>,
+  HelpText<"Instrument function entry only, after inlining, without arguments to the instrumentation call">;
 
 def fxray_instrument : Flag<["-"], "fxray-instrument">, Group<f_Group>,
   Flags<[CC1Option]>,
   HelpText<"Generate XRay instrumentation sleds on function entry and exit">;
 def fnoxray_instrument : Flag<["-"], "fno-xray-instrument">, Group<f_Group>,
   Flags<[CC1Option]>;
 
 def fxray_instruction_threshold_EQ :

cfe/trunk/include/clang/Frontend/CodeGenOptions.def

 ENUM_CODEGENOPT(EmbedBitcode, EmbedBitcodeKind, 2, Embed_Off)
 CODEGENOPT(ForbidGuardVariables , 1, 0) ///< Issue errors if C++ guard variables
                                         ///< are required.
 CODEGENOPT(FunctionSections  , 1, 0) ///< Set when -ffunction-sections is enabled.
 CODEGENOPT(InstrumentFunctions , 1, 0) ///< Set when -finstrument-functions is
                                        ///< enabled.
 CODEGENOPT(InstrumentFunctionsAfterInlining , 1, 0) ///< Set when
                           ///< -finstrument-functions-after-inlining is enabled.
+CODEGENOPT(InstrumentFunctionEntryBare , 1, 0) ///< Set when
+                               ///< -finstrument-function-entry-bare is enabled.
 
 CODEGENOPT(XRayInstrumentFunctions , 1, 0) ///< Set when -fxray-instrument is
                                            ///< enabled.
 
 ///< Set the minimum number of instructions in a function to determine selective
 ///< XRay instrumentation.
 VALUE_CODEGENOPT(XRayInstructionThreshold , 32, 200)
 

cfe/trunk/lib/CodeGen/CodeGenFunction.cpp

if (HasCleanups) {
 
     PopCleanupBlocks(PrologueCleanupDepth);
   }
 
   // Emit function epilog (to return).
   llvm::DebugLoc Loc = EmitReturnBlock();
 
   if (ShouldInstrumentFunction()) {
-    CurFn->addFnAttr(!CGM.getCodeGenOpts().InstrumentFunctionsAfterInlining
-                         ? "instrument-function-exit"
-                         : "instrument-function-exit-inlined",
+    if (CGM.getCodeGenOpts().InstrumentFunctions)
+      CurFn->addFnAttr("instrument-function-exit", "__cyg_profile_func_exit");
+    if (CGM.getCodeGenOpts().InstrumentFunctionsAfterInlining)
+      CurFn->addFnAttr("instrument-function-exit-inlined",
                        "__cyg_profile_func_exit");
   }
 
   // Emit debug descriptor for function end.
   if (CGDebugInfo *DI = getDebugInfo())
     DI->EmitFunctionEnd(Builder, CurFn);
 
   // Reset the debug location to that of the simple 'return' expression, if any
   // rather than that of the end of the function's scope '}'.
if (NormalCleanupDest && isCoroutine()) {
     NormalCleanupDest = nullptr;
   }
 }
 
 /// ShouldInstrumentFunction - Return true if the current function should be
 /// instrumented with __cyg_profile_func_* calls
 bool CodeGenFunction::ShouldInstrumentFunction() {
   if (!CGM.getCodeGenOpts().InstrumentFunctions &&
-      !CGM.getCodeGenOpts().InstrumentFunctionsAfterInlining)
+      !CGM.getCodeGenOpts().InstrumentFunctionsAfterInlining &&
+      !CGM.getCodeGenOpts().InstrumentFunctionEntryBare)
     return false;
   if (!CurFuncDecl || CurFuncDecl->hasAttr<NoInstrumentFunctionAttr>())
     return false;
   return true;
 }
 
 /// ShouldXRayInstrument - Return true if the current function should be
 /// instrumented with XRay nop sleds.
if (CGDebugInfo *DI = getDebugInfo()) {
     for (const VarDecl *VD : Args)
       ArgTypes.push_back(VD->getType());
     QualType FnType = getContext().getFunctionType(
         RetTy, ArgTypes, FunctionProtoType::ExtProtoInfo(CC));
     DI->EmitFunctionStart(GD, Loc, StartLoc, FnType, CurFn, Builder);
   }
 
   if (ShouldInstrumentFunction()) {
-    Fn->addFnAttr(!CGM.getCodeGenOpts().InstrumentFunctionsAfterInlining
-                      ? "instrument-function-entry"
-                      : "instrument-function-entry-inlined",
+    if (CGM.getCodeGenOpts().InstrumentFunctions)
+      CurFn->addFnAttr("instrument-function-entry", "__cyg_profile_func_enter");
+    if (CGM.getCodeGenOpts().InstrumentFunctionsAfterInlining)
+      CurFn->addFnAttr("instrument-function-entry-inlined",
                        "__cyg_profile_func_enter");
+    if (CGM.getCodeGenOpts().InstrumentFunctionEntryBare)
+      CurFn->addFnAttr("instrument-function-entry-inlined",
+                       "__cyg_profile_func_enter_bare");
   }
 
   // Since emitting the mcount call here impacts optimizations such as function
   // inlining, we just add an attribute to insert a mcount call in backend.
   // The attribute "counting-function" is set to mcount function name which is
   // architecture dependent.
   if (CGM.getCodeGenOpts().InstrumentForProfiling) {
     if (CGM.getCodeGenOpts().CallFEntry)

cfe/trunk/lib/Driver/ToolChains/Clang.cpp

if (Args.hasFlag(options::OPT_fdata_sections, options::OPT_fno_data_sections,
                    UseSeparateSections)) {
     CmdArgs.push_back("-fdata-sections");
   }
 
   if (!Args.hasFlag(options::OPT_funique_section_names,
                     options::OPT_fno_unique_section_names, true))
     CmdArgs.push_back("-fno-unique-section-names");
 
-  Args.AddLastArg(CmdArgs, options::OPT_finstrument_functions,
-                  options::OPT_finstrument_functions_after_inlining);
+  if (auto *A = Args.getLastArg(
+      options::OPT_finstrument_functions,
+      options::OPT_finstrument_functions_after_inlining,
+      options::OPT_finstrument_function_entry_bare))
+    A->render(Args, CmdArgs);
 
   addPGOAndCoverageFlags(C, D, Output, Args, CmdArgs);
 
   if (auto *ABICompatArg = Args.getLastArg(options::OPT_fclang_abi_compat_EQ))
     ABICompatArg->render(Args, CmdArgs);
 
   // Add runtime flag for PS4 when PGO or Coverage are enabled.
   if (RawTriple.isPS4CPU())

cfe/trunk/lib/Frontend/CompilerInvocation.cpp

for (const auto &A : Args) {
       }
     }
   }
 
   Opts.PreserveVec3Type = Args.hasArg(OPT_fpreserve_vec3_type);
   Opts.InstrumentFunctions = Args.hasArg(OPT_finstrument_functions);
   Opts.InstrumentFunctionsAfterInlining =
       Args.hasArg(OPT_finstrument_functions_after_inlining);
+  Opts.InstrumentFunctionEntryBare =
+      Args.hasArg(OPT_finstrument_function_entry_bare);
   Opts.XRayInstrumentFunctions = Args.hasArg(OPT_fxray_instrument);
   Opts.XRayInstructionThreshold =
       getLastArgIntValue(Args, OPT_fxray_instruction_threshold_EQ, 200, Diags);
   Opts.InstrumentForProfiling = Args.hasArg(OPT_pg);
   Opts.CallFEntry = Args.hasArg(OPT_mfentry);
   Opts.EmitOpenCLArgMetadata = Args.hasArg(OPT_cl_kernel_arg_info);
 
   if (const Arg *A = Args.getLastArg(OPT_compress_debug_sections,

cfe/trunk/test/CodeGen/instrument-functions.c

 // RUN: %clang_cc1 -S -debug-info-kind=standalone -emit-llvm -o - %s -finstrument-functions -disable-llvm-passes | FileCheck %s
+// RUN: %clang_cc1 -S -debug-info-kind=standalone -emit-llvm -o - %s -finstrument-function-entry-bare -disable-llvm-passes | FileCheck -check-prefix=BARE %s
 
 int test1(int x) {
 // CHECK: @test1(i32 {{.*}}%x) #[[ATTR1:[0-9]+]]
 // CHECK: ret
+
+// BARE: @test1(i32 {{.*}}%x) #[[ATTR1:[0-9]+]]
+// BARE: ret
   return x;
 }
 
 int test2(int) __attribute__((no_instrument_function));
 int test2(int x) {
 // CHECK: @test2(i32 {{.*}}%x) #[[ATTR2:[0-9]+]]
 // CHECK: ret
+
+// BARE: @test2(i32 {{.*}}%x) #[[ATTR2:[0-9]+]]
+// BARE: ret
   return x;
 }
 
 // CHECK: attributes #[[ATTR1]] =
 // CHECK-SAME: "instrument-function-entry"="__cyg_profile_func_enter"
 // CHECK-SAME: "instrument-function-exit"="__cyg_profile_func_exit"
 
+// BARE: attributes #[[ATTR1]] =
+// BARE-SAME: "instrument-function-entry-inlined"="__cyg_profile_func_enter_bare"
+
 // CHECK: attributes #[[ATTR2]] =
 // CHECK-NOT: "instrument-function-entry"
+
+// BARE: attributes #[[ATTR2]] =
+// BARE-NOT: "instrument-function-entry"