Add the cold attribute to error-reporting call sites

Generally speaking, control flow paths with error reporting calls are cold.
So far, error reporting calls are calls to perror and calls to fprintf,
fwrite, etc. with stderr as the stream. This can be extended in the future.

The primary motivation is to improve block placement (the cold attribute
affects the static branch prediction heuristics).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194943 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Transforms/Utils/SimplifyLibCalls.cpp b/lib/Transforms/Utils/SimplifyLibCalls.cpp
index a060c3475e599213e8ee9b098f6e8bca0491b8ea..15b3e66f94ad037a91de1365bb8678b737a840b3 100644 (file)
--- a/lib/Transforms/Utils/SimplifyLibCalls.cpp
+++ b/lib/Transforms/Utils/SimplifyLibCalls.cpp
@@ -27,11 +27,16 @@
 #include "llvm/IR/LLVMContext.h"
 #include "llvm/IR/Module.h"
 #include "llvm/Support/Allocator.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Target/TargetLibraryInfo.h"
 #include "llvm/Transforms/Utils/BuildLibCalls.h"
 
 using namespace llvm;
 
+static cl::opt<bool>
+ColdErrorCalls("error-reporting-is-cold",  cl::init(true),
+  cl::Hidden, cl::desc("Treat error-reporting calls as cold"));
+
 /// This class is the abstract base class for the set of optimizations that
 /// corresponds to one library call.
 namespace {
@@ -1506,6 +1511,54 @@ struct ToAsciiOpt : public LibCallOptimization {
 // Formatting and IO Library Call Optimizations
 //===----------------------------------------------------------------------===//
 
+struct ErrorReportingOpt : public LibCallOptimization {
+  ErrorReportingOpt(int S = -1) : StreamArg(S) {}
+
+  virtual Value *callOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &) {
+    // Error reporting calls should be cold, mark them as such.
+    // This applies even to non-builtin calls: it is only a hint and applies to
+    // functions that the frontend might not understand as builtins.
+
+    // This heuristic was suggested in:
+    // Improving Static Branch Prediction in a Compiler
+    // Brian L. Deitrich, Ben-Chung Cheng, Wen-mei W. Hwu
+    // Proceedings of PACT'98, Oct. 1998, IEEE
+
+    if (!CI->hasFnAttr(Attribute::Cold) && isReportingError(Callee, CI)) {
+      CI->addAttribute(AttributeSet::FunctionIndex, Attribute::Cold);
+    }
+
+    return 0;
+  }
+
+protected:
+  bool isReportingError(Function *Callee, CallInst *CI) {
+    if (!ColdErrorCalls)
+      return false;
+ 
+    if (!Callee || !Callee->isDeclaration())
+      return false;
+
+    if (StreamArg < 0)
+      return true;
+
+    // These functions might be considered cold, but only if their stream
+    // argument is stderr.
+
+    if (StreamArg >= (int) CI->getNumArgOperands())
+      return false;
+    LoadInst *LI = dyn_cast<LoadInst>(CI->getArgOperand(StreamArg));
+    if (!LI)
+      return false;
+    GlobalVariable *GV = dyn_cast<GlobalVariable>(LI->getPointerOperand());
+    if (!GV || !GV->isDeclaration())
+      return false;
+    return GV->getName() == "stderr";
+  }
+
+  int StreamArg;
+};
+
 struct PrintFOpt : public LibCallOptimization {
   Value *optimizeFixedFormatString(Function *Callee, CallInst *CI,
                                    IRBuilder<> &B) {
@@ -1686,6 +1739,9 @@ struct SPrintFOpt : public LibCallOptimization {
 struct FPrintFOpt : public LibCallOptimization {
   Value *optimizeFixedFormatString(Function *Callee, CallInst *CI,
                                    IRBuilder<> &B) {
+    ErrorReportingOpt ER(/* StreamArg = */ 0);
+    (void) ER.callOptimizer(Callee, CI, B);
+
     // All the optimizations depend on the format string.
     StringRef FormatStr;
     if (!getConstantStringInfo(CI->getArgOperand(1), FormatStr))
@@ -1763,6 +1819,9 @@ struct FPrintFOpt : public LibCallOptimization {
 
 struct FWriteOpt : public LibCallOptimization {
   virtual Value *callOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) {
+    ErrorReportingOpt ER(/* StreamArg = */ 3);
+    (void) ER.callOptimizer(Callee, CI, B);
+
     // Require a pointer, an integer, an integer, a pointer, returning integer.
     FunctionType *FT = Callee->getFunctionType();
     if (FT->getNumParams() != 4 || !FT->getParamType(0)->isPointerTy() ||
@@ -1796,6 +1855,9 @@ struct FWriteOpt : public LibCallOptimization {
 
 struct FPutsOpt : public LibCallOptimization {
   virtual Value *callOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) {
+    ErrorReportingOpt ER(/* StreamArg = */ 1);
+    (void) ER.callOptimizer(Callee, CI, B);
+
     // These optimizations require DataLayout.
     if (!TD) return 0;
 
@@ -1924,6 +1986,9 @@ static IsAsciiOpt IsAscii;
 static ToAsciiOpt ToAscii;
 
 // Formatting and IO library call optimizations.
+static ErrorReportingOpt ErrorReporting;
+static ErrorReportingOpt ErrorReporting0(0);
+static ErrorReportingOpt ErrorReporting1(1);
 static PrintFOpt PrintF;
 static SPrintFOpt SPrintF;
 static FPrintFOpt FPrintF;
@@ -2038,6 +2103,13 @@ LibCallOptimization *LibCallSimplifierImpl::lookupOptimization(CallInst *CI) {
         return &FPuts;
       case LibFunc::puts:
         return &Puts;
+      case LibFunc::perror:
+        return &ErrorReporting;
+      case LibFunc::vfprintf:
+      case LibFunc::fiprintf:
+        return &ErrorReporting0;
+      case LibFunc::fputc:
+        return &ErrorReporting1;
       case LibFunc::ceil:
       case LibFunc::fabs:
       case LibFunc::floor:

diff --git a/test/Transforms/InstCombine/err-rep-cold.ll b/test/Transforms/InstCombine/err-rep-cold.ll
new file mode 100644 (file)
index 0000000..0cbafc4
--- /dev/null
+++ b/test/Transforms/InstCombine/err-rep-cold.ll
@@ -0,0 +1,77 @@
+; Test the static branch probability heuristics for error-reporting functions.
+; RUN: opt < %s -instcombine -S | FileCheck %s
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+%struct._IO_FILE = type { i32, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, %struct._IO_marker*, %struct._IO_FILE*, i32, i32, i64, i16, i8, [1 x i8], i8*, i64, i8*, i8*, i8*, i8*, i64, i32, [20 x i8] }
+%struct._IO_marker = type { %struct._IO_marker*, %struct._IO_FILE*, i32 }
+
+@stdout = external global %struct._IO_FILE*
+@stderr = external global %struct._IO_FILE*
+@.str = private unnamed_addr constant [13 x i8] c"an error: %d\00", align 1
+@.str1 = private unnamed_addr constant [9 x i8] c"an error\00", align 1
+
+define i32 @test1(i32 %a) #0 {
+; CHECK-LABEL: @test1
+entry:
+  %cmp = icmp sgt i32 %a, 8
+  br i1 %cmp, label %if.then, label %return
+
+if.then:                                          ; preds = %entry
+  %0 = load %struct._IO_FILE** @stderr, align 8
+  %call = tail call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %0, i8* getelementptr inbounds ([13 x i8]* @.str, i64 0, i64 0), i32 %a) #1
+  br label %return
+
+; CHECK: %call = tail call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %0, i8* getelementptr inbounds ([13 x i8]* @.str, i64 0, i64 0), i32 %a) #[[AT1:[0-9]+]]
+
+return:                                           ; preds = %entry, %if.then
+  %retval.0 = phi i32 [ 1, %if.then ], [ 0, %entry ]
+  ret i32 %retval.0
+}
+
+declare i32 @fprintf(%struct._IO_FILE* nocapture, i8* nocapture readonly, ...) #1
+
+define i32 @test2(i32 %a) #0 {
+; CHECK-LABEL: @test2
+entry:
+  %cmp = icmp sgt i32 %a, 8
+  br i1 %cmp, label %if.then, label %return
+
+if.then:                                          ; preds = %entry
+  %0 = load %struct._IO_FILE** @stderr, align 8
+  %1 = tail call i64 @fwrite(i8* getelementptr inbounds ([9 x i8]* @.str1, i64 0, i64 0), i64 8, i64 1, %struct._IO_FILE* %0)
+  br label %return
+
+; CHECK: tail call i64 @fwrite(i8* getelementptr inbounds ([9 x i8]* @.str1, i64 0, i64 0), i64 8, i64 1, %struct._IO_FILE* %0) #[[AT2:[0-9]+]]
+
+return:                                           ; preds = %entry, %if.then
+  %retval.0 = phi i32 [ 1, %if.then ], [ 0, %entry ]
+  ret i32 %retval.0
+}
+
+declare i64 @fwrite(i8* nocapture, i64, i64, %struct._IO_FILE* nocapture) #1
+
+define i32 @test3(i32 %a) #0 {
+; CHECK-LABEL: @test3
+entry:
+  %cmp = icmp sgt i32 %a, 8
+  br i1 %cmp, label %if.then, label %return
+
+if.then:                                          ; preds = %entry
+  %0 = load %struct._IO_FILE** @stdout, align 8
+  %1 = tail call i64 @fwrite(i8* getelementptr inbounds ([9 x i8]* @.str1, i64 0, i64 0), i64 8, i64 1, %struct._IO_FILE* %0)
+  br label %return
+
+; CHECK-NOT: tail call i64 @fwrite(i8* getelementptr inbounds ([9 x i8]* @.str1, i64 0, i64 0), i64 8, i64 1, %struct._IO_FILE* %0) #[[AT2]]
+
+return:                                           ; preds = %entry, %if.then
+  %retval.0 = phi i32 [ 1, %if.then ], [ 0, %entry ]
+  ret i32 %retval.0
+}
+
+attributes #0 = { nounwind uwtable }
+attributes #1 = { nounwind }
+
+; CHECK: attributes #[[AT1]] = { cold nounwind }
+; CHECK: attributes #[[AT2]] = { cold }
+