Cherry-picking the pow-for-le32 fix from upstream clang.

SVN log from upstream clang:

r185568 | eliben | 2013-07-03 12:19:12 -0700 (Wed, 03 Jul 2013)

Add target hook CodeGen queries when generating builtin pow*.

Without fmath-errno, Clang currently generates calls to @llvm.pow.* intrinsics
when it sees pow*(). This may not be suitable for all targets (for
example le32/PNaCl), so the attached patch adds a target hook that CodeGen
queries. The target can state its preference for having or not having the
intrinsic generated. Non-PNaCl behavior remains unchanged;
PNaCl-specific test added.

BUG= https://code.google.com/p/nativeclient/issues/detail?id=3513
R=dschuff@chromium.org

Review URL: https://codereview.chromium.org/18953003

5 files changed, 47 insertions, 12 deletions

diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index 7f1709e10f..0d5be26a5e 100644
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -1293,10 +1293,17 @@ RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,
   case Builtin::BIpow:
   case Builtin::BIpowf:
   case Builtin::BIpowl: {
-    // LOCALMOD: For PNACl we don't want pow* calls to ever turn into
-    // intrinsics. We want them to be resolved vs. the newlib implementation
-    // within the pexe during bitcode linking.
-    // TODO(eliben): upstream this
+    // Transform a call to pow* into a @llvm.pow.* intrinsic call, but only
+    // if the target agrees.
+    if (getTargetHooks().emitIntrinsicForPow()) {
+      if (!FD->hasAttr<ConstAttr>())
+        break;
+      Value *Base = EmitScalarExpr(E->getArg(0));
+      Value *Exponent = EmitScalarExpr(E->getArg(1));
+      llvm::Type *ArgType = Base->getType();
+      Value *F = CGM.getIntrinsic(Intrinsic::pow, ArgType);
+      return RValue::get(Builder.CreateCall2(F, Base, Exponent));
+    }
     break;
   }
 
diff --git a/lib/CodeGen/TargetInfo.cpp b/lib/CodeGen/TargetInfo.cpp
index b870ae9eb1..6ca2910d79 100644
--- a/lib/CodeGen/TargetInfo.cpp
+++ b/lib/CodeGen/TargetInfo.cpp
@@ -416,6 +416,10 @@ class PNaClTargetCodeGenInfo : public TargetCodeGenInfo {
  public:
   PNaClTargetCodeGenInfo(CodeGen::CodeGenTypes &CGT)
     : TargetCodeGenInfo(new PNaClABIInfo(CGT)) {}
+
+  /// For PNaCl we don't want llvm.pow.* intrinsics to be emitted instead
+  /// of library function calls.
+  bool emitIntrinsicForPow() const { return false; }
 };
 
 void PNaClABIInfo::computeInfo(CGFunctionInfo &FI) const {
diff --git a/lib/CodeGen/TargetInfo.h b/lib/CodeGen/TargetInfo.h
index bb50ce69e3..a682c183f0 100644
--- a/lib/CodeGen/TargetInfo.h
+++ b/lib/CodeGen/TargetInfo.h
@@ -73,6 +73,10 @@ namespace clang {
     ///     through such registers.
     virtual bool extendPointerWithSExt() const { return false; }
 
+    /// Controls whether BIpow* emit an intrinsic call instead of a library
+    /// function call.
+    virtual bool emitIntrinsicForPow() const { return true; }
+
     /// Determines the DWARF register number for the stack pointer, for
     /// exception-handling purposes.  Implements __builtin_dwarf_sp_column.
     ///
diff --git a/test/CodeGen/le32-libcall-pow.c b/test/CodeGen/le32-libcall-pow.c
new file mode 100644
index 0000000000..c2f892c3ae
--- /dev/null
+++ b/test/CodeGen/le32-libcall-pow.c
@@ -0,0 +1,21 @@
+// RUN: %clang_cc1 -fmath-errno -emit-llvm -o - %s -triple le32-unknown-nacl | FileCheck %s
+// RUN: %clang_cc1 -emit-llvm -o - %s -triple le32-unknown-nacl | FileCheck %s
+
+// le32 (PNaCl) never generates intrinsics for pow calls, with or without errno
+
+// CHECK: define void @test_pow
+void test_pow(float a0, double a1, long double a2) {
+  // CHECK: call float @powf
+  float l0 = powf(a0, a0);
+
+  // CHECK: call double @pow
+  double l1 = pow(a1, a1);
+
+  // CHECK: call double @powl
+  long double l2 = powl(a2, a2);
+}
+
+// CHECK: declare float @powf(float, float)
+// CHECK: declare double @pow(double, double)
+// CHECK: declare double @powl(double, double)
+
diff --git a/test/CodeGen/libcalls.c b/test/CodeGen/libcalls.c
index 87e0592337..8f8e18226a 100644
--- a/test/CodeGen/libcalls.c
+++ b/test/CodeGen/libcalls.c
@@ -31,27 +31,25 @@ void test_sqrt(float a0, double a1, long double a2) {
 // CHECK-YES: define void @test_pow
 // CHECK-NO: define void @test_pow
 void test_pow(float a0, double a1, long double a2) {
-  // LOCALMOD: for PNaCl we generate lib calls and not intrinsics, even without
-  // errno. For more details see the LOCALMOD in lib/CodeGen/CGBuiltin.cpp
   // CHECK-YES: call float @powf
-  // CHECK-NO: call float @powf
+  // CHECK-NO: call float @llvm.pow.f32
   float l0 = powf(a0, a0);
 
   // CHECK-YES: call double @pow
-  // CHECK-NO: call double @pow
+  // CHECK-NO: call double @llvm.pow.f64
   double l1 = pow(a1, a1);
 
   // CHECK-YES: call x86_fp80 @powl
-  // CHECK-NO: call x86_fp80 @powl
+  // CHECK-NO: call x86_fp80 @llvm.pow.f80
   long double l2 = powl(a2, a2);
 }
 
 // CHECK-YES: declare float @powf(float, float)
 // CHECK-YES: declare double @pow(double, double)
 // CHECK-YES: declare x86_fp80 @powl(x86_fp80, x86_fp80)
-// CHECK-NO: declare float @powf(float, float)
-// CHECK-NO: declare double @pow(double, double)
-// CHECK-NO: declare x86_fp80 @powl(x86_fp80, x86_fp80)
+// CHECK-NO: declare float @llvm.pow.f32(float, float) [[NUW_RO:#[0-9]+]]
+// CHECK-NO: declare double @llvm.pow.f64(double, double) [[NUW_RO]]
+// CHECK-NO: declare x86_fp80 @llvm.pow.f80(x86_fp80, x86_fp80) [[NUW_RO]]
 
 // CHECK-YES: define void @test_fma
 // CHECK-NO: define void @test_fma
@@ -124,3 +122,4 @@ void test_builtins(double d, float f, long double ld) {
 // CHECK-YES: attributes [[NUW_RN]] = { nounwind readnone }
 
 // CHECK-NO: attributes [[NUW_RN]] = { nounwind readnone{{.*}} }
+// CHECK-NO: attributes [[NUW_RO]] = { nounwind readonly }