switch the flag for using NEON for SP floating point to a subtarget 'feature'.

Re-commit. This time complete with testsuite updates.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99570 91177308-0d34-0410-b5e6-96231b3b80d8

19 files changed, 31 insertions, 45 deletions

diff --git a/lib/Target/ARM/ARM.td b/lib/Target/ARM/ARM.td
index 50d54b436e..931c81c5b6 100644
--- a/lib/Target/ARM/ARM.td
+++ b/lib/Target/ARM/ARM.td
@@ -49,8 +49,14 @@ def FeatureFP16   : SubtargetFeature<"fp16", "HasFP16", "true",
 // FIXME: Currently, this is only flagged for Cortex-A8. It may be true for
 // others as well. We should do more benchmarking and confirm one way or
 // the other.
-def HasSlowVMLx   : SubtargetFeature<"vmlx", "SlowVMLx", "true",
-                                     "Disable VFP MAC instructions">;
+def FeatureHasSlowVMLx   : SubtargetFeature<"vmlx", "SlowVMLx", "true",
+                                            "Disable VFP MAC instructions">;
+// Some processors benefit from using NEON instructions for scalar
+// single-precision FP operations.
+def FeatureNEONForFP : SubtargetFeature<"neonfp", "UseNEONForSinglePrecisionFP",
+                                        "true",
+                                        "Use NEON for single precision FP">;
+
 
 //===----------------------------------------------------------------------===//
 // ARM Processors supported.
@@ -115,7 +121,8 @@ def : Processor<"arm1156t2f-s",    ARMV6Itineraries,
 
 // V7 Processors.
 def : Processor<"cortex-a8",        CortexA8Itineraries,
-                [ArchV7A, FeatureThumb2, FeatureNEON, HasSlowVMLx]>;
+                [ArchV7A, FeatureThumb2, FeatureNEON, FeatureHasSlowVMLx,
+                 FeatureNEONForFP]>;
 def : ProcNoItin<"cortex-a9",       [ArchV7A, FeatureThumb2, FeatureNEON]>;
 
 //===----------------------------------------------------------------------===//
diff --git a/lib/Target/ARM/ARMSubtarget.cpp b/lib/Target/ARM/ARMSubtarget.cpp
index 76ed66fe7d..9e55cd8700 100644
--- a/lib/Target/ARM/ARMSubtarget.cpp
+++ b/lib/Target/ARM/ARMSubtarget.cpp
@@ -22,10 +22,6 @@ using namespace llvm;
 static cl::opt<bool>
 ReserveR9("arm-reserve-r9", cl::Hidden,
           cl::desc("Reserve R9, making it unavailable as GPR"));
-static cl::opt<bool>
-UseNEONFP("arm-use-neon-fp",
-          cl::desc("Use NEON for single-precision FP"),
-          cl::init(false), cl::Hidden);
 
 static cl::opt<bool>
 UseMOVT("arm-use-movt",
@@ -35,7 +31,7 @@ ARMSubtarget::ARMSubtarget(const std::string &TT, const std::string &FS,
                            bool isT)
   : ARMArchVersion(V4)
   , ARMFPUType(None)
-  , UseNEONForSinglePrecisionFP(UseNEONFP)
+  , UseNEONForSinglePrecisionFP(false)
   , SlowVMLx(false)
   , IsThumb(isT)
   , ThumbMode(Thumb1)
@@ -116,14 +112,6 @@ ARMSubtarget::ARMSubtarget(const std::string &TT, const std::string &FS,
 
   if (!isThumb() || hasThumb2())
     PostRAScheduler = true;
-
-  // Set CPU specific features.
-  if (CPUString == "cortex-a8") {
-    // On Cortex-a8, it's faster to perform some single-precision FP
-    // operations with NEON instructions.
-    if (UseNEONFP.getPosition() == 0)
-      UseNEONForSinglePrecisionFP = true;
-  }
 }
 
 /// GVIsIndirectSymbol - true if the GV will be accessed via an indirect symbol.
diff --git a/test/CodeGen/ARM/fabss.ll b/test/CodeGen/ARM/fabss.ll
index e5b5791b3c..f03282bdab 100644
--- a/test/CodeGen/ARM/fabss.ll
+++ b/test/CodeGen/ARM/fabss.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NFP1
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NFP0
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=CORTEXA8
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=CORTEXA9
 
diff --git a/test/CodeGen/ARM/fadds.ll b/test/CodeGen/ARM/fadds.ll
index db18a86ecc..749690e98d 100644
--- a/test/CodeGen/ARM/fadds.ll
+++ b/test/CodeGen/ARM/fadds.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NFP1
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NFP0
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=CORTEXA8
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=CORTEXA9
 
diff --git a/test/CodeGen/ARM/fdivs.ll b/test/CodeGen/ARM/fdivs.ll
index a5c86bf263..0c31495792 100644
--- a/test/CodeGen/ARM/fdivs.ll
+++ b/test/CodeGen/ARM/fdivs.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NFP1
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NFP0
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=CORTEXA8
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=CORTEXA9
 
diff --git a/test/CodeGen/ARM/fmacs.ll b/test/CodeGen/ARM/fmacs.ll
index 904a587393..f8b47b5bac 100644
--- a/test/CodeGen/ARM/fmacs.ll
+++ b/test/CodeGen/ARM/fmacs.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NFP1
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NFP0
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=CORTEXA8
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=CORTEXA9
 
diff --git a/test/CodeGen/ARM/fmscs.ll b/test/CodeGen/ARM/fmscs.ll
index 7b9e029b67..7a70543dee 100644
--- a/test/CodeGen/ARM/fmscs.ll
+++ b/test/CodeGen/ARM/fmscs.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NFP1
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NFP0
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=CORTEXA8
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=CORTEXA9
 
diff --git a/test/CodeGen/ARM/fmuls.ll b/test/CodeGen/ARM/fmuls.ll
index d3c9c82e97..ef4e3e5281 100644
--- a/test/CodeGen/ARM/fmuls.ll
+++ b/test/CodeGen/ARM/fmuls.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NFP1
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NFP0
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=CORTEXA8
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=CORTEXA9
 
diff --git a/test/CodeGen/ARM/fnegs.ll b/test/CodeGen/ARM/fnegs.ll
index d6c22f14a4..c15005e6e8 100644
--- a/test/CodeGen/ARM/fnegs.ll
+++ b/test/CodeGen/ARM/fnegs.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NFP1
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NFP0
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=CORTEXA8
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=CORTEXA9
 
diff --git a/test/CodeGen/ARM/fnmacs.ll b/test/CodeGen/ARM/fnmacs.ll
index 724947ea04..1d1d06a70e 100644
--- a/test/CodeGen/ARM/fnmacs.ll
+++ b/test/CodeGen/ARM/fnmacs.ll
@@ -1,6 +1,6 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NEON
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NEONFP
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NEON
+; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=NEONFP
 
 define float @test(float %acc, float %a, float %b) {
 entry:
diff --git a/test/CodeGen/ARM/fnmscs.ll b/test/CodeGen/ARM/fnmscs.ll
index ad2188218e..6b7cefa641 100644
--- a/test/CodeGen/ARM/fnmscs.ll
+++ b/test/CodeGen/ARM/fnmscs.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s
 
diff --git a/test/CodeGen/ARM/fp_convert.ll b/test/CodeGen/ARM/fp_convert.ll
index 2adac78cf8..1ef9f7f321 100644
--- a/test/CodeGen/ARM/fp_convert.ll
+++ b/test/CodeGen/ARM/fp_convert.ll
@@ -1,6 +1,5 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NEON
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=VFP2
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=VFP2
 ; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=NEON
 ; RUN: llc < %s -march=arm -mcpu=cortex-a9 | FileCheck %s -check-prefix=VFP2
 
diff --git a/test/CodeGen/ARM/fsubs.ll b/test/CodeGen/ARM/fsubs.ll
index ae98be3078..bea8d5f4f3 100644
--- a/test/CodeGen/ARM/fsubs.ll
+++ b/test/CodeGen/ARM/fsubs.ll
@@ -1,6 +1,6 @@
 ; RUN: llc < %s -march=arm -mattr=+vfp2 | FileCheck %s -check-prefix=VFP2
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=1 | FileCheck %s -check-prefix=NFP1
-; RUN: llc < %s -march=arm -mattr=+neon -arm-use-neon-fp=0 | FileCheck %s -check-prefix=NFP0
+; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s -check-prefix=NFP1
+; RUN: llc < %s -march=arm -mattr=+neon | FileCheck %s -check-prefix=NFP0
 
 define float @test(float %a, float %b) {
 entry:
diff --git a/test/CodeGen/Thumb2/2009-08-04-CoalescerBug.ll b/test/CodeGen/Thumb2/2009-08-04-CoalescerBug.ll
index 319d29b790..9c1fdb32e8 100644
--- a/test/CodeGen/Thumb2/2009-08-04-CoalescerBug.ll
+++ b/test/CodeGen/Thumb2/2009-08-04-CoalescerBug.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=thumbv7-apple-darwin -mattr=+neon -arm-use-neon-fp -relocation-model=pic -disable-fp-elim
+; RUN: llc < %s -mtriple=thumbv7-apple-darwin -mcpu=cortex-a8 -relocation-model=pic -disable-fp-elim
 
 	type { %struct.GAP }		; type %0
 	type { i16, i8, i8 }		; type %1
diff --git a/test/CodeGen/Thumb2/2009-08-04-ScavengerAssert.ll b/test/CodeGen/Thumb2/2009-08-04-ScavengerAssert.ll
index a62b61290a..317db64ae4 100644
--- a/test/CodeGen/Thumb2/2009-08-04-ScavengerAssert.ll
+++ b/test/CodeGen/Thumb2/2009-08-04-ScavengerAssert.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=thumbv7-apple-darwin -mattr=+neon -arm-use-neon-fp -relocation-model=pic -disable-fp-elim -O3
+; RUN: llc < %s -mtriple=thumbv7-apple-darwin -mcpu=cortex-a8 -relocation-model=pic -disable-fp-elim -O3
 
 	type { i16, i8, i8 }		; type %0
 	type { [2 x i32], [2 x i32] }		; type %1
diff --git a/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug.ll b/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug.ll
index 76474746ee..2bbc231f96 100644
--- a/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug.ll
+++ b/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug.ll
@@ -1,5 +1,5 @@
-; RUN: llc < %s -mtriple=thumbv7-apple-darwin9 -mattr=+neon -arm-use-neon-fp 
-; RUN: llc < %s -mtriple=thumbv7-apple-darwin9 -mattr=+neon -arm-use-neon-fp | not grep fcpys
+; RUN: llc < %s -mtriple=thumbv7-apple-darwin9 -mcpu=cortex-a8
+; RUN: llc < %s -mtriple=thumbv7-apple-darwin9 -mcpu=cortex-a8 | not grep fcpys
 ; rdar://7117307
 
 	%struct.Hosp = type { i32, i32, i32, %struct.List, %struct.List, %struct.List, %struct.List }
diff --git a/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug2.ll b/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug2.ll
index acf562c74a..8294484766 100644
--- a/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug2.ll
+++ b/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug2.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=thumbv7-apple-darwin9 -mattr=+neon -arm-use-neon-fp
+; RUN: llc < %s -mtriple=thumbv7-apple-darwin9 -mcpu=cortex-a8
 ; rdar://7117307
 
 	%struct.Hosp = type { i32, i32, i32, %struct.List, %struct.List, %struct.List, %struct.List }
diff --git a/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug3.ll b/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug3.ll
index 3ada02676b..b18c972aed 100644
--- a/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug3.ll
+++ b/test/CodeGen/Thumb2/2009-08-04-SubregLoweringBug3.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=thumbv7-apple-darwin9 -mattr=+neon -arm-use-neon-fp
+; RUN: llc < %s -mtriple=thumbv7-apple-darwin9 -mcpu=cortex-a8
 ; rdar://7117307
 
 	%struct.Hosp = type { i32, i32, i32, %struct.List, %struct.List, %struct.List, %struct.List }
diff --git a/test/CodeGen/Thumb2/2009-08-07-NeonFPBug.ll b/test/CodeGen/Thumb2/2009-08-07-NeonFPBug.ll
index 090ed2d81f..96bcbad771 100644
--- a/test/CodeGen/Thumb2/2009-08-07-NeonFPBug.ll
+++ b/test/CodeGen/Thumb2/2009-08-07-NeonFPBug.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=thumbv7-apple-darwin10 -mcpu=cortex-a8 -arm-use-neon-fp
+; RUN: llc < %s -mtriple=thumbv7-apple-darwin10 -mcpu=cortex-a8
 
 	%struct.FILE = type { i8*, i32, i32, i16, i16, %struct.__sbuf, i32, i8*, i32 (i8*)*, i32 (i8*, i8*, i32)*, i64 (i8*, i64, i32)*, i32 (i8*, i8*, i32)*, %struct.__sbuf, %struct.__sFILEX*, i32, [3 x i8], [1 x i8], %struct.__sbuf, i32, i64 }
 	%struct.JHUFF_TBL = type { [17 x i8], [256 x i8], i32 }