Add PREFETCHW codegen support

- Add 'PRFCHW' feature defined in AVX2 ISA extension



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@178040 91177308-0d34-0410-b5e6-96231b3b80d8

lib/Target/X86/X86.td

@@ -122,6 +122,8 @@ def FeatureRTM     : SubtargetFeature<"rtm", "HasRTM", "true",
                                       "Support RTM instructions">;
 def FeatureADX     : SubtargetFeature<"adx", "HasADX", "true",
                                       "Support ADX instructions">;
+def FeaturePRFCHW  : SubtargetFeature<"prfchw", "HasPRFCHW", "true",
+                                      "Support PRFCHW instructions">;
 def FeatureLeaForSP : SubtargetFeature<"lea-sp", "UseLeaForSP", "true",
                                      "Use LEA for adjusting the stack pointer">;
 def FeatureSlowDivide : SubtargetFeature<"idiv-to-divb",

lib/Target/X86/X86Instr3DNow.td

@@ -84,13 +84,16 @@ defm PI2FD    : I3DNow_conv_rm_int<0x0D, "pi2fd">;
 defm PMULHRW  : I3DNow_binop_rm_int<0xB7, "pmulhrw">;
 
 
-def FEMMS : I3DNow<0x0E, RawFrm, (outs), (ins), "femms", [(int_x86_mmx_femms)]>;
+def FEMMS : I3DNow<0x0E, RawFrm, (outs), (ins), "femms",
+                   [(int_x86_mmx_femms)]>;
 
-def PREFETCH  : I3DNow<0x0D, MRM0m, (outs), (ins i32mem:$addr),
+def PREFETCH : I3DNow<0x0D, MRM0m, (outs), (ins i8mem:$addr),
-                       "prefetch\t$addr", []>;
+                      "prefetch\t$addr",
+                      [(prefetch addr:$addr, (i32 0), imm, (i32 1))]>;
 
-def PREFETCHW : I3DNow<0x0D, MRM1m, (outs), (ins i16mem:$addr),
+def PREFETCHW : I<0x0D, MRM1m, (outs), (ins i8mem:$addr), "prefetchw\t$addr",
-                       "prefetchw\t$addr", []>;
+                  [(prefetch addr:$addr, (i32 1), (i32 3), (i32 1))]>, TB,
+                Requires<[HasPrefetchW]>;
 
 // "3DNowA" instructions
 defm PF2IW    : I3DNow_conv_rm_int<0x1C, "pf2iw", "a">;

lib/Target/X86/X86InstrInfo.td

@@ -604,6 +604,8 @@ def HasBMI       : Predicate<"Subtarget->hasBMI()">;
 def HasBMI2      : Predicate<"Subtarget->hasBMI2()">;
 def HasRTM       : Predicate<"Subtarget->hasRTM()">;
 def HasADX       : Predicate<"Subtarget->hasADX()">;
+def HasPRFCHW    : Predicate<"Subtarget->hasPRFCHW()">;
+def HasPrefetchW : Predicate<"Subtarget->has3DNow() || Subtarget->hasPRFCHW()">;
 def FPStackf32   : Predicate<"!Subtarget->hasSSE1()">;
 def FPStackf64   : Predicate<"!Subtarget->hasSSE2()">;
 def HasCmpxchg16b: Predicate<"Subtarget->hasCmpxchg16b()">;

lib/Target/X86/X86Subtarget.cpp

@@ -283,6 +283,10 @@ void X86Subtarget::AutoDetectSubtargetFeatures() {
         HasLZCNT = true;
         ToggleFeature(X86::FeatureLZCNT);
       }
+      if (IsIntel && ((ECX >> 8) & 0x1)) {
+        HasPRFCHW = true;
+        ToggleFeature(X86::FeaturePRFCHW);
+      }
       if (IsAMD) {
         if ((ECX >> 6) & 0x1) {
           HasSSE4A = true;
@@ -440,6 +444,7 @@ void X86Subtarget::initializeEnvironment() {
   HasBMI2 = false;
   HasRTM = false;
   HasADX = false;
+  HasPRFCHW = false;
   IsBTMemSlow = false;
   IsUAMemFast = false;
   HasVectorUAMem = false;

lib/Target/X86/X86Subtarget.h

@@ -124,6 +124,9 @@ protected:
   /// HasADX - Processor has ADX instructions.
   bool HasADX;
 
+  /// HasPRFCHW - Processor has PRFCHW instructions.
+  bool HasPRFCHW;
+
   /// IsBTMemSlow - True if BT (bit test) of memory instructions are slow.
   bool IsBTMemSlow;
 
@@ -254,6 +257,7 @@ public:
   bool hasBMI2() const { return HasBMI2; }
   bool hasRTM() const { return HasRTM; }
   bool hasADX() const { return HasADX; }
+  bool hasPRFCHW() const { return HasPRFCHW; }
   bool isBTMemSlow() const { return IsBTMemSlow; }
   bool isUnalignedMemAccessFast() const { return IsUAMemFast; }
   bool hasVectorUAMem() const { return HasVectorUAMem; }

test/CodeGen/X86/prefetch.ll

@@ -1,5 +1,6 @@
 ; RUN: llc < %s -march=x86 -mattr=+sse | FileCheck %s
 ; RUN: llc < %s -march=x86 -mattr=+avx | FileCheck %s
+; RUN: llc < %s -march=x86 -mattr=+prfchw | FileCheck %s -check-prefix=PRFCHW
 
 ; rdar://10538297
 
@@ -9,10 +10,12 @@ entry:
 ; CHECK: prefetcht1
 ; CHECK: prefetcht0
 ; CHECK: prefetchnta
+; PRFCHW: prefetchw
 	tail call void @llvm.prefetch( i8* %ptr, i32 0, i32 1, i32 1 )
 	tail call void @llvm.prefetch( i8* %ptr, i32 0, i32 2, i32 1 )
 	tail call void @llvm.prefetch( i8* %ptr, i32 0, i32 3, i32 1 )
 	tail call void @llvm.prefetch( i8* %ptr, i32 0, i32 0, i32 1 )
+	tail call void @llvm.prefetch( i8* %ptr, i32 1, i32 3, i32 1 )
 	ret void
 }