Add feature flags for AVX and FMA and fix some SSE4A feature flag

initialization problems.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@74350 91177308-0d34-0410-b5e6-96231b3b80d8

lib/Target/X86/X86.td

@@ -55,6 +55,13 @@ def FeatureSlowBTMem : SubtargetFeature<"slow-bt-mem", "IsBTMemSlow", "true",
 def FeatureSSE4A   : SubtargetFeature<"sse4a", "HasSSE4A", "true",
                                       "Support SSE 4a instructions">;
 
+def FeatureAVX     : SubtargetFeature<"avx", "HasAVX", "true",
+                                      "Enable AVX instructions">;
+def FeatureFMA3    : SubtargetFeature<"fma3", "HasFMA3", "true",
+                                      "Enable three-operand fused multiple-add">;
+def FeatureFMA4    : SubtargetFeature<"fma4", "HasFMA4", "true",
+                                      "Enable four-operand fused multiple-add">;
+
 //===----------------------------------------------------------------------===//
 // X86 processors supported.
 //===----------------------------------------------------------------------===//
@@ -82,6 +89,9 @@ def : Proc<"core2",           [FeatureSSSE3,  Feature64Bit, FeatureSlowBTMem]>;
 def : Proc<"penryn",          [FeatureSSE41,  Feature64Bit, FeatureSlowBTMem]>;
 def : Proc<"atom",            [FeatureSSE3,   Feature64Bit, FeatureSlowBTMem]>;
 def : Proc<"corei7",          [FeatureSSE42,  Feature64Bit, FeatureSlowBTMem]>;
+def : Proc<"nehalem",         [FeatureSSE42,  Feature64Bit, FeatureSlowBTMem]>;
+// Sandy Bridge does not have FMA
+def : Proc<"sandybridge",     [FeatureSSE42,  FeatureAVX,   Feature64Bit]>;
 
 def : Proc<"k6",              [FeatureMMX]>;
 def : Proc<"k6-2",            [FeatureMMX,    Feature3DNow]>;

lib/Target/X86/X86InstrInfo.td

@@ -236,6 +236,10 @@ def HasSSE3      : Predicate<"Subtarget->hasSSE3()">;
 def HasSSSE3     : Predicate<"Subtarget->hasSSSE3()">;
 def HasSSE41     : Predicate<"Subtarget->hasSSE41()">;
 def HasSSE42     : Predicate<"Subtarget->hasSSE42()">;
+def HasSSE4A     : Predicate<"Subtarget->hasSSE4A()">;
+def HasAVX       : Predicate<"Subtarget->hasAVX()">;
+def HasFMA3      : Predicate<"Subtarget->hasFMA3()">;
+def HasFMA4      : Predicate<"Subtarget->hasFMA4()">;
 def FPStackf32   : Predicate<"!Subtarget->hasSSE1()">;
 def FPStackf64   : Predicate<"!Subtarget->hasSSE2()">;
 def In32BitMode  : Predicate<"!Subtarget->is64Bit()">;

lib/Target/X86/X86Subtarget.cpp

@@ -207,6 +207,10 @@ void X86Subtarget::AutoDetectSubtargetFeatures() {
 
   bool IsIntel = memcmp(text.c, "GenuineIntel", 12) == 0;
   bool IsAMD   = !IsIntel && memcmp(text.c, "AuthenticAMD", 12) == 0;
+
+  HasFMA3 = IsIntel && ((ECX >> 12) & 0x1);
+  HasAVX = ((ECX >> 28) & 0x1);
+
   if (IsIntel || IsAMD) {
     // Determine if bit test memory instructions are slow.
     unsigned Family = 0;
@@ -217,6 +221,7 @@ void X86Subtarget::AutoDetectSubtargetFeatures() {
     X86::GetCpuIDAndInfo(0x80000001, &EAX, &EBX, &ECX, &EDX);
     HasX86_64 = (EDX >> 29) & 0x1;
     HasSSE4A = IsAMD && ((ECX >> 6) & 0x1);
+    HasFMA4 = IsAMD && ((ECX >> 16) & 0x1);
   }
 }
 
@@ -342,6 +347,10 @@ X86Subtarget::X86Subtarget(const Module &M, const std::string &FS, bool is64Bit)
   , X86SSELevel(NoMMXSSE)
   , X863DNowLevel(NoThreeDNow)
   , HasX86_64(false)
+  , HasSSE4A(false)
+  , HasAVX(false)
+  , HasFMA3(false)
+  , HasFMA4(false)
   , IsBTMemSlow(false)
   , DarwinVers(0)
   , IsLinux(false)

lib/Target/X86/X86Subtarget.h

@@ -64,12 +64,21 @@ protected:
   ///
   bool HasX86_64;
 
-  /// IsBTMemSlow - True if BT (bit test) of memory instructions are slow.
-  bool IsBTMemSlow;
-  
   /// HasSSE4A - True if the processor supports SSE4A instructions.
   bool HasSSE4A;
 
+  /// HasAVX - Target has AVX instructions
+  bool HasAVX;
+
+  /// HasFMA3 - Target has 3-operand fused multiply-add
+  bool HasFMA3;
+
+  /// HasFMA4 - Target has 4-operand fused multiply-add
+  bool HasFMA4;
+
+  /// IsBTMemSlow - True if BT (bit test) of memory instructions are slow.
+  bool IsBTMemSlow;
+  
   /// DarwinVers - Nonzero if this is a darwin platform: the numeric
   /// version of the platform, e.g. 8 = 10.4 (Tiger), 9 = 10.5 (Leopard), etc.
   unsigned char DarwinVers; // Is any darwin-x86 platform.
@@ -133,6 +142,9 @@ public:
   bool hasSSE4A() const { return HasSSE4A; }
   bool has3DNow() const { return X863DNowLevel >= ThreeDNow; }
   bool has3DNowA() const { return X863DNowLevel >= ThreeDNowA; }
+  bool hasAVX() const { return hasAVX(); }
+  bool hasFMA3() const { return HasFMA3; }
+  bool hasFMA4() const { return HasFMA4; }
 
   bool isBTMemSlow() const { return IsBTMemSlow; }