[X86][FMA4] FMA4 ops can perform unaligned folded loads.

llvm-svn: 240342

diff --git a/llvm/lib/Target/X86/X86InstrInfo.cpp b/llvm/lib/Target/X86/X86InstrInfo.cpp
index 744a7cb..4dee5b7 100644 (file)
--- a/llvm/lib/Target/X86/X86InstrInfo.cpp
+++ b/llvm/lib/Target/X86/X86InstrInfo.cpp
@@ -1577,38 +1577,38 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
     { X86::VPXORYrr,          X86::VPXORYrm,           0 },
 
     // FMA4 foldable patterns
-    { X86::VFMADDSS4rr,       X86::VFMADDSS4mr,        0 },
-    { X86::VFMADDSD4rr,       X86::VFMADDSD4mr,        0 },
-    { X86::VFMADDPS4rr,       X86::VFMADDPS4mr,        0 },
-    { X86::VFMADDPD4rr,       X86::VFMADDPD4mr,        0 },
-    { X86::VFMADDPS4rrY,      X86::VFMADDPS4mrY,       0 },
-    { X86::VFMADDPD4rrY,      X86::VFMADDPD4mrY,       0 },
-    { X86::VFNMADDSS4rr,      X86::VFNMADDSS4mr,       0 },
-    { X86::VFNMADDSD4rr,      X86::VFNMADDSD4mr,       0 },
-    { X86::VFNMADDPS4rr,      X86::VFNMADDPS4mr,       0 },
-    { X86::VFNMADDPD4rr,      X86::VFNMADDPD4mr,       0 },
-    { X86::VFNMADDPS4rrY,     X86::VFNMADDPS4mrY,      0 },
-    { X86::VFNMADDPD4rrY,     X86::VFNMADDPD4mrY,      0 },
-    { X86::VFMSUBSS4rr,       X86::VFMSUBSS4mr,        0 },
-    { X86::VFMSUBSD4rr,       X86::VFMSUBSD4mr,        0 },
-    { X86::VFMSUBPS4rr,       X86::VFMSUBPS4mr,        0 },
-    { X86::VFMSUBPD4rr,       X86::VFMSUBPD4mr,        0 },
-    { X86::VFMSUBPS4rrY,      X86::VFMSUBPS4mrY,       0 },
-    { X86::VFMSUBPD4rrY,      X86::VFMSUBPD4mrY,       0 },
-    { X86::VFNMSUBSS4rr,      X86::VFNMSUBSS4mr,       0 },
-    { X86::VFNMSUBSD4rr,      X86::VFNMSUBSD4mr,       0 },
-    { X86::VFNMSUBPS4rr,      X86::VFNMSUBPS4mr,       0 },
-    { X86::VFNMSUBPD4rr,      X86::VFNMSUBPD4mr,       0 },
-    { X86::VFNMSUBPS4rrY,     X86::VFNMSUBPS4mrY,      0 },
-    { X86::VFNMSUBPD4rrY,     X86::VFNMSUBPD4mrY,      0 },
-    { X86::VFMADDSUBPS4rr,    X86::VFMADDSUBPS4mr,     0 },
-    { X86::VFMADDSUBPD4rr,    X86::VFMADDSUBPD4mr,     0 },
-    { X86::VFMADDSUBPS4rrY,   X86::VFMADDSUBPS4mrY,    0 },
-    { X86::VFMADDSUBPD4rrY,   X86::VFMADDSUBPD4mrY,    0 },
-    { X86::VFMSUBADDPS4rr,    X86::VFMSUBADDPS4mr,     0 },
-    { X86::VFMSUBADDPD4rr,    X86::VFMSUBADDPD4mr,     0 },
-    { X86::VFMSUBADDPS4rrY,   X86::VFMSUBADDPS4mrY,    0 },
-    { X86::VFMSUBADDPD4rrY,   X86::VFMSUBADDPD4mrY,    0 },
+    { X86::VFMADDSS4rr,       X86::VFMADDSS4mr,        TB_ALIGN_NONE },
+    { X86::VFMADDSD4rr,       X86::VFMADDSD4mr,        TB_ALIGN_NONE },
+    { X86::VFMADDPS4rr,       X86::VFMADDPS4mr,        TB_ALIGN_NONE },
+    { X86::VFMADDPD4rr,       X86::VFMADDPD4mr,        TB_ALIGN_NONE },
+    { X86::VFMADDPS4rrY,      X86::VFMADDPS4mrY,       TB_ALIGN_NONE },
+    { X86::VFMADDPD4rrY,      X86::VFMADDPD4mrY,       TB_ALIGN_NONE },
+    { X86::VFNMADDSS4rr,      X86::VFNMADDSS4mr,       TB_ALIGN_NONE },
+    { X86::VFNMADDSD4rr,      X86::VFNMADDSD4mr,       TB_ALIGN_NONE },
+    { X86::VFNMADDPS4rr,      X86::VFNMADDPS4mr,       TB_ALIGN_NONE },
+    { X86::VFNMADDPD4rr,      X86::VFNMADDPD4mr,       TB_ALIGN_NONE },
+    { X86::VFNMADDPS4rrY,     X86::VFNMADDPS4mrY,      TB_ALIGN_NONE },
+    { X86::VFNMADDPD4rrY,     X86::VFNMADDPD4mrY,      TB_ALIGN_NONE },
+    { X86::VFMSUBSS4rr,       X86::VFMSUBSS4mr,        TB_ALIGN_NONE },
+    { X86::VFMSUBSD4rr,       X86::VFMSUBSD4mr,        TB_ALIGN_NONE },
+    { X86::VFMSUBPS4rr,       X86::VFMSUBPS4mr,        TB_ALIGN_NONE },
+    { X86::VFMSUBPD4rr,       X86::VFMSUBPD4mr,        TB_ALIGN_NONE },
+    { X86::VFMSUBPS4rrY,      X86::VFMSUBPS4mrY,       TB_ALIGN_NONE },
+    { X86::VFMSUBPD4rrY,      X86::VFMSUBPD4mrY,       TB_ALIGN_NONE },
+    { X86::VFNMSUBSS4rr,      X86::VFNMSUBSS4mr,       TB_ALIGN_NONE },
+    { X86::VFNMSUBSD4rr,      X86::VFNMSUBSD4mr,       TB_ALIGN_NONE },
+    { X86::VFNMSUBPS4rr,      X86::VFNMSUBPS4mr,       TB_ALIGN_NONE },
+    { X86::VFNMSUBPD4rr,      X86::VFNMSUBPD4mr,       TB_ALIGN_NONE },
+    { X86::VFNMSUBPS4rrY,     X86::VFNMSUBPS4mrY,      TB_ALIGN_NONE },
+    { X86::VFNMSUBPD4rrY,     X86::VFNMSUBPD4mrY,      TB_ALIGN_NONE },
+    { X86::VFMADDSUBPS4rr,    X86::VFMADDSUBPS4mr,     TB_ALIGN_NONE },
+    { X86::VFMADDSUBPD4rr,    X86::VFMADDSUBPD4mr,     TB_ALIGN_NONE },
+    { X86::VFMADDSUBPS4rrY,   X86::VFMADDSUBPS4mrY,    TB_ALIGN_NONE },
+    { X86::VFMADDSUBPD4rrY,   X86::VFMADDSUBPD4mrY,    TB_ALIGN_NONE },
+    { X86::VFMSUBADDPS4rr,    X86::VFMSUBADDPS4mr,     TB_ALIGN_NONE },
+    { X86::VFMSUBADDPD4rr,    X86::VFMSUBADDPD4mr,     TB_ALIGN_NONE },
+    { X86::VFMSUBADDPS4rrY,   X86::VFMSUBADDPS4mrY,    TB_ALIGN_NONE },
+    { X86::VFMSUBADDPD4rrY,   X86::VFMSUBADDPD4mrY,    TB_ALIGN_NONE },
 
     // XOP foldable instructions
     { X86::VPCMOVrr,          X86::VPCMOVmr,            0 },
@@ -1852,38 +1852,38 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
     { X86::VFMSUBADDPDr213rY,     X86::VFMSUBADDPDr213mY,     TB_ALIGN_NONE },
 
     // FMA4 foldable patterns
-    { X86::VFMADDSS4rr,           X86::VFMADDSS4rm,           0           },
-    { X86::VFMADDSD4rr,           X86::VFMADDSD4rm,           0           },
-    { X86::VFMADDPS4rr,           X86::VFMADDPS4rm,           TB_ALIGN_16 },
-    { X86::VFMADDPD4rr,           X86::VFMADDPD4rm,           TB_ALIGN_16 },
-    { X86::VFMADDPS4rrY,          X86::VFMADDPS4rmY,          TB_ALIGN_32 },
-    { X86::VFMADDPD4rrY,          X86::VFMADDPD4rmY,          TB_ALIGN_32 },
-    { X86::VFNMADDSS4rr,          X86::VFNMADDSS4rm,          0           },
-    { X86::VFNMADDSD4rr,          X86::VFNMADDSD4rm,          0           },
-    { X86::VFNMADDPS4rr,          X86::VFNMADDPS4rm,          TB_ALIGN_16 },
-    { X86::VFNMADDPD4rr,          X86::VFNMADDPD4rm,          TB_ALIGN_16 },
-    { X86::VFNMADDPS4rrY,         X86::VFNMADDPS4rmY,         TB_ALIGN_32 },
-    { X86::VFNMADDPD4rrY,         X86::VFNMADDPD4rmY,         TB_ALIGN_32 },
-    { X86::VFMSUBSS4rr,           X86::VFMSUBSS4rm,           0           },
-    { X86::VFMSUBSD4rr,           X86::VFMSUBSD4rm,           0           },
-    { X86::VFMSUBPS4rr,           X86::VFMSUBPS4rm,           TB_ALIGN_16 },
-    { X86::VFMSUBPD4rr,           X86::VFMSUBPD4rm,           TB_ALIGN_16 },
-    { X86::VFMSUBPS4rrY,          X86::VFMSUBPS4rmY,          TB_ALIGN_32 },
-    { X86::VFMSUBPD4rrY,          X86::VFMSUBPD4rmY,          TB_ALIGN_32 },
-    { X86::VFNMSUBSS4rr,          X86::VFNMSUBSS4rm,          0           },
-    { X86::VFNMSUBSD4rr,          X86::VFNMSUBSD4rm,          0           },
-    { X86::VFNMSUBPS4rr,          X86::VFNMSUBPS4rm,          TB_ALIGN_16 },
-    { X86::VFNMSUBPD4rr,          X86::VFNMSUBPD4rm,          TB_ALIGN_16 },
-    { X86::VFNMSUBPS4rrY,         X86::VFNMSUBPS4rmY,         TB_ALIGN_32 },
-    { X86::VFNMSUBPD4rrY,         X86::VFNMSUBPD4rmY,         TB_ALIGN_32 },
-    { X86::VFMADDSUBPS4rr,        X86::VFMADDSUBPS4rm,        TB_ALIGN_16 },
-    { X86::VFMADDSUBPD4rr,        X86::VFMADDSUBPD4rm,        TB_ALIGN_16 },
-    { X86::VFMADDSUBPS4rrY,       X86::VFMADDSUBPS4rmY,       TB_ALIGN_32 },
-    { X86::VFMADDSUBPD4rrY,       X86::VFMADDSUBPD4rmY,       TB_ALIGN_32 },
-    { X86::VFMSUBADDPS4rr,        X86::VFMSUBADDPS4rm,        TB_ALIGN_16 },
-    { X86::VFMSUBADDPD4rr,        X86::VFMSUBADDPD4rm,        TB_ALIGN_16 },
-    { X86::VFMSUBADDPS4rrY,       X86::VFMSUBADDPS4rmY,       TB_ALIGN_32 },
-    { X86::VFMSUBADDPD4rrY,       X86::VFMSUBADDPD4rmY,       TB_ALIGN_32 },
+    { X86::VFMADDSS4rr,           X86::VFMADDSS4rm,           TB_ALIGN_NONE },
+    { X86::VFMADDSD4rr,           X86::VFMADDSD4rm,           TB_ALIGN_NONE },
+    { X86::VFMADDPS4rr,           X86::VFMADDPS4rm,           TB_ALIGN_NONE },
+    { X86::VFMADDPD4rr,           X86::VFMADDPD4rm,           TB_ALIGN_NONE },
+    { X86::VFMADDPS4rrY,          X86::VFMADDPS4rmY,          TB_ALIGN_NONE },
+    { X86::VFMADDPD4rrY,          X86::VFMADDPD4rmY,          TB_ALIGN_NONE },
+    { X86::VFNMADDSS4rr,          X86::VFNMADDSS4rm,          TB_ALIGN_NONE },
+    { X86::VFNMADDSD4rr,          X86::VFNMADDSD4rm,          TB_ALIGN_NONE },
+    { X86::VFNMADDPS4rr,          X86::VFNMADDPS4rm,          TB_ALIGN_NONE },
+    { X86::VFNMADDPD4rr,          X86::VFNMADDPD4rm,          TB_ALIGN_NONE },
+    { X86::VFNMADDPS4rrY,         X86::VFNMADDPS4rmY,         TB_ALIGN_NONE },
+    { X86::VFNMADDPD4rrY,         X86::VFNMADDPD4rmY,         TB_ALIGN_NONE },
+    { X86::VFMSUBSS4rr,           X86::VFMSUBSS4rm,           TB_ALIGN_NONE },
+    { X86::VFMSUBSD4rr,           X86::VFMSUBSD4rm,           TB_ALIGN_NONE },
+    { X86::VFMSUBPS4rr,           X86::VFMSUBPS4rm,           TB_ALIGN_NONE },
+    { X86::VFMSUBPD4rr,           X86::VFMSUBPD4rm,           TB_ALIGN_NONE },
+    { X86::VFMSUBPS4rrY,          X86::VFMSUBPS4rmY,          TB_ALIGN_NONE },
+    { X86::VFMSUBPD4rrY,          X86::VFMSUBPD4rmY,          TB_ALIGN_NONE },
+    { X86::VFNMSUBSS4rr,          X86::VFNMSUBSS4rm,          TB_ALIGN_NONE },
+    { X86::VFNMSUBSD4rr,          X86::VFNMSUBSD4rm,          TB_ALIGN_NONE },
+    { X86::VFNMSUBPS4rr,          X86::VFNMSUBPS4rm,          TB_ALIGN_NONE },
+    { X86::VFNMSUBPD4rr,          X86::VFNMSUBPD4rm,          TB_ALIGN_NONE },
+    { X86::VFNMSUBPS4rrY,         X86::VFNMSUBPS4rmY,         TB_ALIGN_NONE },
+    { X86::VFNMSUBPD4rrY,         X86::VFNMSUBPD4rmY,         TB_ALIGN_NONE },
+    { X86::VFMADDSUBPS4rr,        X86::VFMADDSUBPS4rm,        TB_ALIGN_NONE },
+    { X86::VFMADDSUBPD4rr,        X86::VFMADDSUBPD4rm,        TB_ALIGN_NONE },
+    { X86::VFMADDSUBPS4rrY,       X86::VFMADDSUBPS4rmY,       TB_ALIGN_NONE },
+    { X86::VFMADDSUBPD4rrY,       X86::VFMADDSUBPD4rmY,       TB_ALIGN_NONE },
+    { X86::VFMSUBADDPS4rr,        X86::VFMSUBADDPS4rm,        TB_ALIGN_NONE },
+    { X86::VFMSUBADDPD4rr,        X86::VFMSUBADDPD4rm,        TB_ALIGN_NONE },
+    { X86::VFMSUBADDPS4rrY,       X86::VFMSUBADDPS4rmY,       TB_ALIGN_NONE },
+    { X86::VFMSUBADDPD4rrY,       X86::VFMSUBADDPD4rmY,       TB_ALIGN_NONE },
 
     // XOP foldable instructions
     { X86::VPCMOVrr,              X86::VPCMOVrm,              0 },