[llvm-commits] [llvm] r154798 - in /llvm/trunk/lib/Target/X86: X86ISelLowering.cpp X86InstrFragmentsSIMD.td

[Craig Topper]

Author: ctopper
Date: Mon Apr 16 01:43:40 2012
New Revision: 154798

URL: http://llvm.org/viewvc/llvm-project?rev=154798&view=rev
Log:
Change type profile for vpermv back to using operand type for the mask argument to match intrinsic behavior. Add a bitcast to the lowering code to convert mask from v8i32 to v8f32 for vpermps.

Modified:
    llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
    llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td

Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=154798&r1=154797&r2=154798&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Mon Apr 16 01:43:40 2012
@@ -6655,9 +6655,11 @@
     for (unsigned i = 0; i != 8; ++i) {
       permclMask.push_back(DAG.getConstant((M[i]>=0) ? M[i] : 0, MVT::i32));
     }
+    SDValue Mask = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i32,
+                               &permclMask[0], 8);
+    // Bitcast is for VPERMPS since mask is v8i32 but node takes v8f32
     return DAG.getNode(X86ISD::VPERMV, dl, VT,
-                       DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i32,
-                                   &permclMask[0], 8), V1);
+                       DAG.getNode(ISD::BITCAST, dl, VT, Mask), V1);
   }
 
   if (V2IsUndef && HasAVX2 && (VT == MVT::v4i64 || VT == MVT::v4f64))

Modified: llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td?rev=154798&r1=154797&r2=154798&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td Mon Apr 16 01:43:40 2012
@@ -124,8 +124,6 @@
                                  SDTCisSameAs<0,1>, SDTCisInt<2>]>;
 def SDTShuff3OpI : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
                                  SDTCisSameAs<0,2>, SDTCisInt<3>]>;
-def SDTVPermv : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVT<1, v8i32>,
-                              SDTCisSameAs<0,2>]>;
 
 def SDTVBroadcast : SDTypeProfile<1, 1, [SDTCisVec<0>]>;
 def SDTBlend : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
@@ -157,7 +155,7 @@
 def X86Unpckh : SDNode<"X86ISD::UNPCKH", SDTShuff2Op>;
 
 def X86VPermilp  : SDNode<"X86ISD::VPERMILP", SDTShuff2OpI>;
-def X86VPermv    : SDNode<"X86ISD::VPERMV",   SDTVPermv>;
+def X86VPermv    : SDNode<"X86ISD::VPERMV",   SDTShuff2Op>;
 def X86VPermi    : SDNode<"X86ISD::VPERMI",   SDTShuff2OpI>;
 
 def X86VPerm2x128 : SDNode<"X86ISD::VPERM2X128", SDTShuff3OpI>;