[llvm] r264085 - [SelectionDAG] Ensure constant folded legalized vector element types are compatible with the BUILD_VECTOR type

[Simon Pilgrim via llvm-commits]

Author: rksimon
Date: Tue Mar 22 14:59:53 2016
New Revision: 264085

URL: http://llvm.org/viewvc/llvm-project?rev=264085&view=rev
Log:
[SelectionDAG] Ensure constant folded legalized vector element types are compatible with the BUILD_VECTOR type

Found during fuzz testing - 32-bit x86 targets were legalizing a <2 x i1> compare result to <2 x i32> when <2 x i64> was expected.

Added:
    llvm/trunk/test/CodeGen/X86/widen_compare-1.ll
Modified:
    llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp?rev=264085&r1=264084&r2=264085&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Tue Mar 22 14:59:53 2016
@@ -3403,7 +3403,7 @@ SDValue SelectionDAG::FoldConstantVector
   EVT LegalSVT = VT.getScalarType();
   if (LegalSVT.isInteger()) {
     LegalSVT = TLI->getTypeToTransformTo(*getContext(), LegalSVT);
-    if (LegalSVT.bitsLT(SVT))
+    if (LegalSVT.bitsLT(VT.getScalarType()))
       return SDValue();
   }
 

Added: llvm/trunk/test/CodeGen/X86/widen_compare-1.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/widen_compare-1.ll?rev=264085&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/widen_compare-1.ll (added)
+++ llvm/trunk/test/CodeGen/X86/widen_compare-1.ll Tue Mar 22 14:59:53 2016
@@ -0,0 +1,21 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i686-unknown -mattr=+sse4.2 | FileCheck %s --check-prefix=X86
+; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+sse4.2 | FileCheck %s --check-prefix=X64
+
+; compare v2i16
+
+define <2 x i16> @compare_v2i64_to_v2i16(<2 x i16>* %src) nounwind {
+; X86-LABEL: compare_v2i64_to_v2i16:
+; X86:       # BB#0:
+; X86-NEXT:    movaps {{.*#+}} xmm0 = [65535,0,65535,0]
+; X86-NEXT:    retl
+;
+; X64-LABEL: compare_v2i64_to_v2i16:
+; X64:       # BB#0:
+; X64-NEXT:    movaps {{.*#+}} xmm0 = [65535,65535]
+; X64-NEXT:    retq
+  %val = load <2 x i16>, <2 x i16>* %src, align 4
+  %cmp = icmp uge <2 x i16> %val, %val
+  %sel = select <2 x i1> %cmp, <2 x i16> <i16 -1, i16 -1>, <2 x i16> zeroinitializer
+  ret <2 x i16> %sel
+}