[llvm-commits] [llvm] r150477 - in /llvm/trunk: lib/CodeGen/SelectionDAG/LegalizeTypes.cpp test/CodeGen/X86/2012-02-14-scalar.ll

[Nadav Rotem]

Author: nadav
Date: Tue Feb 14 07:06:32 2012
New Revision: 150477

URL: http://llvm.org/viewvc/llvm-project?rev=150477&view=rev
Log:
Fix PR12000. Some vector operations may use scalar operands with types
that are greater than the vector element type. For example BUILD_VECTOR
of type <1 x i1> with a constant i8 operand.
This patch fixes the assertion.



Added:
    llvm/trunk/test/CodeGen/X86/2012-02-14-scalar.ll
Modified:
    llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp?rev=150477&r1=150476&r2=150477&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp Tue Feb 14 07:06:32 2012
@@ -748,7 +748,11 @@
 }
 
 void DAGTypeLegalizer::SetScalarizedVector(SDValue Op, SDValue Result) {
-  assert(Result.getValueType() == Op.getValueType().getVectorElementType() &&
+  // Note that in some cases vector operation operands may be greater than
+  // the vector element type. For example BUILD_VECTOR of type <1 x i1> with
+  // a constant i8 operand.
+  assert(Result.getValueType().getSizeInBits() >=
+         Op.getValueType().getVectorElementType().getSizeInBits() &&
          "Invalid type for scalarized vector");
   AnalyzeNewValue(Result);
 

Added: llvm/trunk/test/CodeGen/X86/2012-02-14-scalar.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2012-02-14-scalar.ll?rev=150477&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/2012-02-14-scalar.ll (added)
+++ llvm/trunk/test/CodeGen/X86/2012-02-14-scalar.ll Tue Feb 14 07:06:32 2012
@@ -0,0 +1,13 @@
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx -mattr=+avx
+target triple = "x86_64-unknown-linux-gnu"
+; Make sure we are not crashing on this one
+define void @autogen_28112_5000() {
+BB:
+  %S17 = icmp sgt <1 x i64> undef, undef
+  %E19 = extractelement <1 x i1> %S17, i32 0
+  br label %CF
+
+CF:                                               ; preds = %CF, %BB
+  %S23 = select i1 %E19, i8 undef, i8 undef
+  br label %CF
+}