[PATCH] D49574: [CodeGen] Fix ICE in SelectionDAG::computeKnownBits

[Scott Linder via Phabricator via llvm-commits]

scott.linder updated this revision to Diff 156508.
scott.linder added a comment.

Addressed feedback; updated other places where DemandedElts is zero extended to the source vector width.

I do not have a FileCheck in the test because the passing condition is just for there to be no assert during CodeGen. There is nothing wrong with the IR/DAG/ISA at any point when the bug is or isn't present, so I don't know what to check for.


https://reviews.llvm.org/D49574

Files:
  lib/CodeGen/SelectionDAG/SelectionDAG.cpp
  lib/CodeGen/SelectionDAG/TargetLowering.cpp
  test/CodeGen/AMDGPU/extract-subvector-equal-length.ll


Index: test/CodeGen/AMDGPU/extract-subvector-equal-length.ll
===================================================================
--- /dev/null
+++ test/CodeGen/AMDGPU/extract-subvector-equal-length.ll
@@ -0,0 +1,22 @@
+; RUN: llc -march=amdgcn < %s
+
+; Test for ICE in SelectionDAG::computeKnownBits when visiting EXTRACT_SUBVECTOR
+; with equal length vector arguments.
+
+define <3 x i32> @quux() #0 {
+bb:
+  %tmp = shufflevector <4 x i8> <i8 1, i8 2, i8 3, i8 4>, <4 x i8> undef, <3 x i32> <i32 0, i32 1, i32 2>
+  %tmp1 = extractelement <3 x i8> %tmp, i64 0
+  %tmp2 = zext i8 %tmp1 to i32
+  %tmp3 = insertelement <3 x i32> undef, i32 %tmp2, i32 0
+  %tmp4 = extractelement <3 x i8> %tmp, i64 1
+  %tmp5 = zext i8 %tmp4 to i32
+  %tmp6 = insertelement <3 x i32> %tmp3, i32 %tmp5, i32 1
+  %tmp7 = extractelement <3 x i8> %tmp, i64 2
+  %tmp8 = zext i8 %tmp7 to i32
+  %tmp9 = insertelement <3 x i32> %tmp6, i32 %tmp8, i32 2
+  %tmp10 = lshr <3 x i32> %tmp9, <i32 1, i32 1, i32 1>
+  ret <3 x i32> %tmp10
+}
+
+attributes #0 = { noinline optnone }
Index: lib/CodeGen/SelectionDAG/TargetLowering.cpp
===================================================================
--- lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -1489,7 +1489,7 @@
       break;
     // Offset the demanded elts by the subvector index.
     uint64_t SubIdx = Idx.getZExtValue();
-    APInt SrcElts = DemandedElts.zext(NumSrcElts).shl(SubIdx);
+    APInt SrcElts = DemandedElts.zextOrSelf(NumSrcElts).shl(SubIdx);
     APInt SrcUndef, SrcZero;
     if (SimplifyDemandedVectorElts(Src, SrcElts, SrcUndef, SrcZero, TLO,
                                    Depth + 1))
Index: lib/CodeGen/SelectionDAG/SelectionDAG.cpp
===================================================================
--- lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -2346,7 +2346,7 @@
     if (SubIdx && SubIdx->getAPIntValue().ule(NumSrcElts - NumElts)) {
       // Offset the demanded elts by the subvector index.
       uint64_t Idx = SubIdx->getZExtValue();
-      APInt DemandedSrc = DemandedElts.zext(NumSrcElts).shl(Idx);
+      APInt DemandedSrc = DemandedElts.zextOrSelf(NumSrcElts).shl(Idx);
       computeKnownBits(Src, Known, DemandedSrc, Depth + 1);
     } else {
       computeKnownBits(Src, Known, Depth + 1);
@@ -2653,7 +2653,7 @@
   }
   case ISD::ZERO_EXTEND_VECTOR_INREG: {
     EVT InVT = Op.getOperand(0).getValueType();
-    APInt InDemandedElts = DemandedElts.zext(InVT.getVectorNumElements());
+    APInt InDemandedElts = DemandedElts.zextOrSelf(InVT.getVectorNumElements());
     computeKnownBits(Op.getOperand(0), Known, InDemandedElts, Depth + 1);
     Known = Known.zext(BitWidth);
     Known.Zero.setBitsFrom(InVT.getScalarSizeInBits());
@@ -3236,7 +3236,7 @@
   case ISD::SIGN_EXTEND_VECTOR_INREG: {
     SDValue Src = Op.getOperand(0);
     EVT SrcVT = Src.getValueType();
-    APInt DemandedSrcElts = DemandedElts.zext(SrcVT.getVectorNumElements());
+    APInt DemandedSrcElts = DemandedElts.zextOrSelf(SrcVT.getVectorNumElements());
     Tmp = VTBits - SrcVT.getScalarSizeInBits();
     return ComputeNumSignBits(Src, DemandedSrcElts, Depth+1) + Tmp;
   }
@@ -3505,7 +3505,7 @@
     if (SubIdx && SubIdx->getAPIntValue().ule(NumSrcElts - NumElts)) {
       // Offset the demanded elts by the subvector index.
       uint64_t Idx = SubIdx->getZExtValue();
-      APInt DemandedSrc = DemandedElts.zext(NumSrcElts).shl(Idx);
+      APInt DemandedSrc = DemandedElts.zextOrSelf(NumSrcElts).shl(Idx);
       return ComputeNumSignBits(Src, DemandedSrc, Depth + 1);
     }
     return ComputeNumSignBits(Src, Depth + 1);


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D49574.156508.patch
Type: text/x-patch
Size: 3683 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180720/a64c450a/attachment.bin>