[llvm] r301002 - ARM: don't try to create an i8 -> i32 vpaddl.
Tim Northover via llvm-commits
llvm-commits at lists.llvm.org
Fri Apr 21 10:21:59 PDT 2017
Author: tnorthover
Date: Fri Apr 21 12:21:59 2017
New Revision: 301002
URL: http://llvm.org/viewvc/llvm-project?rev=301002&view=rev
Log:
ARM: don't try to create an i8 -> i32 vpaddl.
DAG combine was mistakenly assuming that the step-up it was looking at was
always a doubling, but it can sometimes be a larger extension in which case
we'd crash.
Modified:
llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
llvm/trunk/test/CodeGen/ARM/vpadd.ll
Modified: llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp?rev=301002&r1=301001&r2=301002&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp Fri Apr 21 12:21:59 2017
@@ -9480,8 +9480,11 @@ AddCombineBUILD_VECTORToVPADDL(SDNode *N
return SDValue();
}
- // Don't generate vpaddl+vmovn; we'll match it to vpadd later.
- if (Vec.getValueType().getVectorElementType() == VT.getVectorElementType())
+ // Don't generate vpaddl+vmovn; we'll match it to vpadd later. Also don't try
+ // to handle an i8 -> i32 situation (or similar). vpaddl can only double the
+ // size.
+ if (2 * Vec.getValueType().getVectorElementType().getSizeInBits() !=
+ VT.getVectorElementType().getSizeInBits())
return SDValue();
// Create VPADDL node.
Modified: llvm/trunk/test/CodeGen/ARM/vpadd.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/vpadd.ll?rev=301002&r1=301001&r2=301002&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/vpadd.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/vpadd.ll Fri Apr 21 12:21:59 2017
@@ -485,6 +485,17 @@ define <2 x i16> @fromExtendingExtractVe
ret <2 x i16> %x
}
+; And <2 x i8> to <2 x i32>
+define <2 x i8> @fromExtendingExtractVectorElt_2i8(<8 x i8> %in) {
+; CHECK-LABEL: fromExtendingExtractVectorElt_2i8:
+; CHECK: vadd.i32
+ %tmp1 = shufflevector <8 x i8> %in, <8 x i8> undef, <2 x i32> <i32 0, i32 2>
+ %tmp2 = shufflevector <8 x i8> %in, <8 x i8> undef, <2 x i32> <i32 1, i32 3>
+ %x = add <2 x i8> %tmp2, %tmp1
+ ret <2 x i8> %x
+}
+
+
declare <4 x i16> @llvm.arm.neon.vpaddls.v4i16.v8i8(<8 x i8>) nounwind readnone
declare <2 x i32> @llvm.arm.neon.vpaddls.v2i32.v4i16(<4 x i16>) nounwind readnone
declare <1 x i64> @llvm.arm.neon.vpaddls.v1i64.v2i32(<2 x i32>) nounwind readnone
More information about the llvm-commits
mailing list