forked from OSchip/llvm-project
[Hexagon] SETEQ and SETNE are valid integer condition codes
llvm-svn: 323452
This commit is contained in:
parent
d22f07bbbe
commit
b2c458e648
|
@ -1289,7 +1289,8 @@ SDValue HexagonTargetLowering::LowerSETCC(SDValue Op, SelectionDAG &DAG) const {
|
|||
EVT RHSVT = RHS.getValueType();
|
||||
|
||||
if (LHSVT == MVT::v2i16) {
|
||||
assert(ISD::isSignedIntSetCC(CC) || ISD::isUnsignedIntSetCC(CC));
|
||||
assert(CC == ISD::SETEQ || CC == ISD::SETNE ||
|
||||
ISD::isSignedIntSetCC(CC) || ISD::isUnsignedIntSetCC(CC));
|
||||
unsigned ExtOpc = ISD::isSignedIntSetCC(CC) ? ISD::SIGN_EXTEND
|
||||
: ISD::ZERO_EXTEND;
|
||||
SDValue LX = DAG.getNode(ExtOpc, dl, MVT::v2i32, LHS);
|
||||
|
|
|
@ -0,0 +1,21 @@
|
|||
; RUN: llc -march=hexagon < %s | FileCheck %s
|
||||
; REQUIRES: asserts
|
||||
|
||||
; Check that this testcase doesn't crash.
|
||||
; CHECK: vcmpw.eq
|
||||
|
||||
target datalayout = "e-m:e-p:32:32:32-a:0-n16:32-i64:64:64-i32:32:32-i16:16:16-i1:8:8-f32:32:32-f64:64:64-v32:32:32-v64:64:64-v512:512:512-v1024:1024:1024-v2048:2048:2048"
|
||||
target triple = "hexagon"
|
||||
|
||||
define i32 @fred(<2 x i16>* %a0) #0 {
|
||||
b1:
|
||||
%v2 = load <2 x i16>, <2 x i16>* %a0, align 2
|
||||
%v3 = icmp eq <2 x i16> %v2, zeroinitializer
|
||||
%v4 = zext <2 x i1> %v3 to <2 x i16>
|
||||
%v5 = extractelement <2 x i16> %v4, i32 1
|
||||
%v8 = icmp ne i16 %v5, 1
|
||||
%v9 = zext i1 %v8 to i32
|
||||
ret i32 %v9
|
||||
}
|
||||
|
||||
attributes #0 = { nounwind "target-cpu"="hexagonv60" "target-features"="+hvx-length64b,+hvxv60" }
|
Loading…
Reference in New Issue