forked from OSchip/llvm-project
Fixed a bug in splitting Scatter operation in the Type Legalizer.
After the split of the Scatter operation, the order of the new instructions is well defined - Lo goes before Hi. Otherwise the semantic of Scatter (from LSB to MSB) is broken. I'm chaining 2 nodes to prevent reordering. Differential Revision https://reviews.llvm.org/D37670 llvm-svn: 312894
This commit is contained in:
parent
70a6929f1a
commit
cc477bbcea
|
@ -1970,13 +1970,12 @@ SDValue DAGTypeLegalizer::SplitVecOp_MSCATTER(MaskedScatterSDNode *N,
|
||||||
MachineMemOperand::MOStore, HiMemVT.getStoreSize(),
|
MachineMemOperand::MOStore, HiMemVT.getStoreSize(),
|
||||||
Alignment, N->getAAInfo(), N->getRanges());
|
Alignment, N->getAAInfo(), N->getRanges());
|
||||||
|
|
||||||
SDValue OpsHi[] = {Ch, DataHi, MaskHi, Ptr, IndexHi};
|
// The order of the Scatter operation after split is well defined. The "Hi"
|
||||||
Hi = DAG.getMaskedScatter(DAG.getVTList(MVT::Other), DataHi.getValueType(),
|
// part comes after the "Lo". So these two operations should be chained one
|
||||||
|
// after another.
|
||||||
|
SDValue OpsHi[] = {Lo, DataHi, MaskHi, Ptr, IndexHi};
|
||||||
|
return DAG.getMaskedScatter(DAG.getVTList(MVT::Other), DataHi.getValueType(),
|
||||||
DL, OpsHi, MMO);
|
DL, OpsHi, MMO);
|
||||||
|
|
||||||
// Build a factor node to remember that this store is independent of the
|
|
||||||
// other one.
|
|
||||||
return DAG.getNode(ISD::TokenFactor, DL, MVT::Other, Lo, Hi);
|
|
||||||
}
|
}
|
||||||
|
|
||||||
SDValue DAGTypeLegalizer::SplitVecOp_STORE(StoreSDNode *N, unsigned OpNo) {
|
SDValue DAGTypeLegalizer::SplitVecOp_STORE(StoreSDNode *N, unsigned OpNo) {
|
||||||
|
|
|
@ -5,16 +5,15 @@ target triple = "x86_64-unknown-linux-gnu"
|
||||||
|
|
||||||
; This test checks the order of scatter operations after split.
|
; This test checks the order of scatter operations after split.
|
||||||
; The right order is "from LSB to MSB", otherwise the semantic is broken.
|
; The right order is "from LSB to MSB", otherwise the semantic is broken.
|
||||||
; The submitted version of the test demonstrates the bug.
|
|
||||||
|
|
||||||
define void @test(i64 %x272, <16 x i32*> %x335, <16 x i32> %x270) {
|
define void @test(i64 %x272, <16 x i32*> %x335, <16 x i32> %x270) {
|
||||||
; CHECK-LABEL: test:
|
; CHECK-LABEL: test:
|
||||||
; CHECK: # BB#0:
|
; CHECK: # BB#0:
|
||||||
; CHECK-NEXT: vextracti64x4 $1, %zmm2, %ymm3
|
|
||||||
; CHECK-NEXT: kxnorw %k0, %k0, %k1
|
; CHECK-NEXT: kxnorw %k0, %k0, %k1
|
||||||
; CHECK-NEXT: kxnorw %k0, %k0, %k2
|
; CHECK-NEXT: kxnorw %k0, %k0, %k2
|
||||||
; CHECK-NEXT: vpscatterqd %ymm3, (,%zmm1) {%k2}
|
; CHECK-NEXT: vpscatterqd %ymm2, (,%zmm0) {%k2}
|
||||||
; CHECK-NEXT: vpscatterqd %ymm2, (,%zmm0) {%k1}
|
; CHECK-NEXT: vextracti64x4 $1, %zmm2, %ymm0
|
||||||
|
; CHECK-NEXT: vpscatterqd %ymm0, (,%zmm1) {%k1}
|
||||||
; CHECK-NEXT: vzeroupper
|
; CHECK-NEXT: vzeroupper
|
||||||
; CHECK-NEXT: retq
|
; CHECK-NEXT: retq
|
||||||
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %x270, <16 x i32*> %x335, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
|
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %x270, <16 x i32*> %x335, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
|
||||||
|
|
Loading…
Reference in New Issue