From 5c2c94648e424d1c2ccfd7bd3c7147bb6389198d Mon Sep 17 00:00:00 2001
From: Kerry McLaughlin <kerry.mclaughlin@arm.com>
Date: Wed, 30 Oct 2019 11:44:29 +0000
Subject: [PATCH] [AArch64][SVE] Implement masked store intrinsics

Summary:
Adds support for codegen of masked stores, with non-truncating
and truncating variants.

Reviewers: huntergr, greened, dmgreen, rovka, sdesmalen

Reviewed By: dmgreen, sdesmalen

Subscribers: tschuett, kristof.beyls, hiraditya, rkruppe, psnobl, cfe-commits, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D69378
---
 llvm/lib/Target/AArch64/AArch64InstrInfo.td        |  28 ++++
 llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td     |  30 +++++
 .../Target/AArch64/AArch64TargetTransformInfo.h    |  10 +-
 .../test/CodeGen/AArch64/sve-masked-ldst-nonext.ll | 148 ++++++++++++++++++---
 llvm/test/CodeGen/AArch64/sve-masked-ldst-trunc.ll |  66 +++++++++
 5 files changed, 260 insertions(+), 22 deletions(-)
 create mode 100644 llvm/test/CodeGen/AArch64/sve-masked-ldst-trunc.ll

diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.td b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
index 4e1be26..c0cb471 100644
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
@@ -309,6 +309,34 @@ def zext_masked_load_i32 :
   return cast<MaskedLoadSDNode>(N)->getMemoryVT().getScalarType() == MVT::i32;
 }]>;
 
+// non-truncating masked store fragment.
+def nontrunc_masked_store :
+  PatFrag<(ops node:$val, node:$ptr, node:$pred),
+          (masked_st node:$val, node:$ptr, node:$pred), [{
+  return !cast<MaskedStoreSDNode>(N)->isTruncatingStore();
+}]>;
+// truncating masked store fragments.
+def trunc_masked_store :
+  PatFrag<(ops node:$val, node:$ptr, node:$pred),
+          (masked_st node:$val, node:$ptr, node:$pred), [{
+  return cast<MaskedStoreSDNode>(N)->isTruncatingStore();
+}]>;
+def trunc_masked_store_i8 :
+  PatFrag<(ops node:$val, node:$ptr, node:$pred),
+          (trunc_masked_store node:$val, node:$ptr, node:$pred), [{
+  return cast<MaskedStoreSDNode>(N)->getMemoryVT().getScalarType() == MVT::i8;
+}]>;
+def trunc_masked_store_i16 :
+  PatFrag<(ops node:$val, node:$ptr, node:$pred),
+          (trunc_masked_store node:$val, node:$ptr, node:$pred), [{
+  return cast<MaskedStoreSDNode>(N)->getMemoryVT().getScalarType() == MVT::i16;
+}]>;
+def trunc_masked_store_i32 :
+  PatFrag<(ops node:$val, node:$ptr, node:$pred),
+          (trunc_masked_store node:$val, node:$ptr, node:$pred), [{
+  return cast<MaskedStoreSDNode>(N)->getMemoryVT().getScalarType() == MVT::i32;
+}]>;
+
 // Node definitions.
 def AArch64adrp          : SDNode<"AArch64ISD::ADRP", SDTIntUnaryOp, []>;
 def AArch64adr           : SDNode<"AArch64ISD::ADR", SDTIntUnaryOp, []>;
diff --git a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
index 290bb32..502de47 100644
--- a/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64SVEInstrInfo.td
@@ -1108,6 +1108,36 @@ let Predicates = [HasSVE] in {
   // 16-element contiguous loads
   defm : pred_load<nxv16i8, nxv16i1, nonext_masked_load, LD1B_IMM>;
 
+  multiclass pred_store<ValueType Ty, ValueType PredTy, SDPatternOperator Store,
+                         Instruction RegImmInst> {
+    def _default : Pat<(Store (Ty ZPR:$vec), GPR64:$base, (PredTy PPR:$gp)),
+                       (RegImmInst ZPR:$vec, PPR:$gp, GPR64:$base, (i64 0))>;
+  }
+
+  // 2-element contiguous stores
+  defm : pred_store<nxv2i64, nxv2i1, trunc_masked_store_i8,  ST1B_D_IMM>;
+  defm : pred_store<nxv2i64, nxv2i1, trunc_masked_store_i16, ST1H_D_IMM>;
+  defm : pred_store<nxv2i64, nxv2i1, trunc_masked_store_i32, ST1W_D_IMM>;
+  defm : pred_store<nxv2i64, nxv2i1, nontrunc_masked_store,  ST1D_IMM>;
+  defm : pred_store<nxv2f16, nxv2i1, nontrunc_masked_store,  ST1H_D_IMM>;
+  defm : pred_store<nxv2f32, nxv2i1, nontrunc_masked_store,  ST1W_D_IMM>;
+  defm : pred_store<nxv2f64, nxv2i1, nontrunc_masked_store,  ST1D_IMM>;
+
+  // 4-element contiguous stores
+  defm : pred_store<nxv4i32, nxv4i1, trunc_masked_store_i8,  ST1B_S_IMM>;
+  defm : pred_store<nxv4i32, nxv4i1, trunc_masked_store_i16, ST1H_S_IMM>;
+  defm : pred_store<nxv4i32, nxv4i1, nontrunc_masked_store,  ST1W_IMM>;
+  defm : pred_store<nxv4f16, nxv4i1, nontrunc_masked_store,  ST1H_S_IMM>;
+  defm : pred_store<nxv4f32, nxv4i1, nontrunc_masked_store,  ST1W_IMM>;
+
+  // 8-element contiguous stores
+  defm : pred_store<nxv8i16, nxv8i1, trunc_masked_store_i8, ST1B_H_IMM>;
+  defm : pred_store<nxv8i16, nxv8i1, nontrunc_masked_store, ST1H_IMM>;
+  defm : pred_store<nxv8f16, nxv8i1, nontrunc_masked_store, ST1H_IMM>;
+
+  // 16-element contiguous stores
+  defm : pred_store<nxv16i8, nxv16i1, nontrunc_masked_store, ST1B_IMM>;
+
 }
 
 let Predicates = [HasSVE2] in {
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
index 9956c9c..8f313e3 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
@@ -147,7 +147,7 @@ public:
 
   bool getTgtMemIntrinsic(IntrinsicInst *Inst, MemIntrinsicInfo &Info);
 
-  bool isLegalMaskedLoad(Type *DataType, MaybeAlign Alignment) {
+  bool isLegalMaskedLoadStore(Type *DataType, MaybeAlign Alignment) {
     if (!isa<VectorType>(DataType) || !ST->hasSVE())
       return false;
 
@@ -162,6 +162,14 @@ public:
     return false;
   }
 
+  bool isLegalMaskedLoad(Type *DataType, MaybeAlign Alignment) {
+    return isLegalMaskedLoadStore(DataType, Alignment);
+  }
+
+  bool isLegalMaskedStore(Type *DataType, MaybeAlign Alignment) {
+    return isLegalMaskedLoadStore(DataType, Alignment);
+  }
+
   int getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy, unsigned Factor,
                                  ArrayRef<unsigned> Indices, unsigned Alignment,
                                  unsigned AddressSpace,
diff --git a/llvm/test/CodeGen/AArch64/sve-masked-ldst-nonext.ll b/llvm/test/CodeGen/AArch64/sve-masked-ldst-nonext.ll
index 461d290..b754266 100644
--- a/llvm/test/CodeGen/AArch64/sve-masked-ldst-nonext.ll
+++ b/llvm/test/CodeGen/AArch64/sve-masked-ldst-nonext.ll
@@ -1,79 +1,173 @@
-; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve < %s | FileCheck %s
+; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve -asm-verbose=0 < %s | FileCheck %s
 
 ;
 ; Masked Loads
 ;
 
-define <vscale x 2 x i64> @masked_load_nxv2i64(<vscale x 2 x i64> *%a, <vscale x 2 x i1> %mask) {
+define <vscale x 2 x i64> @masked_load_nxv2i64(<vscale x 2 x i64> *%a, <vscale x 2 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv2i64:
-; CHECK: ld1d { [[IN:z[0-9]+]].d }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1d { z0.d }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 2 x i64> @llvm.masked.load.nxv2i64(<vscale x 2 x i64> *%a, i32 8, <vscale x 2 x i1> %mask, <vscale x 2 x i64> undef)
   ret <vscale x 2 x i64> %load
 }
 
-define <vscale x 4 x i32> @masked_load_nxv4i32(<vscale x 4 x i32> *%a, <vscale x 4 x i1> %mask) {
+define <vscale x 4 x i32> @masked_load_nxv4i32(<vscale x 4 x i32> *%a, <vscale x 4 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv4i32:
-; CHECK: ld1w { [[IN:z[0-9]+]].s }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1w { z0.s }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32(<vscale x 4 x i32> *%a, i32 4, <vscale x 4 x i1> %mask, <vscale x 4 x i32> undef)
   ret <vscale x 4 x i32> %load
 }
 
-define <vscale x 8 x i16> @masked_load_nxv8i16(<vscale x 8 x i16> *%a, <vscale x 8 x i1> %mask) {
+define <vscale x 8 x i16> @masked_load_nxv8i16(<vscale x 8 x i16> *%a, <vscale x 8 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv8i16:
-; CHECK: ld1h { [[IN:z[0-9]+]].h }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1h { z0.h }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 8 x i16> @llvm.masked.load.nxv8i16(<vscale x 8 x i16> *%a, i32 2, <vscale x 8 x i1> %mask, <vscale x 8 x i16> undef)
   ret <vscale x 8 x i16> %load
 }
 
-define <vscale x 16 x i8> @masked_load_nxv16i8(<vscale x 16 x i8> *%a, <vscale x 16 x i1> %mask) {
+define <vscale x 16 x i8> @masked_load_nxv16i8(<vscale x 16 x i8> *%a, <vscale x 16 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv16i8:
-; CHECK: ld1b { [[IN:z[0-9]+]].b }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1b { z0.b }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 16 x i8> @llvm.masked.load.nxv16i8(<vscale x 16 x i8> *%a, i32 1, <vscale x 16 x i1> %mask, <vscale x 16 x i8> undef)
   ret <vscale x 16 x i8> %load
 }
 
-define <vscale x 2 x double> @masked_load_nxv2f64(<vscale x 2 x double> *%a, <vscale x 2 x i1> %mask) {
+define <vscale x 2 x double> @masked_load_nxv2f64(<vscale x 2 x double> *%a, <vscale x 2 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv2f64:
-; CHECK: ld1d { [[IN:z[0-9]+]].d }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1d { z0.d }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 2 x double> @llvm.masked.load.nxv2f64(<vscale x 2 x double> *%a, i32 8, <vscale x 2 x i1> %mask, <vscale x 2 x double> undef)
   ret <vscale x 2 x double> %load
 }
 
-define <vscale x 2 x float> @masked_load_nxv2f32(<vscale x 2 x float> *%a, <vscale x 2 x i1> %mask) {
+define <vscale x 2 x float> @masked_load_nxv2f32(<vscale x 2 x float> *%a, <vscale x 2 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv2f32:
-; CHECK: ld1w { [[IN:z[0-9]+]].d }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1w { z0.d }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 2 x float> @llvm.masked.load.nxv2f32(<vscale x 2 x float> *%a, i32 4, <vscale x 2 x i1> %mask, <vscale x 2 x float> undef)
   ret <vscale x 2 x float> %load
 }
 
-define <vscale x 2 x half> @masked_load_nxv2f16(<vscale x 2 x half> *%a, <vscale x 2 x i1> %mask) {
+define <vscale x 2 x half> @masked_load_nxv2f16(<vscale x 2 x half> *%a, <vscale x 2 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv2f16:
-; CHECK: ld1h { [[IN:z[0-9]+]].d }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1h { z0.d }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 2 x half> @llvm.masked.load.nxv2f16(<vscale x 2 x half> *%a, i32 2, <vscale x 2 x i1> %mask, <vscale x 2 x half> undef)
   ret <vscale x 2 x half> %load
 }
 
-define <vscale x 4 x float> @masked_load_nxv4f32(<vscale x 4 x float> *%a, <vscale x 4 x i1> %mask) {
+define <vscale x 4 x float> @masked_load_nxv4f32(<vscale x 4 x float> *%a, <vscale x 4 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv4f32:
-; CHECK: ld1w { [[IN:z[0-9]+]].s }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1w { z0.s }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 4 x float> @llvm.masked.load.nxv4f32(<vscale x 4 x float> *%a, i32 4, <vscale x 4 x i1> %mask, <vscale x 4 x float> undef)
   ret <vscale x 4 x float> %load
 }
 
-define <vscale x 4 x half> @masked_load_nxv4f16(<vscale x 4 x half> *%a, <vscale x 4 x i1> %mask) {
+define <vscale x 4 x half> @masked_load_nxv4f16(<vscale x 4 x half> *%a, <vscale x 4 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv4f16:
-; CHECK: ld1h { [[IN:z[0-9]+]].s }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1h { z0.s }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 4 x half> @llvm.masked.load.nxv4f16(<vscale x 4 x half> *%a, i32 2, <vscale x 4 x i1> %mask, <vscale x 4 x half> undef)
   ret <vscale x 4 x half> %load
 }
 
-define <vscale x 8 x half> @masked_load_nxv8f16(<vscale x 8 x half> *%a, <vscale x 8 x i1> %mask) {
+define <vscale x 8 x half> @masked_load_nxv8f16(<vscale x 8 x half> *%a, <vscale x 8 x i1> %mask) nounwind {
 ; CHECK-LABEL: masked_load_nxv8f16:
-; CHECK: ld1h { [[IN:z[0-9]+]].h }, [[PG:p[0-9]+]]/z, [x0]
+; CHECK-NEXT: ld1h { z0.h }, p0/z, [x0]
+; CHECK-NEXT: ret
   %load = call <vscale x 8 x half> @llvm.masked.load.nxv8f16(<vscale x 8 x half> *%a, i32 2, <vscale x 8 x i1> %mask, <vscale x 8 x half> undef)
   ret <vscale x 8 x half> %load
 }
 
+;
+; Masked Stores
+;
+
+define void @masked_store_nxv2i64(<vscale x 2 x i64> *%a, <vscale x 2 x i64> %val, <vscale x 2 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv2i64:
+; CHECK-NEXT: st1d { z0.d }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv2i64(<vscale x 2 x i64> %val, <vscale x 2 x i64> *%a, i32 8, <vscale x 2 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv4i32(<vscale x 4 x i32> *%a, <vscale x 4 x i32> %val, <vscale x 4 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv4i32:
+; CHECK-NEXT: st1w { z0.s }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv4i32(<vscale x 4 x i32> %val, <vscale x 4 x i32> *%a, i32 4, <vscale x 4 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv8i16(<vscale x 8 x i16> *%a, <vscale x 8 x i16> %val, <vscale x 8 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv8i16:
+; CHECK-NEXT: st1h { z0.h }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv8i16(<vscale x 8 x i16> %val, <vscale x 8 x i16> *%a, i32 2, <vscale x 8 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv16i8(<vscale x 16 x i8> *%a, <vscale x 16 x i8> %val, <vscale x 16 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv16i8:
+; CHECK-NEXT: st1b { z0.b }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv16i8(<vscale x 16 x i8> %val, <vscale x 16 x i8> *%a, i32 1, <vscale x 16 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv2f64(<vscale x 2 x double> *%a, <vscale x 2 x double> %val, <vscale x 2 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv2f64:
+; CHECK-NEXT: st1d { z0.d }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv2f64(<vscale x 2 x double> %val, <vscale x 2 x double> *%a, i32 8, <vscale x 2 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv2f32(<vscale x 2 x float> *%a, <vscale x 2 x float> %val, <vscale x 2 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv2f32:
+; CHECK-NEXT: st1w { z0.d }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv2f32(<vscale x 2 x float> %val, <vscale x 2 x float> *%a, i32 4, <vscale x 2 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv2f16(<vscale x 2 x half> *%a, <vscale x 2 x half> %val, <vscale x 2 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv2f16:
+; CHECK-NEXT: st1h { z0.d }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv2f16(<vscale x 2 x half> %val, <vscale x 2 x half> *%a, i32 4, <vscale x 2 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv4f32(<vscale x 4 x float> *%a, <vscale x 4 x float> %val, <vscale x 4 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv4f32:
+; CHECK-NEXT: st1w { z0.s }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv4f32(<vscale x 4 x float> %val, <vscale x 4 x float> *%a, i32 4, <vscale x 4 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv4f16(<vscale x 4 x half> *%a, <vscale x 4 x half> %val, <vscale x 4 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv4f16:
+; CHECK-NEXT: st1h { z0.s }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv4f16(<vscale x 4 x half> %val, <vscale x 4 x half> *%a, i32 2, <vscale x 4 x i1> %mask)
+  ret void
+}
+
+define void @masked_store_nxv8f16(<vscale x 8 x half> *%a, <vscale x 8 x half> %val, <vscale x 8 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_store_nxv8f16:
+; CHECK-NEXT: st1h { z0.h }, p0, [x0]
+; CHECK-NEXT: ret
+  call void @llvm.masked.store.nxv8f16(<vscale x 8 x half> %val, <vscale x 8 x half> *%a, i32 2, <vscale x 8 x i1> %mask)
+  ret void
+}
+
 declare <vscale x 2 x i64> @llvm.masked.load.nxv2i64(<vscale x 2 x i64>*, i32, <vscale x 2 x i1>, <vscale x 2 x i64>)
 declare <vscale x 4 x i32> @llvm.masked.load.nxv4i32(<vscale x 4 x i32>*, i32, <vscale x 4 x i1>, <vscale x 4 x i32>)
 declare <vscale x 8 x i16> @llvm.masked.load.nxv8i16(<vscale x 8 x i16>*, i32, <vscale x 8 x i1>, <vscale x 8 x i16>)
@@ -85,3 +179,15 @@ declare <vscale x 2 x half> @llvm.masked.load.nxv2f16(<vscale x 2 x half>*, i32,
 declare <vscale x 4 x float> @llvm.masked.load.nxv4f32(<vscale x 4 x float>*, i32, <vscale x 4 x i1>, <vscale x 4 x float>)
 declare <vscale x 4 x half> @llvm.masked.load.nxv4f16(<vscale x 4 x half>*, i32, <vscale x 4 x i1>, <vscale x 4 x half>)
 declare <vscale x 8 x half> @llvm.masked.load.nxv8f16(<vscale x 8 x half>*, i32, <vscale x 8 x i1>, <vscale x 8 x half>)
+
+declare void @llvm.masked.store.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>*, i32, <vscale x 2 x i1>)
+declare void @llvm.masked.store.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>*, i32, <vscale x 4 x i1>)
+declare void @llvm.masked.store.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>*, i32, <vscale x 8 x i1>)
+declare void @llvm.masked.store.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>*, i32, <vscale x 16 x i1>)
+
+declare void @llvm.masked.store.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>*, i32, <vscale x 2 x i1>)
+declare void @llvm.masked.store.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>*, i32, <vscale x 2 x i1>)
+declare void @llvm.masked.store.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>*, i32, <vscale x 2 x i1>)
+declare void @llvm.masked.store.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>*, i32, <vscale x 4 x i1>)
+declare void @llvm.masked.store.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>*, i32, <vscale x 4 x i1>)
+declare void @llvm.masked.store.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>*, i32, <vscale x 8 x i1>)
diff --git a/llvm/test/CodeGen/AArch64/sve-masked-ldst-trunc.ll b/llvm/test/CodeGen/AArch64/sve-masked-ldst-trunc.ll
new file mode 100644
index 0000000..d2069da
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/sve-masked-ldst-trunc.ll
@@ -0,0 +1,66 @@
+; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve -asm-verbose=0 < %s | FileCheck %s
+
+;
+; Masked Stores
+;
+
+define void @masked_trunc_store_nxv2i8(<vscale x 2 x i64> *%a, <vscale x 2 x i64> %val, <vscale x 2 x i8> *%b, <vscale x 2 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_trunc_store_nxv2i8:
+; CHECK-NEXT: st1b { z0.d }, p0, [x1]
+; CHECK-NEXT: ret
+  %trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i8>
+  call void @llvm.masked.store.nxv2i8(<vscale x 2 x i8> %trunc, <vscale x 2 x i8> *%b, i32 8, <vscale x 2 x i1> %mask)
+  ret void
+}
+
+define void @masked_trunc_store_nxv2i16(<vscale x 2 x i64> *%a, <vscale x 2 x i64> %val, <vscale x 2 x i16> *%b, <vscale x 2 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_trunc_store_nxv2i16:
+; CHECK-NEXT: st1h { z0.d }, p0, [x1]
+; CHECK-NEXT: ret
+  %trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i16>
+  call void @llvm.masked.store.nxv2i16(<vscale x 2 x i16> %trunc, <vscale x 2 x i16> *%b, i32 8, <vscale x 2 x i1> %mask)
+  ret void
+}
+
+define void @masked_trunc_store_nxv2i32(<vscale x 2 x i64> *%a, <vscale x 2 x i64> %val, <vscale x 2 x i32> *%b, <vscale x 2 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_trunc_store_nxv2i32:
+; CHECK-NEXT: st1w { z0.d }, p0, [x1]
+; CHECK-NEXT: ret
+  %trunc = trunc <vscale x 2 x i64> %val to <vscale x 2 x i32>
+  call void @llvm.masked.store.nxv2i32(<vscale x 2 x i32> %trunc, <vscale x 2 x i32> *%b, i32 8, <vscale x 2 x i1> %mask)
+  ret void
+}
+
+define void @masked_trunc_store_nxv4i8(<vscale x 4 x i32> *%a, <vscale x 4 x i32> %val, <vscale x 4 x i8> *%b, <vscale x 4 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_trunc_store_nxv4i8:
+; CHECK-NEXT: st1b { z0.s }, p0, [x1]
+; CHECK-NEXT: ret
+  %trunc = trunc <vscale x 4 x i32> %val to <vscale x 4 x i8>
+  call void @llvm.masked.store.nxv4i8(<vscale x 4 x i8> %trunc, <vscale x 4 x i8> *%b, i32 4, <vscale x 4 x i1> %mask)
+  ret void
+}
+
+define void @masked_trunc_store_nxv4i16(<vscale x 4 x i32> *%a, <vscale x 4 x i32> %val, <vscale x 4 x i16> *%b, <vscale x 4 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_trunc_store_nxv4i16:
+; CHECK-NEXT: st1h { z0.s }, p0, [x1]
+; CHECK-NEXT: ret
+  %trunc = trunc <vscale x 4 x i32> %val to <vscale x 4 x i16>
+  call void @llvm.masked.store.nxv4i16(<vscale x 4 x i16> %trunc, <vscale x 4 x i16> *%b, i32 4, <vscale x 4 x i1> %mask)
+  ret void
+}
+
+define void @masked_trunc_store_nxv8i8(<vscale x 8 x i16> *%a, <vscale x 8 x i16> %val, <vscale x 8 x i8> *%b, <vscale x 8 x i1> %mask) nounwind {
+; CHECK-LABEL: masked_trunc_store_nxv8i8:
+; CHECK-NEXT: st1b { z0.h }, p0, [x1]
+; CHECK-NEXT: ret
+  %trunc = trunc <vscale x 8 x i16> %val to <vscale x 8 x i8>
+  call void @llvm.masked.store.nxv8i8(<vscale x 8 x i8> %trunc, <vscale x 8 x i8> *%b, i32 2, <vscale x 8 x i1> %mask)
+  ret void
+}
+
+declare void @llvm.masked.store.nxv2i8(<vscale x 2 x i8>, <vscale x 2 x i8>*, i32, <vscale x 2 x i1>)
+declare void @llvm.masked.store.nxv2i16(<vscale x 2 x i16>, <vscale x 2 x i16>*, i32, <vscale x 2 x i1>)
+declare void @llvm.masked.store.nxv2i32(<vscale x 2 x i32>, <vscale x 2 x i32>*, i32, <vscale x 2 x i1>)
+declare void @llvm.masked.store.nxv4i8(<vscale x 4 x i8>, <vscale x 4 x i8>*, i32, <vscale x 4 x i1>)
+declare void @llvm.masked.store.nxv4i16(<vscale x 4 x i16>, <vscale x 4 x i16>*, i32, <vscale x 4 x i1>)
+declare void @llvm.masked.store.nxv8i8(<vscale x 8 x i8>, <vscale x 8 x i8>*, i32, <vscale x 8 x i1>)
-- 
2.7.4