diff --git a/llvm/test/Instrumentation/MemorySanitizer/AArch64/neon_vst_origins.ll b/llvm/test/Instrumentation/MemorySanitizer/AArch64/neon_vst_origins.ll
new file mode 100644
index 00000000000000..ff4c4c24bf37a7
--- /dev/null
+++ b/llvm/test/Instrumentation/MemorySanitizer/AArch64/neon_vst_origins.ll
@@ -0,0 +1,158 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
+; Test memory sanitizer instrumentation for Arm NEON VST instructions, with
+; origin tracking. These tests are deliberately shorter than neon_vst.ll, due
+; to the verbosity of the output.
+;
+; RUN: opt < %s -passes=msan -msan-track-origins=2 -S | FileCheck %s
+;
+; Forked from llvm/test/CodeGen/AArch64/arm64-st1.ll
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64--linux-android9001"
+
+; -----------------------------------------------------------------------------------------------------------------------------------------------
+
+define void @st2_16b(<16 x i8> %A, <16 x i8> %B, ptr %P) nounwind sanitize_memory {
+;
+; CHECK-LABEL: define void @st2_16b
+; CHECK-SAME: (<16 x i8> [[A:%.*]], <16 x i8> [[B:%.*]], ptr [[P:%.*]]) #[[ATTR0:[0-9]+]] {
+; CHECK-NEXT:    [[TMP1:%.*]] = load <16 x i8>, ptr @__msan_param_tls, align 8
+; CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr @__msan_param_origin_tls, align 4
+; CHECK-NEXT:    [[TMP3:%.*]] = load <16 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 16) to ptr), align 8
+; CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 16) to ptr), align 4
+; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
+; CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 32) to ptr), align 4
+; CHECK-NEXT:    call void @llvm.donothing()
+; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <16 x i8> [[TMP1]] to i128
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i128 [[TMP7]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP8:%.*]], label [[TMP9:%.*]], !prof [[PROF0:![0-9]+]]
+; CHECK:       8:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP2]]) #[[ATTR4:[0-9]+]]
+; CHECK-NEXT:    unreachable
+; CHECK:       9:
+; CHECK-NEXT:    [[TMP10:%.*]] = bitcast <16 x i8> [[TMP3]] to i128
+; CHECK-NEXT:    [[_MSCMP1:%.*]] = icmp ne i128 [[TMP10]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP1]], label [[TMP11:%.*]], label [[TMP12:%.*]], !prof [[PROF0]]
+; CHECK:       11:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP4]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       12:
+; CHECK-NEXT:    [[_MSCMP2:%.*]] = icmp ne i64 [[TMP5]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP2]], label [[TMP13:%.*]], label [[TMP14:%.*]], !prof [[PROF0]]
+; CHECK:       13:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP6]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       14:
+; CHECK-NEXT:    call void @llvm.aarch64.neon.st2.v16i8.p0(<16 x i8> [[A]], <16 x i8> [[B]], ptr [[P]])
+; CHECK-NEXT:    ret void
+;
+  call void @llvm.aarch64.neon.st2.v16i8.p0(<16 x i8> %A, <16 x i8> %B, ptr %P)
+  ret void
+}
+
+define void @st3_16b(<16 x i8> %A, <16 x i8> %B, <16 x i8> %C, ptr %P) nounwind sanitize_memory {
+;
+; CHECK-LABEL: define void @st3_16b
+; CHECK-SAME: (<16 x i8> [[A:%.*]], <16 x i8> [[B:%.*]], <16 x i8> [[C:%.*]], ptr [[P:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:    [[TMP1:%.*]] = load <16 x i8>, ptr @__msan_param_tls, align 8
+; CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr @__msan_param_origin_tls, align 4
+; CHECK-NEXT:    [[TMP3:%.*]] = load <16 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 16) to ptr), align 8
+; CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 16) to ptr), align 4
+; CHECK-NEXT:    [[TMP5:%.*]] = load <16 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
+; CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 32) to ptr), align 4
+; CHECK-NEXT:    [[TMP7:%.*]] = load i64, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 48) to ptr), align 8
+; CHECK-NEXT:    [[TMP8:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 48) to ptr), align 4
+; CHECK-NEXT:    call void @llvm.donothing()
+; CHECK-NEXT:    [[TMP9:%.*]] = bitcast <16 x i8> [[TMP1]] to i128
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i128 [[TMP9]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP10:%.*]], label [[TMP11:%.*]], !prof [[PROF0]]
+; CHECK:       10:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP2]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       11:
+; CHECK-NEXT:    [[TMP12:%.*]] = bitcast <16 x i8> [[TMP3]] to i128
+; CHECK-NEXT:    [[_MSCMP1:%.*]] = icmp ne i128 [[TMP12]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP1]], label [[TMP13:%.*]], label [[TMP14:%.*]], !prof [[PROF0]]
+; CHECK:       13:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP4]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       14:
+; CHECK-NEXT:    [[TMP15:%.*]] = bitcast <16 x i8> [[TMP5]] to i128
+; CHECK-NEXT:    [[_MSCMP2:%.*]] = icmp ne i128 [[TMP15]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP2]], label [[TMP16:%.*]], label [[TMP17:%.*]], !prof [[PROF0]]
+; CHECK:       16:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP6]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       17:
+; CHECK-NEXT:    [[_MSCMP3:%.*]] = icmp ne i64 [[TMP7]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP3]], label [[TMP18:%.*]], label [[TMP19:%.*]], !prof [[PROF0]]
+; CHECK:       18:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP8]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       19:
+; CHECK-NEXT:    call void @llvm.aarch64.neon.st3.v16i8.p0(<16 x i8> [[A]], <16 x i8> [[B]], <16 x i8> [[C]], ptr [[P]])
+; CHECK-NEXT:    ret void
+;
+  call void @llvm.aarch64.neon.st3.v16i8.p0(<16 x i8> %A, <16 x i8> %B, <16 x i8> %C, ptr %P)
+  ret void
+}
+
+define void @st4_16b(<16 x i8> %A, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, ptr %P) nounwind sanitize_memory {
+;
+; CHECK-LABEL: define void @st4_16b
+; CHECK-SAME: (<16 x i8> [[A:%.*]], <16 x i8> [[B:%.*]], <16 x i8> [[C:%.*]], <16 x i8> [[D:%.*]], ptr [[P:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:    [[TMP1:%.*]] = load <16 x i8>, ptr @__msan_param_tls, align 8
+; CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr @__msan_param_origin_tls, align 4
+; CHECK-NEXT:    [[TMP3:%.*]] = load <16 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 16) to ptr), align 8
+; CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 16) to ptr), align 4
+; CHECK-NEXT:    [[TMP5:%.*]] = load <16 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
+; CHECK-NEXT:    [[TMP6:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 32) to ptr), align 4
+; CHECK-NEXT:    [[TMP7:%.*]] = load <16 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 48) to ptr), align 8
+; CHECK-NEXT:    [[TMP8:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 48) to ptr), align 4
+; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
+; CHECK-NEXT:    [[TMP10:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_origin_tls to i64), i64 64) to ptr), align 4
+; CHECK-NEXT:    call void @llvm.donothing()
+; CHECK-NEXT:    [[TMP11:%.*]] = bitcast <16 x i8> [[TMP1]] to i128
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i128 [[TMP11]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP12:%.*]], label [[TMP13:%.*]], !prof [[PROF0]]
+; CHECK:       12:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP2]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       13:
+; CHECK-NEXT:    [[TMP14:%.*]] = bitcast <16 x i8> [[TMP3]] to i128
+; CHECK-NEXT:    [[_MSCMP1:%.*]] = icmp ne i128 [[TMP14]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP1]], label [[TMP15:%.*]], label [[TMP16:%.*]], !prof [[PROF0]]
+; CHECK:       15:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP4]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       16:
+; CHECK-NEXT:    [[TMP17:%.*]] = bitcast <16 x i8> [[TMP5]] to i128
+; CHECK-NEXT:    [[_MSCMP2:%.*]] = icmp ne i128 [[TMP17]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP2]], label [[TMP18:%.*]], label [[TMP19:%.*]], !prof [[PROF0]]
+; CHECK:       18:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP6]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       19:
+; CHECK-NEXT:    [[TMP20:%.*]] = bitcast <16 x i8> [[TMP7]] to i128
+; CHECK-NEXT:    [[_MSCMP3:%.*]] = icmp ne i128 [[TMP20]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP3]], label [[TMP21:%.*]], label [[TMP22:%.*]], !prof [[PROF0]]
+; CHECK:       21:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP8]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       22:
+; CHECK-NEXT:    [[_MSCMP4:%.*]] = icmp ne i64 [[TMP9]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP4]], label [[TMP23:%.*]], label [[TMP24:%.*]], !prof [[PROF0]]
+; CHECK:       23:
+; CHECK-NEXT:    call void @__msan_warning_with_origin_noreturn(i32 [[TMP10]]) #[[ATTR4]]
+; CHECK-NEXT:    unreachable
+; CHECK:       24:
+; CHECK-NEXT:    call void @llvm.aarch64.neon.st4.v16i8.p0(<16 x i8> [[A]], <16 x i8> [[B]], <16 x i8> [[C]], <16 x i8> [[D]], ptr [[P]])
+; CHECK-NEXT:    ret void
+;
+  call void @llvm.aarch64.neon.st4.v16i8.p0(<16 x i8> %A, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, ptr %P)
+  ret void
+}
+
+declare void @llvm.aarch64.neon.st2.v16i8.p0(<16 x i8>, <16 x i8>, ptr) nounwind sanitize_memory readonly
+declare void @llvm.aarch64.neon.st3.v16i8.p0(<16 x i8>, <16 x i8>, <16 x i8>, ptr) nounwind sanitize_memory readonly
+declare void @llvm.aarch64.neon.st4.v16i8.p0(<16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>, ptr) nounwind sanitize_memory readonly