From 6505124a0c7c648560aad88bac103b0738a8b5f0 Mon Sep 17 00:00:00 2001
From: David Green <david.green@arm.com>
Date: Wed, 12 Feb 2020 18:18:23 +0000
Subject: [PATCH] [ARM] Extra vmovn tests to show BE differences. NFC

---
 llvm/test/CodeGen/ARM/neon-vmovn.ll   | 794 ++++++++++++++++++++++++++++++++++
 llvm/test/CodeGen/Thumb2/mve-vmovn.ll | 322 ++++++++++++++
 2 files changed, 1116 insertions(+)
 create mode 100644 llvm/test/CodeGen/ARM/neon-vmovn.ll

diff --git a/llvm/test/CodeGen/ARM/neon-vmovn.ll b/llvm/test/CodeGen/ARM/neon-vmovn.ll
new file mode 100644
index 0000000..675a38a
--- /dev/null
+++ b/llvm/test/CodeGen/ARM/neon-vmovn.ll
@@ -0,0 +1,794 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=armv8-arm-none-eabi -verify-machineinstrs %s -o - | FileCheck %s
+; RUN: llc -mtriple=armebv8-arm-none-eabi -verify-machineinstrs %s -o - | FileCheck %s --check-prefix=CHECKBE
+
+; This is the same as Thumb2/mve-vmovn.ll, testing the same patterns for neon
+; under both both LE and BE. The vmovn instruction is very different between
+; mve and neon, so these tests are not necessarily expected to generate a (neon)
+; vmovn.
+
+define arm_aapcs_vfpcc <8 x i16> @vmovn32_trunc1(<4 x i32> %src1, <4 x i32> %src2) {
+; CHECK-LABEL: vmovn32_trunc1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vzip.32 q0, q1
+; CHECK-NEXT:    vmovn.i32 d17, q1
+; CHECK-NEXT:    vmovn.i32 d16, q0
+; CHECK-NEXT:    vorr q0, q8, q8
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_trunc1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q8, q1
+; CHECKBE-NEXT:    vrev64.32 q9, q0
+; CHECKBE-NEXT:    vzip.32 q9, q8
+; CHECKBE-NEXT:    vmovn.i32 d17, q8
+; CHECKBE-NEXT:    vmovn.i32 d16, q9
+; CHECKBE-NEXT:    vrev64.16 q0, q8
+; CHECKBE-NEXT:    bx lr
+entry:
+  %strided.vec = shufflevector <4 x i32> %src1, <4 x i32> %src2, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
+  %out = trunc <8 x i32> %strided.vec to <8 x i16>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vmovn32_trunc2(<4 x i32> %src1, <4 x i32> %src2) {
+; CHECK-LABEL: vmovn32_trunc2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vzip.32 q1, q0
+; CHECK-NEXT:    vmovn.i32 d1, q0
+; CHECK-NEXT:    vmovn.i32 d0, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_trunc2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q8, q0
+; CHECKBE-NEXT:    vrev64.32 q9, q1
+; CHECKBE-NEXT:    vzip.32 q9, q8
+; CHECKBE-NEXT:    vmovn.i32 d17, q8
+; CHECKBE-NEXT:    vmovn.i32 d16, q9
+; CHECKBE-NEXT:    vrev64.16 q0, q8
+; CHECKBE-NEXT:    bx lr
+entry:
+  %strided.vec = shufflevector <4 x i32> %src1, <4 x i32> %src2, <8 x i32> <i32 4, i32 0, i32 5, i32 1, i32 6, i32 2, i32 7, i32 3>
+  %out = trunc <8 x i32> %strided.vec to <8 x i16>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vmovn16_trunc1(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: vmovn16_trunc1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vzip.16 q0, q1
+; CHECK-NEXT:    vmovn.i16 d17, q1
+; CHECK-NEXT:    vmovn.i16 d16, q0
+; CHECK-NEXT:    vorr q0, q8, q8
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_trunc1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q8, q1
+; CHECKBE-NEXT:    vrev64.16 q9, q0
+; CHECKBE-NEXT:    vzip.16 q9, q8
+; CHECKBE-NEXT:    vmovn.i16 d17, q8
+; CHECKBE-NEXT:    vmovn.i16 d16, q9
+; CHECKBE-NEXT:    vrev64.8 q0, q8
+; CHECKBE-NEXT:    bx lr
+entry:
+  %strided.vec = shufflevector <8 x i16> %src1, <8 x i16> %src2, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
+  %out = trunc <16 x i16> %strided.vec to <16 x i8>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vmovn16_trunc2(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: vmovn16_trunc2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vzip.16 q1, q0
+; CHECK-NEXT:    vmovn.i16 d1, q0
+; CHECK-NEXT:    vmovn.i16 d0, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_trunc2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q8, q0
+; CHECKBE-NEXT:    vrev64.16 q9, q1
+; CHECKBE-NEXT:    vzip.16 q9, q8
+; CHECKBE-NEXT:    vmovn.i16 d17, q8
+; CHECKBE-NEXT:    vmovn.i16 d16, q9
+; CHECKBE-NEXT:    vrev64.8 q0, q8
+; CHECKBE-NEXT:    bx lr
+entry:
+  %strided.vec = shufflevector <8 x i16> %src1, <8 x i16> %src2, <16 x i32> <i32 8, i32 0, i32 9, i32 1, i32 10, i32 2, i32 11, i32 3, i32 12, i32 4, i32 13, i32 5, i32 14, i32 6, i32 15, i32 7>
+  %out = trunc <16 x i16> %strided.vec to <16 x i8>
+  ret <16 x i8> %out
+}
+
+
+define arm_aapcs_vfpcc <2 x i64> @vmovn64_t1(<2 x i64> %src1, <2 x i64> %src2) {
+; CHECK-LABEL: vmovn64_t1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f64 d1, d2
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_t1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f64 d1, d2
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 0, i32 2>
+  ret <2 x i64> %out
+}
+
+define arm_aapcs_vfpcc <2 x i64> @vmovn64_t2(<2 x i64> %src1, <2 x i64> %src2) {
+; CHECK-LABEL: vmovn64_t2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vorr d3, d0, d0
+; CHECK-NEXT:    vorr q0, q1, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_t2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vorr d3, d0, d0
+; CHECKBE-NEXT:    vorr q0, q1, q1
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 2, i32 0>
+  ret <2 x i64> %out
+}
+
+define arm_aapcs_vfpcc <2 x i64> @vmovn64_b1(<2 x i64> %src1, <2 x i64> %src2) {
+; CHECK-LABEL: vmovn64_b1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f64 d1, d3
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_b1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f64 d1, d3
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 0, i32 3>
+  ret <2 x i64> %out
+}
+
+define arm_aapcs_vfpcc <2 x i64> @vmovn64_b2(<2 x i64> %src1, <2 x i64> %src2) {
+; CHECK-LABEL: vmovn64_b2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f64 d16, d3
+; CHECK-NEXT:    vorr d17, d0, d0
+; CHECK-NEXT:    vorr q0, q8, q8
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_b2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f64 d16, d3
+; CHECKBE-NEXT:    vorr d17, d0, d0
+; CHECKBE-NEXT:    vorr q0, q8, q8
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 3, i32 0>
+  ret <2 x i64> %out
+}
+
+define arm_aapcs_vfpcc <2 x i64> @vmovn64_b3(<2 x i64> %src1, <2 x i64> %src2) {
+; CHECK-LABEL: vmovn64_b3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f64 d0, d1
+; CHECK-NEXT:    vmov.f64 d1, d2
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_b3:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f64 d0, d1
+; CHECKBE-NEXT:    vmov.f64 d1, d2
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 1, i32 2>
+  ret <2 x i64> %out
+}
+
+define arm_aapcs_vfpcc <2 x i64> @vmovn64_b4(<2 x i64> %src1, <2 x i64> %src2) {
+; CHECK-LABEL: vmovn64_b4:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vorr d3, d1, d1
+; CHECK-NEXT:    vorr q0, q1, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_b4:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vorr d3, d1, d1
+; CHECKBE-NEXT:    vorr q0, q1, q1
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 2, i32 1>
+  ret <2 x i64> %out
+}
+
+
+
+define arm_aapcs_vfpcc <4 x i32> @vmovn32_t1(<4 x i32> %src1, <4 x i32> %src2) {
+; CHECK-LABEL: vmovn32_t1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vtrn.32 q0, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_t1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q8, q1
+; CHECKBE-NEXT:    vrev64.32 q9, q0
+; CHECKBE-NEXT:    vtrn.32 q9, q8
+; CHECKBE-NEXT:    vrev64.32 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @vmovn32_t2(<4 x i32> %src1, <4 x i32> %src2) {
+; CHECK-LABEL: vmovn32_t2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vtrn.32 q1, q0
+; CHECK-NEXT:    vorr q0, q1, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_t2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q8, q0
+; CHECKBE-NEXT:    vrev64.32 q9, q1
+; CHECKBE-NEXT:    vtrn.32 q9, q8
+; CHECKBE-NEXT:    vrev64.32 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @vmovn32_b1(<4 x i32> %src1, <4 x i32> %src2) {
+; CHECK-LABEL: vmovn32_b1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev64.32 q8, q0
+; CHECK-NEXT:    vtrn.32 q8, q1
+; CHECK-NEXT:    vorr q0, q1, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_b1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q8, q0
+; CHECKBE-NEXT:    vrev64.32 q9, q1
+; CHECKBE-NEXT:    vrev64.32 q8, q8
+; CHECKBE-NEXT:    vtrn.32 q8, q9
+; CHECKBE-NEXT:    vrev64.32 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @vmovn32_b2(<4 x i32> %src1, <4 x i32> %src2) {
+; CHECK-LABEL: vmovn32_b2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vext.32 q8, q0, q0, #1
+; CHECK-NEXT:    vtrn.32 q8, q1
+; CHECK-NEXT:    vext.32 q0, q1, q1, #1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_b2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q8, q0
+; CHECKBE-NEXT:    vrev64.32 q9, q1
+; CHECKBE-NEXT:    vext.32 q8, q8, q8, #1
+; CHECKBE-NEXT:    vtrn.32 q8, q9
+; CHECKBE-NEXT:    vext.32 q8, q9, q9, #1
+; CHECKBE-NEXT:    vrev64.32 q0, q8
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 5, i32 0, i32 7, i32 2>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @vmovn32_b3(<4 x i32> %src1, <4 x i32> %src2) {
+; CHECK-LABEL: vmovn32_b3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vorr q8, q0, q0
+; CHECK-NEXT:    vtrn.32 q8, q1
+; CHECK-NEXT:    vtrn.32 q0, q8
+; CHECK-NEXT:    vorr q0, q8, q8
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_b3:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q9, q0
+; CHECKBE-NEXT:    vrev64.32 q8, q1
+; CHECKBE-NEXT:    vorr q10, q9, q9
+; CHECKBE-NEXT:    vtrn.32 q10, q8
+; CHECKBE-NEXT:    vtrn.32 q9, q10
+; CHECKBE-NEXT:    vrev64.32 q0, q10
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 1, i32 4, i32 3, i32 6>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @vmovn32_b4(<4 x i32> %src1, <4 x i32> %src2) {
+; CHECK-LABEL: vmovn32_b4:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vorr q8, q0, q0
+; CHECK-NEXT:    vtrn.32 q8, q1
+; CHECK-NEXT:    vtrn.32 q8, q0
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_b4:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q9, q0
+; CHECKBE-NEXT:    vrev64.32 q8, q1
+; CHECKBE-NEXT:    vorr q10, q9, q9
+; CHECKBE-NEXT:    vtrn.32 q10, q8
+; CHECKBE-NEXT:    vtrn.32 q10, q9
+; CHECKBE-NEXT:    vrev64.32 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
+  ret <4 x i32> %out
+}
+
+
+
+
+define arm_aapcs_vfpcc <8 x i16> @vmovn16_t1(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: vmovn16_t1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vtrn.16 q0, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_t1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q8, q1
+; CHECKBE-NEXT:    vrev64.16 q9, q0
+; CHECKBE-NEXT:    vtrn.16 q9, q8
+; CHECKBE-NEXT:    vrev64.16 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vmovn16_t2(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: vmovn16_t2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vtrn.16 q1, q0
+; CHECK-NEXT:    vorr q0, q1, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_t2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q8, q0
+; CHECKBE-NEXT:    vrev64.16 q9, q1
+; CHECKBE-NEXT:    vtrn.16 q9, q8
+; CHECKBE-NEXT:    vrev64.16 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 8, i32 0, i32 10, i32 2, i32 12, i32 4, i32 14, i32 6>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vmovn16_b1(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: vmovn16_b1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev32.16 d16, d1
+; CHECK-NEXT:    vrev32.16 d17, d0
+; CHECK-NEXT:    vtrn.16 d16, d3
+; CHECK-NEXT:    vtrn.16 d17, d2
+; CHECK-NEXT:    vorr q0, q1, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_b1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 d16, d1
+; CHECKBE-NEXT:    vrev64.16 d17, d0
+; CHECKBE-NEXT:    vrev64.16 d19, d3
+; CHECKBE-NEXT:    vrev32.16 d16, d16
+; CHECKBE-NEXT:    vrev64.16 d18, d2
+; CHECKBE-NEXT:    vrev32.16 d17, d17
+; CHECKBE-NEXT:    vtrn.16 d16, d19
+; CHECKBE-NEXT:    vtrn.16 d17, d18
+; CHECKBE-NEXT:    vrev64.16 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vmovn16_b2(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: vmovn16_b2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vorr d17, d3, d3
+; CHECK-NEXT:    vtrn.16 d17, d1
+; CHECK-NEXT:    vorr d16, d2, d2
+; CHECK-NEXT:    vtrn.16 d16, d0
+; CHECK-NEXT:    vtrn.16 d3, d17
+; CHECK-NEXT:    vtrn.16 d2, d16
+; CHECK-NEXT:    vorr q0, q8, q8
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_b2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 d17, d3
+; CHECKBE-NEXT:    vorr d21, d17, d17
+; CHECKBE-NEXT:    vrev64.16 d16, d1
+; CHECKBE-NEXT:    vrev64.16 d19, d2
+; CHECKBE-NEXT:    vrev64.16 d18, d0
+; CHECKBE-NEXT:    vtrn.16 d21, d16
+; CHECKBE-NEXT:    vorr d20, d19, d19
+; CHECKBE-NEXT:    vtrn.16 d20, d18
+; CHECKBE-NEXT:    vtrn.16 d17, d21
+; CHECKBE-NEXT:    vtrn.16 d19, d20
+; CHECKBE-NEXT:    vrev64.16 q0, q10
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 9, i32 0, i32 11, i32 2, i32 13, i32 4, i32 15, i32 6>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vmovn16_b3(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: vmovn16_b3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vorr d17, d1, d1
+; CHECK-NEXT:    vtrn.16 d17, d3
+; CHECK-NEXT:    vorr d16, d0, d0
+; CHECK-NEXT:    vtrn.16 d16, d2
+; CHECK-NEXT:    vtrn.16 d1, d17
+; CHECK-NEXT:    vtrn.16 d0, d16
+; CHECK-NEXT:    vorr q0, q8, q8
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_b3:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 d17, d1
+; CHECKBE-NEXT:    vorr d21, d17, d17
+; CHECKBE-NEXT:    vrev64.16 d16, d3
+; CHECKBE-NEXT:    vrev64.16 d19, d0
+; CHECKBE-NEXT:    vrev64.16 d18, d2
+; CHECKBE-NEXT:    vtrn.16 d21, d16
+; CHECKBE-NEXT:    vorr d20, d19, d19
+; CHECKBE-NEXT:    vtrn.16 d20, d18
+; CHECKBE-NEXT:    vtrn.16 d17, d21
+; CHECKBE-NEXT:    vtrn.16 d19, d20
+; CHECKBE-NEXT:    vrev64.16 q0, q10
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 1, i32 8, i32 3, i32 10, i32 5, i32 12, i32 7, i32 14>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vmovn16_b4(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: vmovn16_b4:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev32.16 d16, d3
+; CHECK-NEXT:    vrev32.16 d17, d2
+; CHECK-NEXT:    vtrn.16 d16, d1
+; CHECK-NEXT:    vtrn.16 d17, d0
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_b4:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 d16, d3
+; CHECKBE-NEXT:    vrev64.16 d17, d2
+; CHECKBE-NEXT:    vrev64.16 d19, d1
+; CHECKBE-NEXT:    vrev32.16 d16, d16
+; CHECKBE-NEXT:    vrev64.16 d18, d0
+; CHECKBE-NEXT:    vrev32.16 d17, d17
+; CHECKBE-NEXT:    vtrn.16 d16, d19
+; CHECKBE-NEXT:    vtrn.16 d17, d18
+; CHECKBE-NEXT:    vrev64.16 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
+  ret <8 x i16> %out
+}
+
+
+define arm_aapcs_vfpcc <16 x i8> @vmovn8_b1(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: vmovn8_b1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vtrn.8 q0, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn8_b1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q8, q1
+; CHECKBE-NEXT:    vrev64.8 q9, q0
+; CHECKBE-NEXT:    vtrn.8 q9, q8
+; CHECKBE-NEXT:    vrev64.8 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vmovn8_b2(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: vmovn8_b2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vtrn.8 q1, q0
+; CHECK-NEXT:    vorr q0, q1, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn8_b2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q8, q0
+; CHECKBE-NEXT:    vrev64.8 q9, q1
+; CHECKBE-NEXT:    vtrn.8 q9, q8
+; CHECKBE-NEXT:    vrev64.8 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 16, i32 0, i32 18, i32 2, i32 20, i32 4, i32 22, i32 6, i32 24, i32 8, i32 26, i32 10, i32 28, i32 12, i32 30, i32 14>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vmovn8_t1(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: vmovn8_t1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vorr q2, q0, q0
+; CHECK-NEXT:    vldr d16, .LCPI24_0
+; CHECK-NEXT:    vorr d6, d3, d3
+; CHECK-NEXT:    vtbl.8 d1, {d5, d6}, d16
+; CHECK-NEXT:    vorr d5, d2, d2
+; CHECK-NEXT:    vtbl.8 d0, {d4, d5}, d16
+; CHECK-NEXT:    bx lr
+; CHECK-NEXT:    .p2align 3
+; CHECK-NEXT:  @ %bb.1:
+; CHECK-NEXT:  .LCPI24_0:
+; CHECK-NEXT:    .byte 0 @ 0x0
+; CHECK-NEXT:    .byte 9 @ 0x9
+; CHECK-NEXT:    .byte 2 @ 0x2
+; CHECK-NEXT:    .byte 11 @ 0xb
+; CHECK-NEXT:    .byte 4 @ 0x4
+; CHECK-NEXT:    .byte 13 @ 0xd
+; CHECK-NEXT:    .byte 6 @ 0x6
+; CHECK-NEXT:    .byte 15 @ 0xf
+;
+; CHECKBE-LABEL: vmovn8_t1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vldr d16, .LCPI24_0
+; CHECKBE-NEXT:    vrev64.8 d19, d3
+; CHECKBE-NEXT:    vrev64.8 d21, d2
+; CHECKBE-NEXT:    vrev64.8 d18, d1
+; CHECKBE-NEXT:    vrev64.8 d16, d16
+; CHECKBE-NEXT:    vrev64.8 d20, d0
+; CHECKBE-NEXT:    vtbl.8 d19, {d18, d19}, d16
+; CHECKBE-NEXT:    vtbl.8 d18, {d20, d21}, d16
+; CHECKBE-NEXT:    vrev64.8 q0, q9
+; CHECKBE-NEXT:    bx lr
+; CHECKBE-NEXT:    .p2align 3
+; CHECKBE-NEXT:  @ %bb.1:
+; CHECKBE-NEXT:  .LCPI24_0:
+; CHECKBE-NEXT:    .byte 0 @ 0x0
+; CHECKBE-NEXT:    .byte 9 @ 0x9
+; CHECKBE-NEXT:    .byte 2 @ 0x2
+; CHECKBE-NEXT:    .byte 11 @ 0xb
+; CHECKBE-NEXT:    .byte 4 @ 0x4
+; CHECKBE-NEXT:    .byte 13 @ 0xd
+; CHECKBE-NEXT:    .byte 6 @ 0x6
+; CHECKBE-NEXT:    .byte 15 @ 0xf
+entry:
+  %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vmovn8_t2(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: vmovn8_t2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    @ kill: def $q1 killed $q1 def $d2_d3_d4
+; CHECK-NEXT:    vldr d18, .LCPI25_0
+; CHECK-NEXT:    vorr d4, d1, d1
+; CHECK-NEXT:    vtbl.8 d17, {d3, d4}, d18
+; CHECK-NEXT:    vorr d3, d0, d0
+; CHECK-NEXT:    vtbl.8 d16, {d2, d3}, d18
+; CHECK-NEXT:    vorr q0, q8, q8
+; CHECK-NEXT:    bx lr
+; CHECK-NEXT:    .p2align 3
+; CHECK-NEXT:  @ %bb.1:
+; CHECK-NEXT:  .LCPI25_0:
+; CHECK-NEXT:    .byte 1 @ 0x1
+; CHECK-NEXT:    .byte 8 @ 0x8
+; CHECK-NEXT:    .byte 3 @ 0x3
+; CHECK-NEXT:    .byte 10 @ 0xa
+; CHECK-NEXT:    .byte 5 @ 0x5
+; CHECK-NEXT:    .byte 12 @ 0xc
+; CHECK-NEXT:    .byte 7 @ 0x7
+; CHECK-NEXT:    .byte 14 @ 0xe
+;
+; CHECKBE-LABEL: vmovn8_t2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vldr d16, .LCPI25_0
+; CHECKBE-NEXT:    vrev64.8 d19, d1
+; CHECKBE-NEXT:    vrev64.8 d21, d0
+; CHECKBE-NEXT:    vrev64.8 d18, d3
+; CHECKBE-NEXT:    vrev64.8 d16, d16
+; CHECKBE-NEXT:    vrev64.8 d20, d2
+; CHECKBE-NEXT:    vtbl.8 d19, {d18, d19}, d16
+; CHECKBE-NEXT:    vtbl.8 d18, {d20, d21}, d16
+; CHECKBE-NEXT:    vrev64.8 q0, q9
+; CHECKBE-NEXT:    bx lr
+; CHECKBE-NEXT:    .p2align 3
+; CHECKBE-NEXT:  @ %bb.1:
+; CHECKBE-NEXT:  .LCPI25_0:
+; CHECKBE-NEXT:    .byte 1 @ 0x1
+; CHECKBE-NEXT:    .byte 8 @ 0x8
+; CHECKBE-NEXT:    .byte 3 @ 0x3
+; CHECKBE-NEXT:    .byte 10 @ 0xa
+; CHECKBE-NEXT:    .byte 5 @ 0x5
+; CHECKBE-NEXT:    .byte 12 @ 0xc
+; CHECKBE-NEXT:    .byte 7 @ 0x7
+; CHECKBE-NEXT:    .byte 14 @ 0xe
+entry:
+  %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 17, i32 0, i32 19, i32 2, i32 21, i32 4, i32 23, i32 6, i32 25, i32 8, i32 27, i32 10, i32 29, i32 12, i32 31, i32 14>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vmovn8_t3(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: vmovn8_t3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vorr q2, q0, q0
+; CHECK-NEXT:    vldr d16, .LCPI26_0
+; CHECK-NEXT:    vorr d6, d3, d3
+; CHECK-NEXT:    vtbl.8 d1, {d5, d6}, d16
+; CHECK-NEXT:    vorr d5, d2, d2
+; CHECK-NEXT:    vtbl.8 d0, {d4, d5}, d16
+; CHECK-NEXT:    bx lr
+; CHECK-NEXT:    .p2align 3
+; CHECK-NEXT:  @ %bb.1:
+; CHECK-NEXT:  .LCPI26_0:
+; CHECK-NEXT:    .byte 1 @ 0x1
+; CHECK-NEXT:    .byte 8 @ 0x8
+; CHECK-NEXT:    .byte 3 @ 0x3
+; CHECK-NEXT:    .byte 10 @ 0xa
+; CHECK-NEXT:    .byte 5 @ 0x5
+; CHECK-NEXT:    .byte 12 @ 0xc
+; CHECK-NEXT:    .byte 7 @ 0x7
+; CHECK-NEXT:    .byte 14 @ 0xe
+;
+; CHECKBE-LABEL: vmovn8_t3:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vldr d16, .LCPI26_0
+; CHECKBE-NEXT:    vrev64.8 d19, d3
+; CHECKBE-NEXT:    vrev64.8 d21, d2
+; CHECKBE-NEXT:    vrev64.8 d18, d1
+; CHECKBE-NEXT:    vrev64.8 d16, d16
+; CHECKBE-NEXT:    vrev64.8 d20, d0
+; CHECKBE-NEXT:    vtbl.8 d19, {d18, d19}, d16
+; CHECKBE-NEXT:    vtbl.8 d18, {d20, d21}, d16
+; CHECKBE-NEXT:    vrev64.8 q0, q9
+; CHECKBE-NEXT:    bx lr
+; CHECKBE-NEXT:    .p2align 3
+; CHECKBE-NEXT:  @ %bb.1:
+; CHECKBE-NEXT:  .LCPI26_0:
+; CHECKBE-NEXT:    .byte 1 @ 0x1
+; CHECKBE-NEXT:    .byte 8 @ 0x8
+; CHECKBE-NEXT:    .byte 3 @ 0x3
+; CHECKBE-NEXT:    .byte 10 @ 0xa
+; CHECKBE-NEXT:    .byte 5 @ 0x5
+; CHECKBE-NEXT:    .byte 12 @ 0xc
+; CHECKBE-NEXT:    .byte 7 @ 0x7
+; CHECKBE-NEXT:    .byte 14 @ 0xe
+entry:
+  %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 1, i32 16, i32 3, i32 18, i32 5, i32 20, i32 7, i32 22, i32 9, i32 24, i32 11, i32 26, i32 13, i32 28, i32 15, i32 30>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vmovn8_t4(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: vmovn8_t4:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    @ kill: def $q1 killed $q1 def $d2_d3_d4
+; CHECK-NEXT:    vldr d18, .LCPI27_0
+; CHECK-NEXT:    vorr d4, d1, d1
+; CHECK-NEXT:    vtbl.8 d17, {d3, d4}, d18
+; CHECK-NEXT:    vorr d3, d0, d0
+; CHECK-NEXT:    vtbl.8 d16, {d2, d3}, d18
+; CHECK-NEXT:    vorr q0, q8, q8
+; CHECK-NEXT:    bx lr
+; CHECK-NEXT:    .p2align 3
+; CHECK-NEXT:  @ %bb.1:
+; CHECK-NEXT:  .LCPI27_0:
+; CHECK-NEXT:    .byte 0 @ 0x0
+; CHECK-NEXT:    .byte 9 @ 0x9
+; CHECK-NEXT:    .byte 2 @ 0x2
+; CHECK-NEXT:    .byte 11 @ 0xb
+; CHECK-NEXT:    .byte 4 @ 0x4
+; CHECK-NEXT:    .byte 13 @ 0xd
+; CHECK-NEXT:    .byte 6 @ 0x6
+; CHECK-NEXT:    .byte 15 @ 0xf
+;
+; CHECKBE-LABEL: vmovn8_t4:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vldr d16, .LCPI27_0
+; CHECKBE-NEXT:    vrev64.8 d19, d1
+; CHECKBE-NEXT:    vrev64.8 d21, d0
+; CHECKBE-NEXT:    vrev64.8 d18, d3
+; CHECKBE-NEXT:    vrev64.8 d16, d16
+; CHECKBE-NEXT:    vrev64.8 d20, d2
+; CHECKBE-NEXT:    vtbl.8 d19, {d18, d19}, d16
+; CHECKBE-NEXT:    vtbl.8 d18, {d20, d21}, d16
+; CHECKBE-NEXT:    vrev64.8 q0, q9
+; CHECKBE-NEXT:    bx lr
+; CHECKBE-NEXT:    .p2align 3
+; CHECKBE-NEXT:  @ %bb.1:
+; CHECKBE-NEXT:  .LCPI27_0:
+; CHECKBE-NEXT:    .byte 0 @ 0x0
+; CHECKBE-NEXT:    .byte 9 @ 0x9
+; CHECKBE-NEXT:    .byte 2 @ 0x2
+; CHECKBE-NEXT:    .byte 11 @ 0xb
+; CHECKBE-NEXT:    .byte 4 @ 0x4
+; CHECKBE-NEXT:    .byte 13 @ 0xd
+; CHECKBE-NEXT:    .byte 6 @ 0x6
+; CHECKBE-NEXT:    .byte 15 @ 0xf
+entry:
+  %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @test(<8 x i16> %src1, <8 x i16> %src2) {
+; CHECK-LABEL: test:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vtrn.8 q0, q1
+; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: test:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q8, q1
+; CHECKBE-NEXT:    vrev64.8 q9, q0
+; CHECKBE-NEXT:    vtrn.8 q9, q8
+; CHECKBE-NEXT:    vrev64.8 q0, q9
+; CHECKBE-NEXT:    bx lr
+entry:
+  %a0 = extractelement <8 x i16> %src1, i32 0
+  %a1 = extractelement <8 x i16> %src1, i32 1
+  %a2 = extractelement <8 x i16> %src1, i32 2
+  %a3 = extractelement <8 x i16> %src1, i32 3
+  %a4 = extractelement <8 x i16> %src1, i32 4
+  %a5 = extractelement <8 x i16> %src1, i32 5
+  %a6 = extractelement <8 x i16> %src1, i32 6
+  %a7 = extractelement <8 x i16> %src1, i32 7
+
+  %b0 = extractelement <8 x i16> %src2, i32 0
+  %b1 = extractelement <8 x i16> %src2, i32 1
+  %b2 = extractelement <8 x i16> %src2, i32 2
+  %b3 = extractelement <8 x i16> %src2, i32 3
+  %b4 = extractelement <8 x i16> %src2, i32 4
+  %b5 = extractelement <8 x i16> %src2, i32 5
+  %b6 = extractelement <8 x i16> %src2, i32 6
+  %b7 = extractelement <8 x i16> %src2, i32 7
+
+  %s0 = trunc i16 %a0 to i8
+  %s1 = trunc i16 %a1 to i8
+  %s2 = trunc i16 %a2 to i8
+  %s3 = trunc i16 %a3 to i8
+  %s4 = trunc i16 %a4 to i8
+  %s5 = trunc i16 %a5 to i8
+  %s6 = trunc i16 %a6 to i8
+  %s7 = trunc i16 %a7 to i8
+  %t0 = trunc i16 %b0 to i8
+  %t1 = trunc i16 %b1 to i8
+  %t2 = trunc i16 %b2 to i8
+  %t3 = trunc i16 %b3 to i8
+  %t4 = trunc i16 %b4 to i8
+  %t5 = trunc i16 %b5 to i8
+  %t6 = trunc i16 %b6 to i8
+  %t7 = trunc i16 %b7 to i8
+
+  %r0 = insertelement <16 x i8> undef, i8 %s0, i32 0
+  %r1 = insertelement <16 x i8> %r0, i8 %s1, i32 2
+  %r2 = insertelement <16 x i8> %r1, i8 %s2, i32 4
+  %r3 = insertelement <16 x i8> %r2, i8 %s3, i32 6
+  %r4 = insertelement <16 x i8> %r3, i8 %s4, i32 8
+  %r5 = insertelement <16 x i8> %r4, i8 %s5, i32 10
+  %r6 = insertelement <16 x i8> %r5, i8 %s6, i32 12
+  %r7 = insertelement <16 x i8> %r6, i8 %s7, i32 14
+  %r10 = insertelement <16 x i8> %r7, i8 %t0, i32 1
+  %r11 = insertelement <16 x i8> %r10, i8 %t1, i32 3
+  %r12 = insertelement <16 x i8> %r11, i8 %t2, i32 5
+  %r13 = insertelement <16 x i8> %r12, i8 %t3, i32 7
+  %r14 = insertelement <16 x i8> %r13, i8 %t4, i32 9
+  %r15 = insertelement <16 x i8> %r14, i8 %t5, i32 11
+  %r16 = insertelement <16 x i8> %r15, i8 %t6, i32 13
+  %r17 = insertelement <16 x i8> %r16, i8 %t7, i32 15
+
+  ret <16 x i8> %r17
+}
diff --git a/llvm/test/CodeGen/Thumb2/mve-vmovn.ll b/llvm/test/CodeGen/Thumb2/mve-vmovn.ll
index 20f790d..ec34bdd 100644
--- a/llvm/test/CodeGen/Thumb2/mve-vmovn.ll
+++ b/llvm/test/CodeGen/Thumb2/mve-vmovn.ll
@@ -1,11 +1,20 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=thumbv8.1m.main-arm-none-eabi -mattr=+mve -verify-machineinstrs %s -o - | FileCheck %s
+; RUN: llc -mtriple=thumbebv8.1m.main-arm-none-eabi -mattr=+mve -verify-machineinstrs %s -o - | FileCheck %s --check-prefix=CHECKBE
 
 define arm_aapcs_vfpcc <8 x i16> @vmovn32_trunc1(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-LABEL: vmovn32_trunc1:
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    vmovnt.i32 q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_trunc1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q2, q1
+; CHECKBE-NEXT:    vrev64.16 q1, q0
+; CHECKBE-NEXT:    vmovnt.i32 q1, q2
+; CHECKBE-NEXT:    vrev64.16 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %strided.vec = shufflevector <4 x i32> %src1, <4 x i32> %src2, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
   %out = trunc <8 x i32> %strided.vec to <8 x i16>
@@ -18,6 +27,14 @@ define arm_aapcs_vfpcc <8 x i16> @vmovn32_trunc2(<4 x i32> %src1, <4 x i32> %src
 ; CHECK-NEXT:    vmovnt.i32 q1, q0
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_trunc2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q2, q0
+; CHECKBE-NEXT:    vrev64.16 q3, q1
+; CHECKBE-NEXT:    vmovnt.i32 q3, q2
+; CHECKBE-NEXT:    vrev64.16 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %strided.vec = shufflevector <4 x i32> %src1, <4 x i32> %src2, <8 x i32> <i32 4, i32 0, i32 5, i32 1, i32 6, i32 2, i32 7, i32 3>
   %out = trunc <8 x i32> %strided.vec to <8 x i16>
@@ -29,6 +46,14 @@ define arm_aapcs_vfpcc <16 x i8> @vmovn16_trunc1(<8 x i16> %src1, <8 x i16> %src
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    vmovnt.i16 q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_trunc1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q2, q1
+; CHECKBE-NEXT:    vrev64.8 q1, q0
+; CHECKBE-NEXT:    vmovnt.i16 q1, q2
+; CHECKBE-NEXT:    vrev64.8 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %strided.vec = shufflevector <8 x i16> %src1, <8 x i16> %src2, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
   %out = trunc <16 x i16> %strided.vec to <16 x i8>
@@ -41,6 +66,14 @@ define arm_aapcs_vfpcc <16 x i8> @vmovn16_trunc2(<8 x i16> %src1, <8 x i16> %src
 ; CHECK-NEXT:    vmovnt.i16 q1, q0
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_trunc2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q2, q0
+; CHECKBE-NEXT:    vrev64.8 q3, q1
+; CHECKBE-NEXT:    vmovnt.i16 q3, q2
+; CHECKBE-NEXT:    vrev64.8 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %strided.vec = shufflevector <8 x i16> %src1, <8 x i16> %src2, <16 x i32> <i32 8, i32 0, i32 9, i32 1, i32 10, i32 2, i32 11, i32 3, i32 12, i32 4, i32 13, i32 5, i32 14, i32 6, i32 15, i32 7>
   %out = trunc <16 x i16> %strided.vec to <16 x i8>
@@ -54,6 +87,12 @@ define arm_aapcs_vfpcc <2 x i64> @vmovn64_t1(<2 x i64> %src1, <2 x i64> %src2) {
 ; CHECK-NEXT:    vmov.f32 s2, s4
 ; CHECK-NEXT:    vmov.f32 s3, s5
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_t1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f32 s2, s4
+; CHECKBE-NEXT:    vmov.f32 s3, s5
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 0, i32 2>
   ret <2 x i64> %out
@@ -66,6 +105,13 @@ define arm_aapcs_vfpcc <2 x i64> @vmovn64_t2(<2 x i64> %src1, <2 x i64> %src2) {
 ; CHECK-NEXT:    vmov.f32 s7, s1
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_t2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f32 s6, s0
+; CHECKBE-NEXT:    vmov.f32 s7, s1
+; CHECKBE-NEXT:    vmov q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 2, i32 0>
   ret <2 x i64> %out
@@ -77,6 +123,12 @@ define arm_aapcs_vfpcc <2 x i64> @vmovn64_b1(<2 x i64> %src1, <2 x i64> %src2) {
 ; CHECK-NEXT:    vmov.f32 s2, s6
 ; CHECK-NEXT:    vmov.f32 s3, s7
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_b1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f32 s2, s6
+; CHECKBE-NEXT:    vmov.f32 s3, s7
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 0, i32 3>
   ret <2 x i64> %out
@@ -91,6 +143,15 @@ define arm_aapcs_vfpcc <2 x i64> @vmovn64_b2(<2 x i64> %src1, <2 x i64> %src2) {
 ; CHECK-NEXT:    vmov.f32 s7, s1
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_b2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f32 s4, s6
+; CHECKBE-NEXT:    vmov.f32 s5, s7
+; CHECKBE-NEXT:    vmov.f32 s6, s0
+; CHECKBE-NEXT:    vmov.f32 s7, s1
+; CHECKBE-NEXT:    vmov q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 3, i32 0>
   ret <2 x i64> %out
@@ -104,6 +165,14 @@ define arm_aapcs_vfpcc <2 x i64> @vmovn64_b3(<2 x i64> %src1, <2 x i64> %src2) {
 ; CHECK-NEXT:    vmov.f32 s2, s4
 ; CHECK-NEXT:    vmov.f32 s3, s5
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_b3:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f32 s0, s2
+; CHECKBE-NEXT:    vmov.f32 s1, s3
+; CHECKBE-NEXT:    vmov.f32 s2, s4
+; CHECKBE-NEXT:    vmov.f32 s3, s5
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 1, i32 2>
   ret <2 x i64> %out
@@ -116,6 +185,13 @@ define arm_aapcs_vfpcc <2 x i64> @vmovn64_b4(<2 x i64> %src1, <2 x i64> %src2) {
 ; CHECK-NEXT:    vmov.f32 s7, s3
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn64_b4:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vmov.f32 s6, s2
+; CHECKBE-NEXT:    vmov.f32 s7, s3
+; CHECKBE-NEXT:    vmov q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <2 x i64> %src1, <2 x i64> %src2, <2 x i32> <i32 2, i32 1>
   ret <2 x i64> %out
@@ -129,6 +205,15 @@ define arm_aapcs_vfpcc <4 x i32> @vmovn32_t1(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-NEXT:    vmov.f32 s1, s4
 ; CHECK-NEXT:    vmov.f32 s3, s6
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_t1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q2, q1
+; CHECKBE-NEXT:    vrev64.32 q1, q0
+; CHECKBE-NEXT:    vmov.f32 s5, s8
+; CHECKBE-NEXT:    vmov.f32 s7, s10
+; CHECKBE-NEXT:    vrev64.32 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
   ret <4 x i32> %out
@@ -141,6 +226,15 @@ define arm_aapcs_vfpcc <4 x i32> @vmovn32_t2(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-NEXT:    vmov.f32 s7, s2
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_t2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q2, q0
+; CHECKBE-NEXT:    vrev64.32 q3, q1
+; CHECKBE-NEXT:    vmov.f32 s13, s8
+; CHECKBE-NEXT:    vmov.f32 s15, s10
+; CHECKBE-NEXT:    vrev64.32 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 4, i32 0, i32 6, i32 2>
   ret <4 x i32> %out
@@ -152,6 +246,15 @@ define arm_aapcs_vfpcc <4 x i32> @vmovn32_b1(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-NEXT:    vmov.f32 s1, s5
 ; CHECK-NEXT:    vmov.f32 s3, s7
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_b1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q2, q1
+; CHECKBE-NEXT:    vrev64.32 q1, q0
+; CHECKBE-NEXT:    vmov.f32 s5, s9
+; CHECKBE-NEXT:    vmov.f32 s7, s11
+; CHECKBE-NEXT:    vrev64.32 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
   ret <4 x i32> %out
@@ -166,6 +269,17 @@ define arm_aapcs_vfpcc <4 x i32> @vmovn32_b2(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-NEXT:    vmov.f32 s11, s2
 ; CHECK-NEXT:    vmov q0, q2
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_b2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q2, q0
+; CHECKBE-NEXT:    vrev64.32 q0, q1
+; CHECKBE-NEXT:    vmov.f32 s4, s1
+; CHECKBE-NEXT:    vmov.f32 s5, s8
+; CHECKBE-NEXT:    vmov.f32 s6, s3
+; CHECKBE-NEXT:    vmov.f32 s7, s10
+; CHECKBE-NEXT:    vrev64.32 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 5, i32 0, i32 7, i32 2>
   ret <4 x i32> %out
@@ -180,6 +294,17 @@ define arm_aapcs_vfpcc <4 x i32> @vmovn32_b3(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-NEXT:    vmov.f32 s11, s6
 ; CHECK-NEXT:    vmov q0, q2
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_b3:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q2, q1
+; CHECKBE-NEXT:    vrev64.32 q1, q0
+; CHECKBE-NEXT:    vmov.f32 s12, s5
+; CHECKBE-NEXT:    vmov.f32 s13, s8
+; CHECKBE-NEXT:    vmov.f32 s14, s7
+; CHECKBE-NEXT:    vmov.f32 s15, s10
+; CHECKBE-NEXT:    vrev64.32 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 1, i32 4, i32 3, i32 6>
   ret <4 x i32> %out
@@ -192,6 +317,15 @@ define arm_aapcs_vfpcc <4 x i32> @vmovn32_b4(<4 x i32> %src1, <4 x i32> %src2) {
 ; CHECK-NEXT:    vmov.f32 s7, s3
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn32_b4:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.32 q2, q0
+; CHECKBE-NEXT:    vrev64.32 q3, q1
+; CHECKBE-NEXT:    vmov.f32 s13, s9
+; CHECKBE-NEXT:    vmov.f32 s15, s11
+; CHECKBE-NEXT:    vrev64.32 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <4 x i32> %src1, <4 x i32> %src2, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
   ret <4 x i32> %out
@@ -205,6 +339,14 @@ define arm_aapcs_vfpcc <8 x i16> @vmovn16_t1(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    vmovnt.i32 q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_t1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q2, q1
+; CHECKBE-NEXT:    vrev64.16 q1, q0
+; CHECKBE-NEXT:    vmovnt.i32 q1, q2
+; CHECKBE-NEXT:    vrev64.16 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
   ret <8 x i16> %out
@@ -216,6 +358,14 @@ define arm_aapcs_vfpcc <8 x i16> @vmovn16_t2(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-NEXT:    vmovnt.i32 q1, q0
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_t2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q2, q0
+; CHECKBE-NEXT:    vrev64.16 q3, q1
+; CHECKBE-NEXT:    vmovnt.i32 q3, q2
+; CHECKBE-NEXT:    vrev64.16 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 8, i32 0, i32 10, i32 2, i32 12, i32 4, i32 14, i32 6>
   ret <8 x i16> %out
@@ -227,6 +377,14 @@ define arm_aapcs_vfpcc <8 x i16> @vmovn16_b1(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-NEXT:    vmovnb.i32 q1, q0
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_b1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q2, q0
+; CHECKBE-NEXT:    vrev64.16 q3, q1
+; CHECKBE-NEXT:    vmovnb.i32 q3, q2
+; CHECKBE-NEXT:    vrev64.16 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
   ret <8 x i16> %out
@@ -253,6 +411,29 @@ define arm_aapcs_vfpcc <8 x i16> @vmovn16_b2(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-NEXT:    vmov.u16 r0, q2[6]
 ; CHECK-NEXT:    vmov.16 q0[7], r0
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_b2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q2, q1
+; CHECKBE-NEXT:    vrev64.16 q3, q0
+; CHECKBE-NEXT:    vmov.u16 r0, q2[1]
+; CHECKBE-NEXT:    vmov.16 q1[0], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q3[0]
+; CHECKBE-NEXT:    vmov.16 q1[1], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q2[3]
+; CHECKBE-NEXT:    vmov.16 q1[2], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q3[2]
+; CHECKBE-NEXT:    vmov.16 q1[3], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q2[5]
+; CHECKBE-NEXT:    vmov.16 q1[4], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q3[4]
+; CHECKBE-NEXT:    vmov.16 q1[5], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q2[7]
+; CHECKBE-NEXT:    vmov.16 q1[6], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q3[6]
+; CHECKBE-NEXT:    vmov.16 q1[7], r0
+; CHECKBE-NEXT:    vrev64.16 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 9, i32 0, i32 11, i32 2, i32 13, i32 4, i32 15, i32 6>
   ret <8 x i16> %out
@@ -279,6 +460,29 @@ define arm_aapcs_vfpcc <8 x i16> @vmovn16_b3(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK-NEXT:    vmov.u16 r0, q1[6]
 ; CHECK-NEXT:    vmov.16 q0[7], r0
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_b3:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q3, q0
+; CHECKBE-NEXT:    vrev64.16 q0, q1
+; CHECKBE-NEXT:    vmov.u16 r0, q3[1]
+; CHECKBE-NEXT:    vmov.16 q2[0], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q0[0]
+; CHECKBE-NEXT:    vmov.16 q2[1], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q3[3]
+; CHECKBE-NEXT:    vmov.16 q2[2], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q0[2]
+; CHECKBE-NEXT:    vmov.16 q2[3], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q3[5]
+; CHECKBE-NEXT:    vmov.16 q2[4], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q0[4]
+; CHECKBE-NEXT:    vmov.16 q2[5], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q3[7]
+; CHECKBE-NEXT:    vmov.16 q2[6], r0
+; CHECKBE-NEXT:    vmov.u16 r0, q0[6]
+; CHECKBE-NEXT:    vmov.16 q2[7], r0
+; CHECKBE-NEXT:    vrev64.16 q0, q2
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 1, i32 8, i32 3, i32 10, i32 5, i32 12, i32 7, i32 14>
   ret <8 x i16> %out
@@ -289,6 +493,14 @@ define arm_aapcs_vfpcc <8 x i16> @vmovn16_b4(<8 x i16> %src1, <8 x i16> %src2) {
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    vmovnb.i32 q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn16_b4:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.16 q2, q1
+; CHECKBE-NEXT:    vrev64.16 q1, q0
+; CHECKBE-NEXT:    vmovnb.i32 q1, q2
+; CHECKBE-NEXT:    vrev64.16 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <8 x i16> %src1, <8 x i16> %src2, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
   ret <8 x i16> %out
@@ -300,6 +512,14 @@ define arm_aapcs_vfpcc <16 x i8> @vmovn8_b1(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    vmovnt.i16 q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn8_b1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q2, q1
+; CHECKBE-NEXT:    vrev64.8 q1, q0
+; CHECKBE-NEXT:    vmovnt.i16 q1, q2
+; CHECKBE-NEXT:    vrev64.8 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
   ret <16 x i8> %out
@@ -311,6 +531,14 @@ define arm_aapcs_vfpcc <16 x i8> @vmovn8_b2(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-NEXT:    vmovnt.i16 q1, q0
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn8_b2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q2, q0
+; CHECKBE-NEXT:    vrev64.8 q3, q1
+; CHECKBE-NEXT:    vmovnt.i16 q3, q2
+; CHECKBE-NEXT:    vrev64.8 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 16, i32 0, i32 18, i32 2, i32 20, i32 4, i32 22, i32 6, i32 24, i32 8, i32 26, i32 10, i32 28, i32 12, i32 30, i32 14>
   ret <16 x i8> %out
@@ -322,6 +550,14 @@ define arm_aapcs_vfpcc <16 x i8> @vmovn8_t1(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-NEXT:    vmovnb.i16 q1, q0
 ; CHECK-NEXT:    vmov q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn8_t1:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q2, q0
+; CHECKBE-NEXT:    vrev64.8 q3, q1
+; CHECKBE-NEXT:    vmovnb.i16 q3, q2
+; CHECKBE-NEXT:    vrev64.8 q0, q3
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
   ret <16 x i8> %out
@@ -364,6 +600,45 @@ define arm_aapcs_vfpcc <16 x i8> @vmovn8_t2(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-NEXT:    vmov.u8 r0, q2[14]
 ; CHECK-NEXT:    vmov.8 q0[15], r0
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn8_t2:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q2, q1
+; CHECKBE-NEXT:    vrev64.8 q3, q0
+; CHECKBE-NEXT:    vmov.u8 r0, q2[1]
+; CHECKBE-NEXT:    vmov.8 q1[0], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[0]
+; CHECKBE-NEXT:    vmov.8 q1[1], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q2[3]
+; CHECKBE-NEXT:    vmov.8 q1[2], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[2]
+; CHECKBE-NEXT:    vmov.8 q1[3], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q2[5]
+; CHECKBE-NEXT:    vmov.8 q1[4], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[4]
+; CHECKBE-NEXT:    vmov.8 q1[5], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q2[7]
+; CHECKBE-NEXT:    vmov.8 q1[6], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[6]
+; CHECKBE-NEXT:    vmov.8 q1[7], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q2[9]
+; CHECKBE-NEXT:    vmov.8 q1[8], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[8]
+; CHECKBE-NEXT:    vmov.8 q1[9], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q2[11]
+; CHECKBE-NEXT:    vmov.8 q1[10], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[10]
+; CHECKBE-NEXT:    vmov.8 q1[11], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q2[13]
+; CHECKBE-NEXT:    vmov.8 q1[12], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[12]
+; CHECKBE-NEXT:    vmov.8 q1[13], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q2[15]
+; CHECKBE-NEXT:    vmov.8 q1[14], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[14]
+; CHECKBE-NEXT:    vmov.8 q1[15], r0
+; CHECKBE-NEXT:    vrev64.8 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 17, i32 0, i32 19, i32 2, i32 21, i32 4, i32 23, i32 6, i32 25, i32 8, i32 27, i32 10, i32 29, i32 12, i32 31, i32 14>
   ret <16 x i8> %out
@@ -406,6 +681,45 @@ define arm_aapcs_vfpcc <16 x i8> @vmovn8_t3(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK-NEXT:    vmov.u8 r0, q1[14]
 ; CHECK-NEXT:    vmov.8 q0[15], r0
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn8_t3:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q3, q0
+; CHECKBE-NEXT:    vrev64.8 q0, q1
+; CHECKBE-NEXT:    vmov.u8 r0, q3[1]
+; CHECKBE-NEXT:    vmov.8 q2[0], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q0[0]
+; CHECKBE-NEXT:    vmov.8 q2[1], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[3]
+; CHECKBE-NEXT:    vmov.8 q2[2], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q0[2]
+; CHECKBE-NEXT:    vmov.8 q2[3], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[5]
+; CHECKBE-NEXT:    vmov.8 q2[4], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q0[4]
+; CHECKBE-NEXT:    vmov.8 q2[5], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[7]
+; CHECKBE-NEXT:    vmov.8 q2[6], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q0[6]
+; CHECKBE-NEXT:    vmov.8 q2[7], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[9]
+; CHECKBE-NEXT:    vmov.8 q2[8], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q0[8]
+; CHECKBE-NEXT:    vmov.8 q2[9], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[11]
+; CHECKBE-NEXT:    vmov.8 q2[10], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q0[10]
+; CHECKBE-NEXT:    vmov.8 q2[11], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[13]
+; CHECKBE-NEXT:    vmov.8 q2[12], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q0[12]
+; CHECKBE-NEXT:    vmov.8 q2[13], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q3[15]
+; CHECKBE-NEXT:    vmov.8 q2[14], r0
+; CHECKBE-NEXT:    vmov.u8 r0, q0[14]
+; CHECKBE-NEXT:    vmov.8 q2[15], r0
+; CHECKBE-NEXT:    vrev64.8 q0, q2
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 1, i32 16, i32 3, i32 18, i32 5, i32 20, i32 7, i32 22, i32 9, i32 24, i32 11, i32 26, i32 13, i32 28, i32 15, i32 30>
   ret <16 x i8> %out
@@ -416,6 +730,14 @@ define arm_aapcs_vfpcc <16 x i8> @vmovn8_t4(<16 x i8> %src1, <16 x i8> %src2) {
 ; CHECK:       @ %bb.0: @ %entry
 ; CHECK-NEXT:    vmovnb.i16 q0, q1
 ; CHECK-NEXT:    bx lr
+;
+; CHECKBE-LABEL: vmovn8_t4:
+; CHECKBE:       @ %bb.0: @ %entry
+; CHECKBE-NEXT:    vrev64.8 q2, q1
+; CHECKBE-NEXT:    vrev64.8 q1, q0
+; CHECKBE-NEXT:    vmovnb.i16 q1, q2
+; CHECKBE-NEXT:    vrev64.8 q0, q1
+; CHECKBE-NEXT:    bx lr
 entry:
   %out = shufflevector <16 x i8> %src1, <16 x i8> %src2, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
   ret <16 x i8> %out
-- 
2.7.4