From: Amaury Sechet <deadalnix@gmail.com>
Date: Tue, 8 Oct 2019 16:16:26 +0000 (+0000)
Subject: (Re)generate various tests. NFC
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=265327e7cf0d070bcc30155adaa97871e3eaa656;p=llvm

(Re)generate various tests. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@374074 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/test/CodeGen/AArch64/arm64-rev.ll b/test/CodeGen/AArch64/arm64-rev.ll
index a04fe05137b..8ceb60432e5 100644
--- a/test/CodeGen/AArch64/arm64-rev.ll
+++ b/test/CodeGen/AArch64/arm64-rev.ll
@@ -8,10 +8,11 @@ define i32 @test_rev_w(i32 %a) nounwind {
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    rev w0, w0
 ; CHECK-NEXT:    ret
-; GISEL-LABEL: test_rev_w:
-; GISEL:       // %bb.0: // %entry
-; GISEL-NEXT:    rev w0, w0
-; GISEL-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev_w:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    rev w0, w0
+; FALLBACK-NEXT:    ret
 entry:
   %0 = tail call i32 @llvm.bswap.i32(i32 %a)
   ret i32 %0
@@ -23,10 +24,11 @@ define i64 @test_rev_x(i64 %a) nounwind {
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    rev x0, x0
 ; CHECK-NEXT:    ret
-; GISEL-LABEL: test_rev_x:
-; GISEL:       // %bb.0: // %entry
-; GISEL-NEXT:    rev x0, x0
-; GISEL-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev_x:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    rev x0, x0
+; FALLBACK-NEXT:    ret
 entry:
   %0 = tail call i64 @llvm.bswap.i64(i64 %a)
   ret i64 %0
@@ -40,6 +42,13 @@ define i32 @test_rev_w_srl16(i16 %a) {
 ; CHECK-NEXT:    and w8, w0, #0xffff
 ; CHECK-NEXT:    rev16 w0, w8
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev_w_srl16:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    and w8, w0, #0xffff
+; FALLBACK-NEXT:    rev w8, w8
+; FALLBACK-NEXT:    lsr w0, w8, #16
+; FALLBACK-NEXT:    ret
 entry:
   %0 = zext i16 %a to i32
   %1 = tail call i32 @llvm.bswap.i32(i32 %0)
@@ -53,6 +62,13 @@ define i32 @test_rev_w_srl16_load(i16 *%a) {
 ; CHECK-NEXT:    ldrh w8, [x0]
 ; CHECK-NEXT:    rev16 w0, w8
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev_w_srl16_load:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    ldrh w8, [x0]
+; FALLBACK-NEXT:    rev w8, w8
+; FALLBACK-NEXT:    lsr w0, w8, #16
+; FALLBACK-NEXT:    ret
 entry:
   %0 = load i16, i16 *%a
   %1 = zext i16 %0 to i32
@@ -68,6 +84,14 @@ define i32 @test_rev_w_srl16_add(i8 %a, i8 %b) {
 ; CHECK-NEXT:    add w8, w8, w1, uxtb
 ; CHECK-NEXT:    rev16 w0, w8
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev_w_srl16_add:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    and w8, w1, #0xff
+; FALLBACK-NEXT:    add w8, w8, w0, uxtb
+; FALLBACK-NEXT:    rev w8, w8
+; FALLBACK-NEXT:    lsr w0, w8, #16
+; FALLBACK-NEXT:    ret
 entry:
   %0 = zext i8 %a to i32
   %1 = zext i8 %b to i32
@@ -85,6 +109,14 @@ define i64 @test_rev_x_srl32(i32 %a) {
 ; CHECK-NEXT:    mov w8, w0
 ; CHECK-NEXT:    rev32 x0, x8
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev_x_srl32:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    // kill: def $w0 killed $w0 def $x0
+; FALLBACK-NEXT:    ubfx x8, x0, #0, #32
+; FALLBACK-NEXT:    rev x8, x8
+; FALLBACK-NEXT:    lsr x0, x8, #32
+; FALLBACK-NEXT:    ret
 entry:
   %0 = zext i32 %a to i64
   %1 = tail call i64 @llvm.bswap.i64(i64 %0)
@@ -98,6 +130,13 @@ define i64 @test_rev_x_srl32_load(i32 *%a) {
 ; CHECK-NEXT:    ldr w8, [x0]
 ; CHECK-NEXT:    rev32 x0, x8
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev_x_srl32_load:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    ldr w8, [x0]
+; FALLBACK-NEXT:    rev x8, x8
+; FALLBACK-NEXT:    lsr x0, x8, #32
+; FALLBACK-NEXT:    ret
 entry:
   %0 = load i32, i32 *%a
   %1 = zext i32 %0 to i64
@@ -112,6 +151,14 @@ define i64 @test_rev_x_srl32_shift(i64 %a) {
 ; CHECK-NEXT:    ubfx x8, x0, #2, #29
 ; CHECK-NEXT:    rev32 x0, x8
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev_x_srl32_shift:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    lsl x8, x0, #33
+; FALLBACK-NEXT:    lsr x8, x8, #35
+; FALLBACK-NEXT:    rev x8, x8
+; FALLBACK-NEXT:    lsr x0, x8, #32
+; FALLBACK-NEXT:    ret
 entry:
   %0 = shl i64 %a, 33
   %1 = lshr i64 %0, 35
@@ -128,6 +175,19 @@ define i32 @test_rev16_w(i32 %X) nounwind {
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    rev16 w0, w0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev16_w:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    lsr w8, w0, #8
+; FALLBACK-NEXT:    lsl w9, w0, #8
+; FALLBACK-NEXT:    and w10, w8, #0xff0000
+; FALLBACK-NEXT:    and w11, w9, #0xff000000
+; FALLBACK-NEXT:    and w9, w9, #0xff00
+; FALLBACK-NEXT:    orr w10, w11, w10
+; FALLBACK-NEXT:    and w8, w8, #0xff
+; FALLBACK-NEXT:    orr w9, w10, w9
+; FALLBACK-NEXT:    orr w0, w9, w8
+; FALLBACK-NEXT:    ret
 entry:
   %tmp1 = lshr i32 %X, 8
   %X15 = bitcast i32 %X to i32
@@ -151,6 +211,13 @@ define i64 @test_rev16_x(i64 %a) nounwind {
 ; CHECK-NEXT:    rev x8, x0
 ; CHECK-NEXT:    ror x0, x8, #16
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev16_x:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    rev x8, x0
+; FALLBACK-NEXT:    lsl x9, x8, #48
+; FALLBACK-NEXT:    orr x0, x9, x8, lsr #16
+; FALLBACK-NEXT:    ret
 entry:
   %0 = tail call i64 @llvm.bswap.i64(i64 %a)
   %1 = lshr i64 %0, 16
@@ -164,6 +231,13 @@ define i64 @test_rev32_x(i64 %a) nounwind {
 ; CHECK:       // %bb.0: // %entry
 ; CHECK-NEXT:    rev32 x0, x0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_rev32_x:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    rev x8, x0
+; FALLBACK-NEXT:    lsl x9, x8, #32
+; FALLBACK-NEXT:    orr x0, x9, x8, lsr #32
+; FALLBACK-NEXT:    ret
 entry:
   %0 = tail call i64 @llvm.bswap.i64(i64 %a)
   %1 = lshr i64 %0, 32
@@ -178,6 +252,12 @@ define <8 x i8> @test_vrev64D8(<8 x i8>* %A) nounwind {
 ; CHECK-NEXT:    ldr d0, [x0]
 ; CHECK-NEXT:    rev64.8b v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64D8:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr d0, [x0]
+; FALLBACK-NEXT:    rev64.8b v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <8 x i8>, <8 x i8>* %A
 	%tmp2 = shufflevector <8 x i8> %tmp1, <8 x i8> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 	ret <8 x i8> %tmp2
@@ -189,6 +269,12 @@ define <4 x i16> @test_vrev64D16(<4 x i16>* %A) nounwind {
 ; CHECK-NEXT:    ldr d0, [x0]
 ; CHECK-NEXT:    rev64.4h v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64D16:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr d0, [x0]
+; FALLBACK-NEXT:    rev64.4h v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <4 x i16>, <4 x i16>* %A
 	%tmp2 = shufflevector <4 x i16> %tmp1, <4 x i16> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
 	ret <4 x i16> %tmp2
@@ -200,6 +286,17 @@ define <2 x i32> @test_vrev64D32(<2 x i32>* %A) nounwind {
 ; CHECK-NEXT:    ldr d0, [x0]
 ; CHECK-NEXT:    rev64.2s v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64D32:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr d0, [x0]
+; FALLBACK-NEXT:    adrp x8, .LCPI13_0
+; FALLBACK-NEXT:    ldr d1, [x8, :lo12:.LCPI13_0]
+; FALLBACK-NEXT:    mov.s v2[1], w8
+; FALLBACK-NEXT:    mov.d v0[1], v2[0]
+; FALLBACK-NEXT:    tbl.16b v0, { v0 }, v1
+; FALLBACK-NEXT:    // kill: def $d0 killed $d0 killed $q0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <2 x i32>, <2 x i32>* %A
 	%tmp2 = shufflevector <2 x i32> %tmp1, <2 x i32> undef, <2 x i32> <i32 1, i32 0>
 	ret <2 x i32> %tmp2
@@ -211,6 +308,17 @@ define <2 x float> @test_vrev64Df(<2 x float>* %A) nounwind {
 ; CHECK-NEXT:    ldr d0, [x0]
 ; CHECK-NEXT:    rev64.2s v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64Df:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr d0, [x0]
+; FALLBACK-NEXT:    adrp x8, .LCPI14_0
+; FALLBACK-NEXT:    ldr d1, [x8, :lo12:.LCPI14_0]
+; FALLBACK-NEXT:    mov.s v2[1], w8
+; FALLBACK-NEXT:    mov.d v0[1], v2[0]
+; FALLBACK-NEXT:    tbl.16b v0, { v0 }, v1
+; FALLBACK-NEXT:    // kill: def $d0 killed $d0 killed $q0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <2 x float>, <2 x float>* %A
 	%tmp2 = shufflevector <2 x float> %tmp1, <2 x float> undef, <2 x i32> <i32 1, i32 0>
 	ret <2 x float> %tmp2
@@ -222,6 +330,12 @@ define <16 x i8> @test_vrev64Q8(<16 x i8>* %A) nounwind {
 ; CHECK-NEXT:    ldr q0, [x0]
 ; CHECK-NEXT:    rev64.16b v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64Q8:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    rev64.16b v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <16 x i8>, <16 x i8>* %A
 	%tmp2 = shufflevector <16 x i8> %tmp1, <16 x i8> undef, <16 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8>
 	ret <16 x i8> %tmp2
@@ -233,6 +347,12 @@ define <8 x i16> @test_vrev64Q16(<8 x i16>* %A) nounwind {
 ; CHECK-NEXT:    ldr q0, [x0]
 ; CHECK-NEXT:    rev64.8h v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64Q16:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    rev64.8h v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <8 x i16>, <8 x i16>* %A
 	%tmp2 = shufflevector <8 x i16> %tmp1, <8 x i16> undef, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
 	ret <8 x i16> %tmp2
@@ -244,6 +364,14 @@ define <4 x i32> @test_vrev64Q32(<4 x i32>* %A) nounwind {
 ; CHECK-NEXT:    ldr q0, [x0]
 ; CHECK-NEXT:    rev64.4s v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64Q32:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    adrp x8, .LCPI17_0
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    ldr q2, [x8, :lo12:.LCPI17_0]
+; FALLBACK-NEXT:    tbl.16b v0, { v0, v1 }, v2
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <4 x i32>, <4 x i32>* %A
 	%tmp2 = shufflevector <4 x i32> %tmp1, <4 x i32> undef, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
 	ret <4 x i32> %tmp2
@@ -255,6 +383,14 @@ define <4 x float> @test_vrev64Qf(<4 x float>* %A) nounwind {
 ; CHECK-NEXT:    ldr q0, [x0]
 ; CHECK-NEXT:    rev64.4s v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64Qf:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    adrp x8, .LCPI18_0
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    ldr q2, [x8, :lo12:.LCPI18_0]
+; FALLBACK-NEXT:    tbl.16b v0, { v0, v1 }, v2
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <4 x float>, <4 x float>* %A
 	%tmp2 = shufflevector <4 x float> %tmp1, <4 x float> undef, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
 	ret <4 x float> %tmp2
@@ -266,6 +402,12 @@ define <8 x i8> @test_vrev32D8(<8 x i8>* %A) nounwind {
 ; CHECK-NEXT:    ldr d0, [x0]
 ; CHECK-NEXT:    rev32.8b v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev32D8:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr d0, [x0]
+; FALLBACK-NEXT:    rev32.8b v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <8 x i8>, <8 x i8>* %A
 	%tmp2 = shufflevector <8 x i8> %tmp1, <8 x i8> undef, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
 	ret <8 x i8> %tmp2
@@ -277,6 +419,12 @@ define <4 x i16> @test_vrev32D16(<4 x i16>* %A) nounwind {
 ; CHECK-NEXT:    ldr d0, [x0]
 ; CHECK-NEXT:    rev32.4h v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev32D16:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr d0, [x0]
+; FALLBACK-NEXT:    rev32.4h v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <4 x i16>, <4 x i16>* %A
 	%tmp2 = shufflevector <4 x i16> %tmp1, <4 x i16> undef, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
 	ret <4 x i16> %tmp2
@@ -288,6 +436,12 @@ define <16 x i8> @test_vrev32Q8(<16 x i8>* %A) nounwind {
 ; CHECK-NEXT:    ldr q0, [x0]
 ; CHECK-NEXT:    rev32.16b v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev32Q8:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    rev32.16b v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <16 x i8>, <16 x i8>* %A
 	%tmp2 = shufflevector <16 x i8> %tmp1, <16 x i8> undef, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4, i32 11, i32 10, i32 9, i32 8, i32 15, i32 14, i32 13, i32 12>
 	ret <16 x i8> %tmp2
@@ -299,6 +453,12 @@ define <8 x i16> @test_vrev32Q16(<8 x i16>* %A) nounwind {
 ; CHECK-NEXT:    ldr q0, [x0]
 ; CHECK-NEXT:    rev32.8h v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev32Q16:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    rev32.8h v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <8 x i16>, <8 x i16>* %A
 	%tmp2 = shufflevector <8 x i16> %tmp1, <8 x i16> undef, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
 	ret <8 x i16> %tmp2
@@ -310,6 +470,12 @@ define <8 x i8> @test_vrev16D8(<8 x i8>* %A) nounwind {
 ; CHECK-NEXT:    ldr d0, [x0]
 ; CHECK-NEXT:    rev16.8b v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev16D8:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr d0, [x0]
+; FALLBACK-NEXT:    rev16.8b v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <8 x i8>, <8 x i8>* %A
 	%tmp2 = shufflevector <8 x i8> %tmp1, <8 x i8> undef, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
 	ret <8 x i8> %tmp2
@@ -321,6 +487,12 @@ define <16 x i8> @test_vrev16Q8(<16 x i8>* %A) nounwind {
 ; CHECK-NEXT:    ldr q0, [x0]
 ; CHECK-NEXT:    rev16.16b v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev16Q8:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    rev16.16b v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <16 x i8>, <16 x i8>* %A
 	%tmp2 = shufflevector <16 x i8> %tmp1, <16 x i8> undef, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
 	ret <16 x i8> %tmp2
@@ -334,6 +506,12 @@ define <8 x i8> @test_vrev64D8_undef(<8 x i8>* %A) nounwind {
 ; CHECK-NEXT:    ldr d0, [x0]
 ; CHECK-NEXT:    rev64.8b v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64D8_undef:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr d0, [x0]
+; FALLBACK-NEXT:    rev64.8b v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <8 x i8>, <8 x i8>* %A
 	%tmp2 = shufflevector <8 x i8> %tmp1, <8 x i8> undef, <8 x i32> <i32 7, i32 undef, i32 undef, i32 4, i32 3, i32 2, i32 1, i32 0>
 	ret <8 x i8> %tmp2
@@ -345,6 +523,12 @@ define <8 x i16> @test_vrev32Q16_undef(<8 x i16>* %A) nounwind {
 ; CHECK-NEXT:    ldr q0, [x0]
 ; CHECK-NEXT:    rev32.8h v0, v0
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev32Q16_undef:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    rev32.8h v0, v0
+; FALLBACK-NEXT:    ret
 	%tmp1 = load <8 x i16>, <8 x i16>* %A
 	%tmp2 = shufflevector <8 x i16> %tmp1, <8 x i16> undef, <8 x i32> <i32 undef, i32 0, i32 undef, i32 2, i32 5, i32 4, i32 7, i32 undef>
 	ret <8 x i16> %tmp2
@@ -359,6 +543,14 @@ define void @test_vrev64(<4 x i16>* nocapture %source, <2 x i16>* nocapture %dst
 ; CHECK-NEXT:    st1.h { v0 }[5], [x8]
 ; CHECK-NEXT:    st1.h { v0 }[6], [x1]
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev64:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    ldr q0, [x0]
+; FALLBACK-NEXT:    add x8, x1, #2 // =2
+; FALLBACK-NEXT:    st1.h { v0 }[5], [x8]
+; FALLBACK-NEXT:    st1.h { v0 }[6], [x1]
+; FALLBACK-NEXT:    ret
 entry:
   %0 = bitcast <4 x i16>* %source to <8 x i16>*
   %tmp2 = load <8 x i16>, <8 x i16>* %0, align 4
@@ -381,6 +573,19 @@ define void @float_vrev64(float* nocapture %source, <4 x float>* nocapture %dest
 ; CHECK-NEXT:    rev64.4s v0, v0
 ; CHECK-NEXT:    str q0, [x1, #176]
 ; CHECK-NEXT:    ret
+;
+; FALLBACK-LABEL: float_vrev64:
+; FALLBACK:       // %bb.0: // %entry
+; FALLBACK-NEXT:    fmov s0, wzr
+; FALLBACK-NEXT:    mov.s v0[1], v0[0]
+; FALLBACK-NEXT:    mov.s v0[2], v0[0]
+; FALLBACK-NEXT:    adrp x8, .LCPI28_0
+; FALLBACK-NEXT:    mov.s v0[3], v0[0]
+; FALLBACK-NEXT:    ldr q1, [x0]
+; FALLBACK-NEXT:    ldr q2, [x8, :lo12:.LCPI28_0]
+; FALLBACK-NEXT:    tbl.16b v0, { v0, v1 }, v2
+; FALLBACK-NEXT:    str q0, [x1, #176]
+; FALLBACK-NEXT:    ret
 entry:
   %0 = bitcast float* %source to <4 x float>*
   %tmp2 = load <4 x float>, <4 x float>* %0, align 4
@@ -396,10 +601,11 @@ define <4 x i32> @test_vrev32_bswap(<4 x i32> %source) nounwind {
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    rev32.16b v0, v0
 ; CHECK-NEXT:    ret
-; GISEL-LABEL: test_vrev32_bswap:
-; GISEL:       // %bb.0:
-; GISEL-NEXT:    rev32.16b v0, v0
-; GISEL-NEXT:    ret
+;
+; FALLBACK-LABEL: test_vrev32_bswap:
+; FALLBACK:       // %bb.0:
+; FALLBACK-NEXT:    rev32.16b v0, v0
+; FALLBACK-NEXT:    ret
   %bswap = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %source)
   ret <4 x i32> %bswap
 }
diff --git a/test/CodeGen/AMDGPU/lshr.v2i16.ll b/test/CodeGen/AMDGPU/lshr.v2i16.ll
index e88550fb92c..5275a819080 100644
--- a/test/CodeGen/AMDGPU/lshr.v2i16.ll
+++ b/test/CodeGen/AMDGPU/lshr.v2i16.ll
@@ -1,46 +1,134 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX9 %s
 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,VI,CIVI %s
 ; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,CI,CIVI %s
 
-; GCN-LABEL: {{^}}s_lshr_v2i16:
-; GFX9: s_load_dword [[LHS:s[0-9]+]]
-; GFX9: s_load_dword [[RHS:s[0-9]+]]
-; GFX9: v_mov_b32_e32 [[VLHS:v[0-9]+]], [[LHS]]
-; GFX9: v_pk_lshrrev_b16 [[RESULT:v[0-9]+]], [[RHS]], [[VLHS]]
-
-; CIVI: s_load_dword [[LHS:s[0-9]+]]
-; CIVI: s_load_dword [[RHS:s[0-9]+]]
-; CIVI: s_lshr_b32 s{{[0-9]+}}, s{{[0-9]+}}, 16
-; CIVI: s_lshr_b32 s{{[0-9]+}}, s{{[0-9]+}}, 16
-; CIVI: s_lshr_b32 s{{[0-9]+}}, s{{[0-9]+}}, s{{[0-9]+}}
-; CIVI-DAG: v_bfe_u32 v{{[0-9]+}}, s{{[0-9]+}}, v{{[0-9]+}}, 16
-; CIVI-DAG: s_lshl_b32
-; CIVI: v_or_b32_e32
 define amdgpu_kernel void @s_lshr_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> %lhs, <2 x i16> %rhs) #0 {
+; GFX9-LABEL: s_lshr_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x24
+; GFX9-NEXT:    s_load_dword s4, s[0:1], 0x2c
+; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x30
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NEXT:    v_mov_b32_e32 v2, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_pk_lshrrev_b16 v2, s0, v2
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: s_lshr_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x24
+; VI-NEXT:    s_load_dword s5, s[0:1], 0x2c
+; VI-NEXT:    s_load_dword s0, s[0:1], 0x30
+; VI-NEXT:    s_mov_b32 s4, 0xffff
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_and_b32 s1, s5, s4
+; VI-NEXT:    s_and_b32 s4, s0, s4
+; VI-NEXT:    s_lshr_b32 s5, s5, 16
+; VI-NEXT:    s_lshr_b32 s0, s0, 16
+; VI-NEXT:    s_lshr_b32 s0, s5, s0
+; VI-NEXT:    v_mov_b32_e32 v0, s4
+; VI-NEXT:    v_bfe_u32 v0, s1, v0, 16
+; VI-NEXT:    s_lshl_b32 s0, s0, 16
+; VI-NEXT:    v_or_b32_e32 v2, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v0, s2
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: s_lshr_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
+; CI-NEXT:    s_load_dword s2, s[0:1], 0xb
+; CI-NEXT:    s_load_dword s0, s[0:1], 0xc
+; CI-NEXT:    s_mov_b32 s3, 0xffff
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, -1
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_lshr_b32 s1, s2, 16
+; CI-NEXT:    s_lshr_b32 s8, s0, 16
+; CI-NEXT:    s_and_b32 s0, s0, s3
+; CI-NEXT:    v_mov_b32_e32 v0, s0
+; CI-NEXT:    s_lshr_b32 s0, s1, s8
+; CI-NEXT:    s_and_b32 s2, s2, s3
+; CI-NEXT:    v_bfe_u32 v0, s2, v0, 16
+; CI-NEXT:    s_lshl_b32 s0, s0, 16
+; CI-NEXT:    v_or_b32_e32 v0, s0, v0
+; CI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; CI-NEXT:    s_endpgm
   %result = lshr <2 x i16> %lhs, %rhs
   store <2 x i16> %result, <2 x i16> addrspace(1)* %out
   ret void
 }
 
-; GCN-LABEL: {{^}}v_lshr_v2i16:
-; GCN: {{buffer|flat|global}}_load_dword [[LHS:v[0-9]+]]
-; GCN: {{buffer|flat|global}}_load_dword [[RHS:v[0-9]+]]
-; GFX9: v_pk_lshrrev_b16 [[RESULT:v[0-9]+]], [[RHS]], [[LHS]]
-
-; VI: v_lshrrev_b16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; VI: v_lshrrev_b16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
-; VI: v_or_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-
-; CI: s_mov_b32 [[MASK:s[0-9]+]], 0xffff{{$}}
-; CI-DAG: v_lshrrev_b32_e32 v{{[0-9]+}}, 16, [[LHS]]
-; CI-DAG: v_lshrrev_b32_e32 v{{[0-9]+}}, 16, [[RHS]]
-; CI: v_and_b32_e32 v{{[0-9]+}}, [[MASK]], v{{[0-9]+}}
-; CI: v_and_b32_e32 v{{[0-9]+}}, [[MASK]], v{{[0-9]+}}
-; CI: v_bfe_u32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}, 16
-; CI: v_lshrrev_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; CI: v_lshlrev_b32_e32 v{{[0-9]+}}, 16, v{{[0-9]+}}
-; CI: v_or_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
 define amdgpu_kernel void @v_lshr_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: v_lshr_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    global_load_dword v4, v[0:1], off offset:4
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshrrev_b16 v2, v4, v3
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: v_lshr_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v4, 2, v0
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v4
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    v_add_u32_e32 v2, vcc, 4, v0
+; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v5, v[0:1]
+; VI-NEXT:    flat_load_dword v2, v[2:3]
+; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v4
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshrrev_b16_e32 v3, v2, v5
+; VI-NEXT:    v_lshrrev_b16_sdwa v2, v2, v5 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-NEXT:    v_or_b32_e32 v2, v3, v2
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: v_lshr_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    buffer_load_dword v3, v[0:1], s[4:7], 0 addr64 offset:4
+; CI-NEXT:    s_mov_b32 s8, 0xffff
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(1)
+; CI-NEXT:    v_lshrrev_b32_e32 v4, 16, v2
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v5, 16, v3
+; CI-NEXT:    v_and_b32_e32 v2, s8, v2
+; CI-NEXT:    v_and_b32_e32 v3, s8, v3
+; CI-NEXT:    v_bfe_u32 v2, v2, v3, 16
+; CI-NEXT:    v_lshrrev_b32_e32 v3, v5, v4
+; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT:    v_or_b32_e32 v2, v2, v3
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -53,11 +141,71 @@ define amdgpu_kernel void @v_lshr_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16>
   ret void
 }
 
-; GCN-LABEL: {{^}}lshr_v_s_v2i16:
-; GFX9: s_load_dword [[RHS:s[0-9]+]]
-; GFX9: {{buffer|flat|global}}_load_dword [[LHS:v[0-9]+]]
-; GFX9: v_pk_lshrrev_b16 [[RESULT:v[0-9]+]], [[RHS]], [[LHS]]
 define amdgpu_kernel void @lshr_v_s_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in, <2 x i16> %sgpr) #0 {
+; GFX9-LABEL: lshr_v_s_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x34
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s7
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s6, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s4, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshrrev_b16 v2, s0, v3
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: lshr_v_s_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    s_load_dword s0, s[0:1], 0x34
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s7
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s6, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v3, v[0:1]
+; VI-NEXT:    s_lshr_b32 s1, s0, 16
+; VI-NEXT:    v_mov_b32_e32 v4, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshrrev_b16_e32 v2, s0, v3
+; VI-NEXT:    v_lshrrev_b16_sdwa v3, v4, v3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-NEXT:    v_or_b32_e32 v2, v2, v3
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: lshr_v_s_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
+; CI-NEXT:    s_load_dword s8, s[0:1], 0xd
+; CI-NEXT:    s_mov_b32 s3, 0xf000
+; CI-NEXT:    s_mov_b32 s2, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[0:1], s[6:7]
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_lshr_b32 s9, s8, 16
+; CI-NEXT:    s_mov_b32 s10, 0xffff
+; CI-NEXT:    s_and_b32 s8, s8, s10
+; CI-NEXT:    s_mov_b64 s[6:7], s[2:3]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v3, 16, v2
+; CI-NEXT:    v_and_b32_e32 v2, s10, v2
+; CI-NEXT:    v_lshrrev_b32_e32 v3, s9, v3
+; CI-NEXT:    v_bfe_u32 v2, v2, s8, 16
+; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT:    v_or_b32_e32 v2, v2, v3
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -68,11 +216,71 @@ define amdgpu_kernel void @lshr_v_s_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16
   ret void
 }
 
-; GCN-LABEL: {{^}}lshr_s_v_v2i16:
-; GFX9: s_load_dword [[LHS:s[0-9]+]]
-; GFX9: {{buffer|flat|global}}_load_dword [[RHS:v[0-9]+]]
-; GFX9: v_pk_lshrrev_b16 [[RESULT:v[0-9]+]], [[RHS]], [[LHS]]
 define amdgpu_kernel void @lshr_s_v_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in, <2 x i16> %sgpr) #0 {
+; GFX9-LABEL: lshr_s_v_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x34
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s7
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s6, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s4, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshrrev_b16 v2, v3, s0
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: lshr_s_v_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    s_load_dword s0, s[0:1], 0x34
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s7
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s6, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v3, v[0:1]
+; VI-NEXT:    s_lshr_b32 s1, s0, 16
+; VI-NEXT:    v_mov_b32_e32 v4, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshrrev_b16_e64 v2, v3, s0
+; VI-NEXT:    v_lshrrev_b16_sdwa v3, v3, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
+; VI-NEXT:    v_or_b32_e32 v2, v2, v3
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: lshr_s_v_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
+; CI-NEXT:    s_load_dword s8, s[0:1], 0xd
+; CI-NEXT:    s_mov_b32 s3, 0xf000
+; CI-NEXT:    s_mov_b32 s2, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[0:1], s[6:7]
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_lshr_b32 s9, s8, 16
+; CI-NEXT:    s_mov_b32 s10, 0xffff
+; CI-NEXT:    s_and_b32 s8, s8, s10
+; CI-NEXT:    s_mov_b64 s[6:7], s[2:3]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v3, 16, v2
+; CI-NEXT:    v_and_b32_e32 v2, s10, v2
+; CI-NEXT:    v_lshr_b32_e32 v3, s9, v3
+; CI-NEXT:    v_bfe_u32 v2, s8, v2, 16
+; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT:    v_or_b32_e32 v2, v2, v3
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -83,10 +291,64 @@ define amdgpu_kernel void @lshr_s_v_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16
   ret void
 }
 
-; GCN-LABEL: {{^}}lshr_imm_v_v2i16:
-; GCN: {{buffer|flat|global}}_load_dword [[RHS:v[0-9]+]]
-; GFX9: v_pk_lshrrev_b16 [[RESULT:v[0-9]+]], [[RHS]], 8
 define amdgpu_kernel void @lshr_imm_v_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: lshr_imm_v_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshrrev_b16 v2, v3, 8 op_sel_hi:[1,0]
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: lshr_imm_v_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    v_mov_b32_e32 v3, 8
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v4, v[0:1]
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshrrev_b16_e64 v2, v4, 8
+; VI-NEXT:    v_lshrrev_b16_sdwa v3, v4, v3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
+; VI-NEXT:    v_or_b32_e32 v2, v2, v3
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: lshr_imm_v_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v3, 16, v2
+; CI-NEXT:    v_and_b32_e32 v2, 0xffff, v2
+; CI-NEXT:    v_lshr_b32_e32 v3, 8, v3
+; CI-NEXT:    v_bfe_u32 v2, 8, v2, 16
+; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT:    v_or_b32_e32 v2, v2, v3
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -97,10 +359,59 @@ define amdgpu_kernel void @lshr_imm_v_v2i16(<2 x i16> addrspace(1)* %out, <2 x i
   ret void
 }
 
-; GCN-LABEL: {{^}}lshr_v_imm_v2i16:
-; GCN: {{buffer|flat|global}}_load_dword [[LHS:v[0-9]+]]
-; GFX9: v_pk_lshrrev_b16 [[RESULT:v[0-9]+]], 8, [[LHS]]
 define amdgpu_kernel void @lshr_v_imm_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: lshr_v_imm_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshrrev_b16 v2, 8, v3 op_sel_hi:[0,1]
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: lshr_v_imm_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v3, v[0:1]
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshrrev_b32_e32 v2, 24, v3
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 16, v2
+; VI-NEXT:    v_or_b32_sdwa v2, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:DWORD
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: lshr_v_imm_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v2, 8, v2
+; CI-NEXT:    v_and_b32_e32 v2, 0xff00ff, v2
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -111,13 +422,84 @@ define amdgpu_kernel void @lshr_v_imm_v2i16(<2 x i16> addrspace(1)* %out, <2 x i
   ret void
 }
 
-; GCN-LABEL: {{^}}v_lshr_v4i16:
-; GCN: {{buffer|flat|global}}_load_dwordx2
-; GCN: {{buffer|flat|global}}_load_dwordx2
-; GFX9: v_pk_lshrrev_b16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; GFX9: v_pk_lshrrev_b16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; GCN: {{buffer|flat|global}}_store_dwordx2
 define amdgpu_kernel void @v_lshr_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: v_lshr_v4i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v4
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dwordx2 v[2:3], v[0:1], off
+; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off offset:8
+; GFX9-NEXT:    v_mov_b32_e32 v5, s1
+; GFX9-NEXT:    v_add_co_u32_e32 v4, vcc, s0, v4
+; GFX9-NEXT:    v_addc_co_u32_e32 v5, vcc, 0, v5, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshrrev_b16 v1, v1, v3
+; GFX9-NEXT:    v_pk_lshrrev_b16 v0, v0, v2
+; GFX9-NEXT:    global_store_dwordx2 v[4:5], v[0:1], off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: v_lshr_v4i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v4
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    v_add_u32_e32 v2, vcc, 8, v0
+; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dwordx2 v[0:1], v[0:1]
+; VI-NEXT:    flat_load_dwordx2 v[2:3], v[2:3]
+; VI-NEXT:    v_mov_b32_e32 v5, s1
+; VI-NEXT:    v_add_u32_e32 v4, vcc, s0, v4
+; VI-NEXT:    v_addc_u32_e32 v5, vcc, 0, v5, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshrrev_b16_e32 v6, v3, v1
+; VI-NEXT:    v_lshrrev_b16_sdwa v1, v3, v1 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-NEXT:    v_lshrrev_b16_e32 v3, v2, v0
+; VI-NEXT:    v_lshrrev_b16_sdwa v0, v2, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-NEXT:    v_or_b32_e32 v1, v6, v1
+; VI-NEXT:    v_or_b32_e32 v0, v3, v0
+; VI-NEXT:    flat_store_dwordx2 v[4:5], v[0:1]
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: v_lshr_v4i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dwordx2 v[2:3], v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    buffer_load_dwordx2 v[4:5], v[0:1], s[4:7], 0 addr64 offset:8
+; CI-NEXT:    s_mov_b32 s8, 0xffff
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(1)
+; CI-NEXT:    v_lshrrev_b32_e32 v6, 16, v2
+; CI-NEXT:    v_lshrrev_b32_e32 v7, 16, v3
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v8, 16, v4
+; CI-NEXT:    v_lshrrev_b32_e32 v9, 16, v5
+; CI-NEXT:    v_and_b32_e32 v2, s8, v2
+; CI-NEXT:    v_and_b32_e32 v4, s8, v4
+; CI-NEXT:    v_and_b32_e32 v3, s8, v3
+; CI-NEXT:    v_and_b32_e32 v5, s8, v5
+; CI-NEXT:    v_bfe_u32 v3, v3, v5, 16
+; CI-NEXT:    v_lshrrev_b32_e32 v5, v9, v7
+; CI-NEXT:    v_bfe_u32 v2, v2, v4, 16
+; CI-NEXT:    v_lshrrev_b32_e32 v4, v8, v6
+; CI-NEXT:    v_lshlrev_b32_e32 v5, 16, v5
+; CI-NEXT:    v_lshlrev_b32_e32 v4, 16, v4
+; CI-NEXT:    v_or_b32_e32 v3, v3, v5
+; CI-NEXT:    v_or_b32_e32 v2, v2, v4
+; CI-NEXT:    buffer_store_dwordx2 v[2:3], v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <4 x i16>, <4 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -130,12 +512,66 @@ define amdgpu_kernel void @v_lshr_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16>
   ret void
 }
 
-; GCN-LABEL: {{^}}lshr_v_imm_v4i16:
-; GCN: {{buffer|flat|global}}_load_dwordx2
-; GFX9: v_pk_lshrrev_b16 v{{[0-9]+}}, 8, v{{[0-9]+}}
-; GFX9: v_pk_lshrrev_b16 v{{[0-9]+}}, 8, v{{[0-9]+}}
-; GCN: {{buffer|flat|global}}_store_dwordx2
 define amdgpu_kernel void @lshr_v_imm_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: lshr_v_imm_v4i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
+; GFX9-NEXT:    v_mov_b32_e32 v3, s1
+; GFX9-NEXT:    v_add_co_u32_e32 v2, vcc, s0, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v3, vcc, 0, v3, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshrrev_b16 v1, 8, v1 op_sel_hi:[0,1]
+; GFX9-NEXT:    v_pk_lshrrev_b16 v0, 8, v0 op_sel_hi:[0,1]
+; GFX9-NEXT:    global_store_dwordx2 v[2:3], v[0:1], off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: lshr_v_imm_v4i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dwordx2 v[0:1], v[0:1]
+; VI-NEXT:    v_mov_b32_e32 v3, s1
+; VI-NEXT:    v_add_u32_e32 v2, vcc, s0, v2
+; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v3, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshrrev_b32_e32 v4, 24, v1
+; VI-NEXT:    v_lshrrev_b32_e32 v5, 24, v0
+; VI-NEXT:    v_lshlrev_b32_e32 v4, 16, v4
+; VI-NEXT:    v_lshlrev_b32_e32 v5, 16, v5
+; VI-NEXT:    v_or_b32_sdwa v1, v1, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:DWORD
+; VI-NEXT:    v_or_b32_sdwa v0, v0, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:DWORD
+; VI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: lshr_v_imm_v4i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dwordx2 v[2:3], v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_mov_b32 s8, 0xff00ff
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v3, 8, v3
+; CI-NEXT:    v_lshrrev_b32_e32 v2, 8, v2
+; CI-NEXT:    v_and_b32_e32 v3, s8, v3
+; CI-NEXT:    v_and_b32_e32 v2, s8, v2
+; CI-NEXT:    buffer_store_dwordx2 v[2:3], v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <4 x i16>, <4 x i16> addrspace(1)* %in, i64 %tid.ext
diff --git a/test/CodeGen/AMDGPU/shl.v2i16.ll b/test/CodeGen/AMDGPU/shl.v2i16.ll
index b116e2ee6a2..c6816d351f2 100644
--- a/test/CodeGen/AMDGPU/shl.v2i16.ll
+++ b/test/CodeGen/AMDGPU/shl.v2i16.ll
@@ -1,60 +1,135 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX9 %s
 ; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,VI,CIVI %s
 ; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,CI,CIVI %s
 
-; GCN-LABEL: {{^}}s_shl_v2i16:
-; GFX9: s_load_dword [[LHS:s[0-9]+]]
-; GFX9: s_load_dword [[RHS:s[0-9]+]]
-; GFX9: v_mov_b32_e32 [[VLHS:v[0-9]+]], [[LHS]]
-; GFX9: v_pk_lshlrev_b16 [[RESULT:v[0-9]+]], [[RHS]], [[VLHS]]
-
-; VI: s_load_dword s
-; VI: s_load_dword s
-; VI: s_lshr_b32
-; VI: s_lshr_b32
-; VI: s_and_b32
-; VI: s_and_b32
-; VI: s_lshl_b32
-; VI: s_lshl_b32
-; VI: s_lshl_b32
-; VI: s_and_b32
-; VI: s_or_b32
-
-; CI: s_load_dword s
-; CI: s_load_dword s
-; CI: s_lshr_b32
-; CI: s_and_b32
-; CI: s_lshr_b32
-; CI: s_lshl_b32
-; CI: s_lshl_b32
-; CI: s_lshl_b32
-; CI: s_and_b32
-; CI: s_or_b32
-; CI: _store_dword
 define amdgpu_kernel void @s_shl_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> %lhs, <2 x i16> %rhs) #0 {
+; GFX9-LABEL: s_shl_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
+; GFX9-NEXT:    s_load_dword s2, s[0:1], 0x2c
+; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x30
+; GFX9-NEXT:    s_mov_b32 s7, 0xf000
+; GFX9-NEXT:    s_mov_b32 s6, -1
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NEXT:    v_pk_lshlrev_b16 v0, s0, v0
+; GFX9-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: s_shl_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
+; VI-NEXT:    s_load_dword s2, s[0:1], 0x2c
+; VI-NEXT:    s_load_dword s0, s[0:1], 0x30
+; VI-NEXT:    s_mov_b32 s3, 0xffff
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_lshr_b32 s1, s2, 16
+; VI-NEXT:    s_lshr_b32 s8, s0, 16
+; VI-NEXT:    s_and_b32 s2, s2, s3
+; VI-NEXT:    s_and_b32 s0, s0, s3
+; VI-NEXT:    s_lshl_b32 s0, s2, s0
+; VI-NEXT:    s_lshl_b32 s1, s1, s8
+; VI-NEXT:    s_lshl_b32 s1, s1, 16
+; VI-NEXT:    s_and_b32 s0, s0, s3
+; VI-NEXT:    s_or_b32 s0, s0, s1
+; VI-NEXT:    v_mov_b32_e32 v0, s0
+; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: s_shl_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
+; CI-NEXT:    s_load_dword s2, s[0:1], 0xb
+; CI-NEXT:    s_load_dword s0, s[0:1], 0xc
+; CI-NEXT:    s_mov_b32 s3, 0xffff
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, -1
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_lshr_b32 s1, s2, 16
+; CI-NEXT:    s_and_b32 s8, s0, s3
+; CI-NEXT:    s_lshr_b32 s0, s0, 16
+; CI-NEXT:    s_lshl_b32 s0, s1, s0
+; CI-NEXT:    s_lshl_b32 s1, s2, s8
+; CI-NEXT:    s_lshl_b32 s0, s0, 16
+; CI-NEXT:    s_and_b32 s1, s1, s3
+; CI-NEXT:    s_or_b32 s0, s1, s0
+; CI-NEXT:    v_mov_b32_e32 v0, s0
+; CI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; CI-NEXT:    s_endpgm
   %result = shl <2 x i16> %lhs, %rhs
   store <2 x i16> %result, <2 x i16> addrspace(1)* %out
   ret void
 }
 
-; GCN-LABEL: {{^}}v_shl_v2i16:
-; GCN: {{buffer|flat|global}}_load_dword [[LHS:v[0-9]+]]
-; GCN: {{buffer|flat|global}}_load_dword [[RHS:v[0-9]+]]
-; GFX9: v_pk_lshlrev_b16 [[RESULT:v[0-9]+]], [[RHS]], [[LHS]]
-
-; VI: v_lshlrev_b16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; VI: v_lshlrev_b16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
-; VI: v_or_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-
-; CI: s_mov_b32 [[MASK:s[0-9]+]], 0xffff{{$}}
-; CI: v_lshrrev_b32_e32 v{{[0-9]+}}, 16, [[LHS]]
-; CI: v_lshrrev_b32_e32 v{{[0-9]+}}, 16, v{{[0-9]+}}
-; CI: v_lshlrev_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; CI: v_lshlrev_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; CI: v_lshlrev_b32_e32 v{{[0-9]+}}, 16, v{{[0-9]+}}
-; CI: v_and_b32_e32 v{{[0-9]+}}, [[MASK]], v{{[0-9]+}}
-; CI: v_or_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
 define amdgpu_kernel void @v_shl_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: v_shl_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    global_load_dword v4, v[0:1], off offset:4
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshlrev_b16 v2, v4, v3
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: v_shl_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v4, 2, v0
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v4
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    v_add_u32_e32 v2, vcc, 4, v0
+; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v5, v[0:1]
+; VI-NEXT:    flat_load_dword v2, v[2:3]
+; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v4
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshlrev_b16_e32 v3, v2, v5
+; VI-NEXT:    v_lshlrev_b16_sdwa v2, v2, v5 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-NEXT:    v_or_b32_e32 v2, v3, v2
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: v_shl_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    buffer_load_dword v3, v[0:1], s[4:7], 0 addr64 offset:4
+; CI-NEXT:    s_mov_b32 s8, 0xffff
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(1)
+; CI-NEXT:    v_lshrrev_b32_e32 v4, 16, v2
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_and_b32_e32 v5, s8, v3
+; CI-NEXT:    v_lshrrev_b32_e32 v3, 16, v3
+; CI-NEXT:    v_lshlrev_b32_e32 v3, v3, v4
+; CI-NEXT:    v_lshlrev_b32_e32 v2, v5, v2
+; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT:    v_and_b32_e32 v2, s8, v2
+; CI-NEXT:    v_or_b32_e32 v2, v2, v3
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -67,11 +142,71 @@ define amdgpu_kernel void @v_shl_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> a
   ret void
 }
 
-; GCN-LABEL: {{^}}shl_v_s_v2i16:
-; GFX9: s_load_dword [[RHS:s[0-9]+]]
-; GFX9: {{buffer|flat|global}}_load_dword [[LHS:v[0-9]+]]
-; GFX9: v_pk_lshlrev_b16 [[RESULT:v[0-9]+]], [[RHS]], [[LHS]]
 define amdgpu_kernel void @shl_v_s_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in, <2 x i16> %sgpr) #0 {
+; GFX9-LABEL: shl_v_s_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x34
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s7
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s6, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s4, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshlrev_b16 v2, s0, v3
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: shl_v_s_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    s_load_dword s0, s[0:1], 0x34
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s7
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s6, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v3, v[0:1]
+; VI-NEXT:    s_lshr_b32 s1, s0, 16
+; VI-NEXT:    v_mov_b32_e32 v4, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshlrev_b16_e32 v2, s0, v3
+; VI-NEXT:    v_lshlrev_b16_sdwa v3, v4, v3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-NEXT:    v_or_b32_e32 v2, v2, v3
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: shl_v_s_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
+; CI-NEXT:    s_load_dword s8, s[0:1], 0xd
+; CI-NEXT:    s_mov_b32 s3, 0xf000
+; CI-NEXT:    s_mov_b32 s2, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[0:1], s[6:7]
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_mov_b32 s9, 0xffff
+; CI-NEXT:    s_lshr_b32 s10, s8, 16
+; CI-NEXT:    s_and_b32 s8, s8, s9
+; CI-NEXT:    s_mov_b64 s[6:7], s[2:3]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v3, 16, v2
+; CI-NEXT:    v_lshlrev_b32_e32 v2, s8, v2
+; CI-NEXT:    v_lshlrev_b32_e32 v3, s10, v3
+; CI-NEXT:    v_and_b32_e32 v2, s9, v2
+; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT:    v_or_b32_e32 v2, v2, v3
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -82,11 +217,71 @@ define amdgpu_kernel void @shl_v_s_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16>
   ret void
 }
 
-; GCN-LABEL: {{^}}shl_s_v_v2i16:
-; GFX9: s_load_dword [[LHS:s[0-9]+]]
-; GFX9: {{buffer|flat|global}}_load_dword [[RHS:v[0-9]+]]
-; GFX9: v_pk_lshlrev_b16 [[RESULT:v[0-9]+]], [[RHS]], [[LHS]]
 define amdgpu_kernel void @shl_s_v_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in, <2 x i16> %sgpr) #0 {
+; GFX9-LABEL: shl_s_v_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x34
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s7
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s6, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s4, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshlrev_b16 v2, v3, s0
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: shl_s_v_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    s_load_dword s0, s[0:1], 0x34
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s7
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s6, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v3, v[0:1]
+; VI-NEXT:    s_lshr_b32 s1, s0, 16
+; VI-NEXT:    v_mov_b32_e32 v4, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshlrev_b16_e64 v2, v3, s0
+; VI-NEXT:    v_lshlrev_b16_sdwa v3, v3, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
+; VI-NEXT:    v_or_b32_e32 v2, v2, v3
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: shl_s_v_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
+; CI-NEXT:    s_load_dword s8, s[0:1], 0xd
+; CI-NEXT:    s_mov_b32 s3, 0xf000
+; CI-NEXT:    s_mov_b32 s2, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[0:1], s[6:7]
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_mov_b32 s0, 0xffff
+; CI-NEXT:    s_lshr_b32 s1, s8, 16
+; CI-NEXT:    s_mov_b64 s[6:7], s[2:3]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_and_b32_e32 v3, s0, v2
+; CI-NEXT:    v_lshrrev_b32_e32 v2, 16, v2
+; CI-NEXT:    v_lshl_b32_e32 v2, s1, v2
+; CI-NEXT:    v_lshl_b32_e32 v3, s8, v3
+; CI-NEXT:    v_lshlrev_b32_e32 v2, 16, v2
+; CI-NEXT:    v_and_b32_e32 v3, s0, v3
+; CI-NEXT:    v_or_b32_e32 v2, v3, v2
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -97,10 +292,66 @@ define amdgpu_kernel void @shl_s_v_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16>
   ret void
 }
 
-; GCN-LABEL: {{^}}shl_imm_v_v2i16:
-; GCN: {{buffer|flat|global}}_load_dword [[RHS:v[0-9]+]]
-; GFX9: v_pk_lshlrev_b16 [[RESULT:v[0-9]+]], [[RHS]], 8
 define amdgpu_kernel void @shl_imm_v_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: shl_imm_v_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshlrev_b16 v2, v3, 8 op_sel_hi:[1,0]
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: shl_imm_v_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    v_mov_b32_e32 v3, 8
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v4, v[0:1]
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshlrev_b16_e64 v2, v4, 8
+; VI-NEXT:    v_lshlrev_b16_sdwa v3, v4, v3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
+; VI-NEXT:    v_or_b32_e32 v2, v2, v3
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: shl_imm_v_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_mov_b32 s4, 0xffff
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_and_b32_e32 v3, s4, v2
+; CI-NEXT:    v_lshrrev_b32_e32 v2, 16, v2
+; CI-NEXT:    v_lshl_b32_e32 v2, 8, v2
+; CI-NEXT:    v_lshl_b32_e32 v3, 8, v3
+; CI-NEXT:    v_lshlrev_b32_e32 v2, 16, v2
+; CI-NEXT:    v_and_b32_e32 v3, s4, v3
+; CI-NEXT:    v_or_b32_e32 v2, v3, v2
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -111,10 +362,60 @@ define amdgpu_kernel void @shl_imm_v_v2i16(<2 x i16> addrspace(1)* %out, <2 x i1
   ret void
 }
 
-; GCN-LABEL: {{^}}shl_v_imm_v2i16:
-; GCN: {{buffer|flat|global}}_load_dword [[LHS:v[0-9]+]]
-; GFX9: v_pk_lshlrev_b16 [[RESULT:v[0-9]+]], 8, [[LHS]]
 define amdgpu_kernel void @shl_v_imm_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: shl_v_imm_v2i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dword v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshlrev_b16 v2, 8, v3 op_sel_hi:[0,1]
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: shl_v_imm_v2i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dword v3, v[0:1]
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 8, v3
+; VI-NEXT:    v_and_b32_e32 v2, 0xff000000, v2
+; VI-NEXT:    v_lshlrev_b16_e32 v3, 8, v3
+; VI-NEXT:    v_or_b32_e32 v2, v3, v2
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: shl_v_imm_v2i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshlrev_b32_e32 v2, 8, v2
+; CI-NEXT:    v_and_b32_e32 v2, 0xff00ff00, v2
+; CI-NEXT:    buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -125,13 +426,84 @@ define amdgpu_kernel void @shl_v_imm_v2i16(<2 x i16> addrspace(1)* %out, <2 x i1
   ret void
 }
 
-; GCN-LABEL: {{^}}v_shl_v4i16:
-; GCN: {{buffer|flat|global}}_load_dwordx2
-; GCN: {{buffer|flat|global}}_load_dwordx2
-; GFX9: v_pk_lshlrev_b16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; GFX9: v_pk_lshlrev_b16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
-; GCN: {{buffer|flat|global}}_store_dwordx2
 define amdgpu_kernel void @v_shl_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: v_shl_v4i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v4
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dwordx2 v[2:3], v[0:1], off
+; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off offset:8
+; GFX9-NEXT:    v_mov_b32_e32 v5, s1
+; GFX9-NEXT:    v_add_co_u32_e32 v4, vcc, s0, v4
+; GFX9-NEXT:    v_addc_co_u32_e32 v5, vcc, 0, v5, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshlrev_b16 v1, v1, v3
+; GFX9-NEXT:    v_pk_lshlrev_b16 v0, v0, v2
+; GFX9-NEXT:    global_store_dwordx2 v[4:5], v[0:1], off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: v_shl_v4i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v4
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    v_add_u32_e32 v2, vcc, 8, v0
+; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dwordx2 v[0:1], v[0:1]
+; VI-NEXT:    flat_load_dwordx2 v[2:3], v[2:3]
+; VI-NEXT:    v_mov_b32_e32 v5, s1
+; VI-NEXT:    v_add_u32_e32 v4, vcc, s0, v4
+; VI-NEXT:    v_addc_u32_e32 v5, vcc, 0, v5, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshlrev_b16_e32 v6, v3, v1
+; VI-NEXT:    v_lshlrev_b16_sdwa v1, v3, v1 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-NEXT:    v_lshlrev_b16_e32 v3, v2, v0
+; VI-NEXT:    v_lshlrev_b16_sdwa v0, v2, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-NEXT:    v_or_b32_e32 v1, v6, v1
+; VI-NEXT:    v_or_b32_e32 v0, v3, v0
+; VI-NEXT:    flat_store_dwordx2 v[4:5], v[0:1]
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: v_shl_v4i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dwordx2 v[2:3], v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    buffer_load_dwordx2 v[4:5], v[0:1], s[4:7], 0 addr64 offset:8
+; CI-NEXT:    s_mov_b32 s8, 0xffff
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(1)
+; CI-NEXT:    v_lshrrev_b32_e32 v6, 16, v2
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_and_b32_e32 v8, s8, v4
+; CI-NEXT:    v_lshrrev_b32_e32 v4, 16, v4
+; CI-NEXT:    v_and_b32_e32 v9, s8, v5
+; CI-NEXT:    v_lshrrev_b32_e32 v7, 16, v3
+; CI-NEXT:    v_lshrrev_b32_e32 v5, 16, v5
+; CI-NEXT:    v_lshlrev_b32_e32 v5, v5, v7
+; CI-NEXT:    v_lshlrev_b32_e32 v3, v9, v3
+; CI-NEXT:    v_lshlrev_b32_e32 v4, v4, v6
+; CI-NEXT:    v_lshlrev_b32_e32 v2, v8, v2
+; CI-NEXT:    v_lshlrev_b32_e32 v5, 16, v5
+; CI-NEXT:    v_and_b32_e32 v3, s8, v3
+; CI-NEXT:    v_lshlrev_b32_e32 v4, 16, v4
+; CI-NEXT:    v_and_b32_e32 v2, s8, v2
+; CI-NEXT:    v_or_b32_e32 v3, v3, v5
+; CI-NEXT:    v_or_b32_e32 v2, v2, v4
+; CI-NEXT:    buffer_store_dwordx2 v[2:3], v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <4 x i16>, <4 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -144,12 +516,73 @@ define amdgpu_kernel void @v_shl_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16> a
   ret void
 }
 
-; GCN-LABEL: {{^}}shl_v_imm_v4i16:
-; GCN: {{buffer|flat|global}}_load_dwordx2
-; GFX9: v_pk_lshlrev_b16 v{{[0-9]+}}, 8, v{{[0-9]+}}
-; GFX9: v_pk_lshlrev_b16 v{{[0-9]+}}, 8, v{{[0-9]+}}
-; GCN: {{buffer|flat|global}}_store_dwordx2
 define amdgpu_kernel void @shl_v_imm_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16> addrspace(1)* %in) #0 {
+; GFX9-LABEL: shl_v_imm_v4i16:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
+; GFX9-NEXT:    v_mov_b32_e32 v3, s1
+; GFX9-NEXT:    v_add_co_u32_e32 v2, vcc, s0, v2
+; GFX9-NEXT:    v_addc_co_u32_e32 v3, vcc, 0, v3, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_lshlrev_b16 v1, 8, v1 op_sel_hi:[0,1]
+; GFX9-NEXT:    v_pk_lshlrev_b16 v0, 8, v0 op_sel_hi:[0,1]
+; GFX9-NEXT:    global_store_dwordx2 v[2:3], v[0:1], off
+; GFX9-NEXT:    s_endpgm
+;
+; VI-LABEL: shl_v_imm_v4i16:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
+; VI-NEXT:    s_mov_b32 s4, 0xff000000
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v1, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v2
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    flat_load_dwordx2 v[0:1], v[0:1]
+; VI-NEXT:    v_mov_b32_e32 v3, s1
+; VI-NEXT:    v_add_u32_e32 v2, vcc, s0, v2
+; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v3, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_lshlrev_b32_e32 v4, 8, v1
+; VI-NEXT:    v_lshlrev_b16_e32 v5, 8, v0
+; VI-NEXT:    v_lshlrev_b32_e32 v0, 8, v0
+; VI-NEXT:    v_and_b32_e32 v0, s4, v0
+; VI-NEXT:    v_lshlrev_b16_e32 v1, 8, v1
+; VI-NEXT:    v_and_b32_e32 v4, s4, v4
+; VI-NEXT:    v_or_b32_e32 v1, v1, v4
+; VI-NEXT:    v_or_b32_e32 v0, v5, v0
+; VI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
+; VI-NEXT:    s_endpgm
+;
+; CI-LABEL: shl_v_imm_v4i16:
+; CI:       ; %bb.0:
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s7, 0xf000
+; CI-NEXT:    s_mov_b32 s6, 0
+; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
+; CI-NEXT:    v_mov_b32_e32 v1, 0
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; CI-NEXT:    buffer_load_dwordx2 v[2:3], v[0:1], s[4:7], 0 addr64
+; CI-NEXT:    s_mov_b32 s8, 0xff00
+; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; CI-NEXT:    s_waitcnt vmcnt(0)
+; CI-NEXT:    v_lshrrev_b32_e32 v4, 8, v3
+; CI-NEXT:    v_lshlrev_b32_e32 v3, 8, v3
+; CI-NEXT:    v_and_b32_e32 v4, s8, v4
+; CI-NEXT:    v_lshlrev_b32_e32 v2, 8, v2
+; CI-NEXT:    v_and_b32_e32 v3, s8, v3
+; CI-NEXT:    v_lshlrev_b32_e32 v4, 16, v4
+; CI-NEXT:    v_or_b32_e32 v3, v3, v4
+; CI-NEXT:    v_and_b32_e32 v2, 0xff00ff00, v2
+; CI-NEXT:    buffer_store_dwordx2 v[2:3], v[0:1], s[0:3], 0 addr64
+; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <4 x i16>, <4 x i16> addrspace(1)* %in, i64 %tid.ext
diff --git a/test/CodeGen/ARM/rev.ll b/test/CodeGen/ARM/rev.ll
index a36526ff1fb..b97dbc844e0 100644
--- a/test/CodeGen/ARM/rev.ll
+++ b/test/CodeGen/ARM/rev.ll
@@ -1,8 +1,11 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=arm-eabi -mattr=+v6 %s -o - | FileCheck %s
 
 define i32 @test1(i32 %X) nounwind {
-; CHECK-LABEL: test1
-; CHECK: rev16 r0, r0
+; CHECK-LABEL: test1:
+; CHECK:       @ %bb.0:
+; CHECK-NEXT:    rev16 r0, r0
+; CHECK-NEXT:    bx lr
         %tmp1 = lshr i32 %X, 8
         %X15 = bitcast i32 %X to i32
         %tmp4 = shl i32 %X15, 8
@@ -17,8 +20,10 @@ define i32 @test1(i32 %X) nounwind {
 }
 
 define i32 @test2(i32 %X) nounwind {
-; CHECK-LABEL: test2
-; CHECK: revsh r0, r0
+; CHECK-LABEL: test2:
+; CHECK:       @ %bb.0:
+; CHECK-NEXT:    revsh r0, r0
+; CHECK-NEXT:    bx lr
         %tmp1 = lshr i32 %X, 8
         %tmp1.upgrd.1 = trunc i32 %tmp1 to i16
         %tmp3 = trunc i32 %X to i16
@@ -31,9 +36,11 @@ define i32 @test2(i32 %X) nounwind {
 
 ; rdar://9147637
 define i32 @test3(i16 zeroext %a) nounwind {
-entry:
 ; CHECK-LABEL: test3:
-; CHECK: revsh r0, r0
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    revsh r0, r0
+; CHECK-NEXT:    bx lr
+entry:
   %0 = tail call i16 @llvm.bswap.i16(i16 %a)
   %1 = sext i16 %0 to i32
   ret i32 %1
@@ -42,9 +49,11 @@ entry:
 declare i16 @llvm.bswap.i16(i16) nounwind readnone
 
 define i32 @test4(i16 zeroext %a) nounwind {
-entry:
 ; CHECK-LABEL: test4:
-; CHECK: revsh r0, r0
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    revsh r0, r0
+; CHECK-NEXT:    bx lr
+entry:
   %conv = zext i16 %a to i32
   %shr9 = lshr i16 %a, 8
   %conv2 = zext i16 %shr9 to i32
@@ -57,9 +66,11 @@ entry:
 
 ; rdar://9609059
 define i32 @test5(i32 %i) nounwind readnone {
+; CHECK-LABEL: test5:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    revsh r0, r0
+; CHECK-NEXT:    bx lr
 entry:
-; CHECK-LABEL: test5
-; CHECK: revsh r0, r0
   %shl = shl i32 %i, 24
   %shr = ashr exact i32 %shl, 16
   %shr23 = lshr i32 %i, 8
@@ -70,9 +81,11 @@ entry:
 
 ; rdar://9609108
 define i32 @test6(i32 %x) nounwind readnone {
+; CHECK-LABEL: test6:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    rev16 r0, r0
+; CHECK-NEXT:    bx lr
 entry:
-; CHECK-LABEL: test6
-; CHECK: rev16 r0, r0
   %and = shl i32 %x, 8
   %shl = and i32 %and, 65280
   %and2 = lshr i32 %x, 8
@@ -87,10 +100,12 @@ entry:
 
 ; rdar://9164521
 define i32 @test7(i32 %a) nounwind readnone {
+; CHECK-LABEL: test7:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    rev r0, r0
+; CHECK-NEXT:    lsr r0, r0, #16
+; CHECK-NEXT:    bx lr
 entry:
-; CHECK-LABEL: test7
-; CHECK: rev r0, r0
-; CHECK: lsr r0, r0, #16
   %and = lshr i32 %a, 8
   %shr3 = and i32 %and, 255
   %and2 = shl i32 %a, 8
@@ -100,9 +115,11 @@ entry:
 }
 
 define i32 @test8(i32 %a) nounwind readnone {
+; CHECK-LABEL: test8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    revsh r0, r0
+; CHECK-NEXT:    bx lr
 entry:
-; CHECK-LABEL: test8
-; CHECK: revsh r0, r0
   %and = lshr i32 %a, 8
   %shr4 = and i32 %and, 255
   %and2 = shl i32 %a, 8
@@ -114,9 +131,11 @@ entry:
 
 ; rdar://10750814
 define zeroext i16 @test9(i16 zeroext %v) nounwind readnone {
+; CHECK-LABEL: test9:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    rev16 r0, r0
+; CHECK-NEXT:    bx lr
 entry:
-; CHECK-LABEL: test9
-; CHECK: rev16 r0, r0
   %conv = zext i16 %v to i32
   %shr4 = lshr i32 %conv, 8
   %shl = shl nuw nsw i32 %conv, 8
diff --git a/test/CodeGen/Thumb/rev.ll b/test/CodeGen/Thumb/rev.ll
index 3e947022e60..9ac65bf5c8e 100644
--- a/test/CodeGen/Thumb/rev.ll
+++ b/test/CodeGen/Thumb/rev.ll
@@ -1,8 +1,11 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=thumb-eabi -mattr=+v6 %s -o - | FileCheck %s
 
 define i32 @test1(i32 %X) nounwind {
-; CHECK: test1
-; CHECK: rev16 r0, r0
+; CHECK-LABEL: test1:
+; CHECK:       @ %bb.0:
+; CHECK-NEXT:    rev16 r0, r0
+; CHECK-NEXT:    bx lr
         %tmp1 = lshr i32 %X, 8
         %X15 = bitcast i32 %X to i32
         %tmp4 = shl i32 %X15, 8
@@ -17,8 +20,10 @@ define i32 @test1(i32 %X) nounwind {
 }
 
 define i32 @test2(i32 %X) nounwind {
-; CHECK: test2
-; CHECK: revsh r0, r0
+; CHECK-LABEL: test2:
+; CHECK:       @ %bb.0:
+; CHECK-NEXT:    revsh r0, r0
+; CHECK-NEXT:    bx lr
         %tmp1 = lshr i32 %X, 8
         %tmp1.upgrd.1 = trunc i32 %tmp1 to i16
         %tmp3 = trunc i32 %X to i16
@@ -31,9 +36,11 @@ define i32 @test2(i32 %X) nounwind {
 
 ; rdar://9147637
 define i32 @test3(i16 zeroext %a) nounwind {
-entry:
 ; CHECK-LABEL: test3:
-; CHECK: revsh r0, r0
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    revsh r0, r0
+; CHECK-NEXT:    bx lr
+entry:
   %0 = tail call i16 @llvm.bswap.i16(i16 %a)
   %1 = sext i16 %0 to i32
   ret i32 %1
@@ -42,9 +49,11 @@ entry:
 declare i16 @llvm.bswap.i16(i16) nounwind readnone
 
 define i32 @test4(i16 zeroext %a) nounwind {
-entry:
 ; CHECK-LABEL: test4:
-; CHECK: revsh r0, r0
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    revsh r0, r0
+; CHECK-NEXT:    bx lr
+entry:
   %conv = zext i16 %a to i32
   %shr9 = lshr i16 %a, 8
   %conv2 = zext i16 %shr9 to i32