From d2725878ee60829a1214f3689b4178d608b512eb Mon Sep 17 00:00:00 2001
From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Wed, 13 May 2020 14:31:18 +0100
Subject: [PATCH] [InstCombine] Add vector tests for the
 or(shl(zext(x),32)|zext(y)) concat combines

---
 test/Transforms/InstCombine/or-concat.ll | 112 +++++++++++++++++++++++
 1 file changed, 112 insertions(+)

diff --git a/test/Transforms/InstCombine/or-concat.ll b/test/Transforms/InstCombine/or-concat.ll
index 4148e4900f7..4bbe5896a7f 100644
--- a/test/Transforms/InstCombine/or-concat.ll
+++ b/test/Transforms/InstCombine/or-concat.ll
@@ -28,6 +28,23 @@ define i64 @concat_bswap32_unary_split(i64 %a0) {
   ret i64 %9
 }
 
+define <2 x i64> @concat_bswap32_unary_split_vector(<2 x i64> %a0) {
+; CHECK-LABEL: @concat_bswap32_unary_split_vector(
+; CHECK-NEXT:    [[TMP1:%.*]] = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> [[A0:%.*]])
+; CHECK-NEXT:    ret <2 x i64> [[TMP1]]
+;
+  %1 = lshr <2 x i64> %a0, <i64 32, i64 32>
+  %2 = trunc <2 x i64> %1 to <2 x i32>
+  %3 = trunc <2 x i64> %a0 to <2 x i32>
+  %4 = tail call <2 x i32> @llvm.bswap.v2i32(<2 x i32> %2)
+  %5 = tail call <2 x i32> @llvm.bswap.v2i32(<2 x i32> %3)
+  %6 = zext <2 x i32> %4 to <2 x i64>
+  %7 = zext <2 x i32> %5 to <2 x i64>
+  %8 = shl nuw <2 x i64> %7, <i64 32, i64 32>
+  %9 = or <2 x i64> %6, %8
+  ret <2 x i64> %9
+}
+
 define i64 @concat_bswap32_unary_flip(i64 %a0) {
 ; CHECK-LABEL: @concat_bswap32_unary_flip(
 ; CHECK-NEXT:    [[TMP1:%.*]] = lshr i64 [[A0:%.*]], 32
@@ -48,6 +65,26 @@ define i64 @concat_bswap32_unary_flip(i64 %a0) {
   ret i64 %9
 }
 
+define <2 x i64> @concat_bswap32_unary_flip_vector(<2 x i64> %a0) {
+; CHECK-LABEL: @concat_bswap32_unary_flip_vector(
+; CHECK-NEXT:    [[TMP1:%.*]] = lshr <2 x i64> [[A0:%.*]], <i64 32, i64 32>
+; CHECK-NEXT:    [[TMP2:%.*]] = shl <2 x i64> [[A0]], <i64 32, i64 32>
+; CHECK-NEXT:    [[TMP3:%.*]] = or <2 x i64> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[TMP4:%.*]] = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> [[TMP3]])
+; CHECK-NEXT:    ret <2 x i64> [[TMP4]]
+;
+  %1 = lshr <2 x i64> %a0, <i64 32, i64 32>
+  %2 = trunc <2 x i64> %1 to <2 x i32>
+  %3 = trunc <2 x i64> %a0 to <2 x i32>
+  %4 = tail call <2 x i32> @llvm.bswap.v2i32(<2 x i32> %2)
+  %5 = tail call <2 x i32> @llvm.bswap.v2i32(<2 x i32> %3)
+  %6 = zext <2 x i32> %4 to <2 x i64>
+  %7 = zext <2 x i32> %5 to <2 x i64>
+  %8 = shl nuw <2 x i64> %6, <i64 32, i64 32>
+  %9 = or <2 x i64> %7, %8
+  ret <2 x i64> %9
+}
+
 define i64 @concat_bswap32_binary(i32 %a0, i32 %a1) {
 ; CHECK-LABEL: @concat_bswap32_binary(
 ; CHECK-NEXT:    [[TMP1:%.*]] = zext i32 [[A1:%.*]] to i64
@@ -66,7 +103,26 @@ define i64 @concat_bswap32_binary(i32 %a0, i32 %a1) {
   ret i64 %6
 }
 
+define <2 x i64> @concat_bswap32_binary_vector(<2 x i32> %a0, <2 x i32> %a1) {
+; CHECK-LABEL: @concat_bswap32_binary_vector(
+; CHECK-NEXT:    [[TMP1:%.*]] = zext <2 x i32> [[A1:%.*]] to <2 x i64>
+; CHECK-NEXT:    [[TMP2:%.*]] = zext <2 x i32> [[A0:%.*]] to <2 x i64>
+; CHECK-NEXT:    [[TMP3:%.*]] = shl nuw <2 x i64> [[TMP2]], <i64 32, i64 32>
+; CHECK-NEXT:    [[TMP4:%.*]] = or <2 x i64> [[TMP3]], [[TMP1]]
+; CHECK-NEXT:    [[TMP5:%.*]] = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> [[TMP4]])
+; CHECK-NEXT:    ret <2 x i64> [[TMP5]]
+;
+  %1 = tail call <2 x i32> @llvm.bswap.v2i32(<2 x i32> %a0)
+  %2 = tail call <2 x i32> @llvm.bswap.v2i32(<2 x i32> %a1)
+  %3 = zext <2 x i32> %1 to <2 x i64>
+  %4 = zext <2 x i32> %2 to <2 x i64>
+  %5 = shl nuw <2 x i64> %4, <i64 32, i64 32>
+  %6 = or <2 x i64> %3, %5
+  ret <2 x i64> %6
+}
+
 declare i32 @llvm.bswap.i32(i32)
+declare <2 x i32> @llvm.bswap.v2i32(<2 x i32>)
 
 ; BITREVERSE
 
@@ -87,6 +143,23 @@ define i64 @concat_bitreverse32_unary_split(i64 %a0) {
   ret i64 %9
 }
 
+define <2 x i64> @concat_bitreverse32_unary_split_vector(<2 x i64> %a0) {
+; CHECK-LABEL: @concat_bitreverse32_unary_split_vector(
+; CHECK-NEXT:    [[TMP1:%.*]] = call <2 x i64> @llvm.bitreverse.v2i64(<2 x i64> [[A0:%.*]])
+; CHECK-NEXT:    ret <2 x i64> [[TMP1]]
+;
+  %1 = lshr <2 x i64> %a0, <i64 32, i64 32>
+  %2 = trunc <2 x i64> %1 to <2 x i32>
+  %3 = trunc <2 x i64> %a0 to <2 x i32>
+  %4 = tail call <2 x i32> @llvm.bitreverse.v2i32(<2 x i32> %2)
+  %5 = tail call <2 x i32> @llvm.bitreverse.v2i32(<2 x i32> %3)
+  %6 = zext <2 x i32> %4 to <2 x i64>
+  %7 = zext <2 x i32> %5 to <2 x i64>
+  %8 = shl nuw <2 x i64> %7, <i64 32, i64 32>
+  %9 = or <2 x i64> %6, %8
+  ret <2 x i64> %9
+}
+
 define i64 @concat_bitreverse32_unary_flip(i64 %a0) {
 ; CHECK-LABEL: @concat_bitreverse32_unary_flip(
 ; CHECK-NEXT:    [[TMP1:%.*]] = lshr i64 [[A0:%.*]], 32
@@ -107,6 +180,26 @@ define i64 @concat_bitreverse32_unary_flip(i64 %a0) {
   ret i64 %9
 }
 
+define <2 x i64> @concat_bitreverse32_unary_flip_vector(<2 x i64> %a0) {
+; CHECK-LABEL: @concat_bitreverse32_unary_flip_vector(
+; CHECK-NEXT:    [[TMP1:%.*]] = lshr <2 x i64> [[A0:%.*]], <i64 32, i64 32>
+; CHECK-NEXT:    [[TMP2:%.*]] = shl <2 x i64> [[A0]], <i64 32, i64 32>
+; CHECK-NEXT:    [[TMP3:%.*]] = or <2 x i64> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[TMP4:%.*]] = call <2 x i64> @llvm.bitreverse.v2i64(<2 x i64> [[TMP3]])
+; CHECK-NEXT:    ret <2 x i64> [[TMP4]]
+;
+  %1 = lshr <2 x i64> %a0, <i64 32, i64 32>
+  %2 = trunc <2 x i64> %1 to <2 x i32>
+  %3 = trunc <2 x i64> %a0 to <2 x i32>
+  %4 = tail call <2 x i32> @llvm.bitreverse.v2i32(<2 x i32> %2)
+  %5 = tail call <2 x i32> @llvm.bitreverse.v2i32(<2 x i32> %3)
+  %6 = zext <2 x i32> %4 to <2 x i64>
+  %7 = zext <2 x i32> %5 to <2 x i64>
+  %8 = shl nuw <2 x i64> %6, <i64 32, i64 32>
+  %9 = or <2 x i64> %7, %8
+  ret <2 x i64> %9
+}
+
 define i64 @concat_bitreverse32_binary(i32 %a0, i32 %a1) {
 ; CHECK-LABEL: @concat_bitreverse32_binary(
 ; CHECK-NEXT:    [[TMP1:%.*]] = zext i32 [[A1:%.*]] to i64
@@ -125,4 +218,23 @@ define i64 @concat_bitreverse32_binary(i32 %a0, i32 %a1) {
   ret i64 %6
 }
 
+define <2 x i64> @concat_bitreverse32_binary_vector(<2 x i32> %a0, <2 x i32> %a1) {
+; CHECK-LABEL: @concat_bitreverse32_binary_vector(
+; CHECK-NEXT:    [[TMP1:%.*]] = zext <2 x i32> [[A1:%.*]] to <2 x i64>
+; CHECK-NEXT:    [[TMP2:%.*]] = zext <2 x i32> [[A0:%.*]] to <2 x i64>
+; CHECK-NEXT:    [[TMP3:%.*]] = shl nuw <2 x i64> [[TMP2]], <i64 32, i64 32>
+; CHECK-NEXT:    [[TMP4:%.*]] = or <2 x i64> [[TMP3]], [[TMP1]]
+; CHECK-NEXT:    [[TMP5:%.*]] = call <2 x i64> @llvm.bitreverse.v2i64(<2 x i64> [[TMP4]])
+; CHECK-NEXT:    ret <2 x i64> [[TMP5]]
+;
+  %1 = tail call <2 x i32> @llvm.bitreverse.v2i32(<2 x i32> %a0)
+  %2 = tail call <2 x i32> @llvm.bitreverse.v2i32(<2 x i32> %a1)
+  %3 = zext <2 x i32> %1 to <2 x i64>
+  %4 = zext <2 x i32> %2 to <2 x i64>
+  %5 = shl nuw <2 x i64> %4, <i64 32, i64 32>
+  %6 = or <2 x i64> %3, %5
+  ret <2 x i64> %6
+}
+
 declare i32 @llvm.bitreverse.i32(i32)
+declare <2 x i32> @llvm.bitreverse.v2i32(<2 x i32>)