3 files changed, 258 insertions, 93 deletions
diff --git a/llvm/test/CodeGen/PowerPC/bittest.ll b/llvm/test/CodeGen/PowerPC/bittest.ll
new file mode 100644
index 0000000..cba56e3
--- /dev/null
+++ b/llvm/test/CodeGen/PowerPC/bittest.ll
@@ -0,0 +1,193 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -verify-machineinstrs < %s -O3 -mcpu=ppc -mtriple powerpc-ibm-aix \
+; RUN:     -ppc-asm-full-reg-names | FileCheck %s
+
+define i32 @foo(i32 noundef signext %x) {
+; CHECK-LABEL: foo:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    mflr r0
+; CHECK-NEXT:    stwu r1, -64(r1)
+; CHECK-NEXT:    stw r0, 72(r1)
+; CHECK-NEXT:    cmpwi r3, 8
+; CHECK-NEXT:    stw r31, 60(r1) # 4-byte Folded Spill
+; CHECK-NEXT:    mr r31, r3
+; CHECK-NEXT:    li r3, 0
+; CHECK-NEXT:    ble cr0, L..BB0_4
+; CHECK-NEXT:  # %bb.1: # %entry
+; CHECK-NEXT:    cmpwi r31, 11
+; CHECK-NEXT:    bge cr0, L..BB0_7
+; CHECK-NEXT:  # %bb.2: # %entry
+; CHECK-NEXT:    cmplwi r31, 9
+; CHECK-NEXT:    beq cr0, L..BB0_9
+; CHECK-NEXT:  # %bb.3: # %entry
+; CHECK-NEXT:    cmplwi r31, 10
+; CHECK-NEXT:    beq cr0, L..BB0_11
+; CHECK-NEXT:    b L..BB0_13
+; CHECK-NEXT:  L..BB0_4: # %entry
+; CHECK-NEXT:    cmplwi r31, 4
+; CHECK-NEXT:    beq cr0, L..BB0_12
+; CHECK-NEXT:  # %bb.5: # %entry
+; CHECK-NEXT:    cmplwi r31, 7
+; CHECK-NEXT:    beq cr0, L..BB0_11
+; CHECK-NEXT:  # %bb.6: # %entry
+; CHECK-NEXT:    cmplwi r31, 8
+; CHECK-NEXT:    beq cr0, L..BB0_10
+; CHECK-NEXT:    b L..BB0_13
+; CHECK-NEXT:  L..BB0_7: # %entry
+; CHECK-NEXT:    beq cr0, L..BB0_10
+; CHECK-NEXT:  # %bb.8: # %entry
+; CHECK-NEXT:    cmplwi r31, 12
+; CHECK-NEXT:    bne cr0, L..BB0_13
+; CHECK-NEXT:  L..BB0_9: # %sw.bb2
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    bl .foo3[PR]
+; CHECK-NEXT:    nop
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    b L..BB0_13
+; CHECK-NEXT:  L..BB0_10: # %sw.bb1
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    bl .foo2[PR]
+; CHECK-NEXT:    nop
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    b L..BB0_13
+; CHECK-NEXT:  L..BB0_11: # %sw.bb
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    bl .foo1[PR]
+; CHECK-NEXT:    nop
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    b L..BB0_13
+; CHECK-NEXT:  L..BB0_12: # %sw.bb3
+; CHECK-NEXT:    li r3, 4
+; CHECK-NEXT:    bl .foo4[PR]
+; CHECK-NEXT:    nop
+; CHECK-NEXT:    li r3, 4
+; CHECK-NEXT:  L..BB0_13: # %return
+; CHECK-NEXT:    lwz r31, 60(r1) # 4-byte Folded Reload
+; CHECK-NEXT:    addi r1, r1, 64
+; CHECK-NEXT:    lwz r0, 8(r1)
+; CHECK-NEXT:    mtlr r0
+; CHECK-NEXT:    blr
+entry:
+  switch i32 %x, label %return [
+  i32 7, label %sw.bb
+  i32 10, label %sw.bb
+  i32 8, label %sw.bb1
+  i32 11, label %sw.bb1
+  i32 9, label %sw.bb2
+  i32 12, label %sw.bb2
+  i32 4, label %sw.bb3
+  ]
+
+sw.bb:                                            ; preds = %entry, %entry
+  tail call void @foo1(i32 noundef signext %x)
+  br label %return
+
+sw.bb1:                                           ; preds = %entry, %entry
+  tail call void @foo2(i32 noundef signext %x)
+  br label %return
+
+sw.bb2:                                           ; preds = %entry, %entry
+  tail call void @foo3(i32 noundef signext %x)
+  br label %return
+
+sw.bb3:                                           ; preds = %entry
+  tail call void @foo4(i32 noundef signext 4)
+  br label %return
+
+return:                                           ; preds = %sw.bb, %sw.bb1, %sw.bb2, %sw.bb3, %entry
+  %retval.0 = phi i32 [ 0, %entry ], [ 4, %sw.bb3 ], [ %x, %sw.bb2 ], [ %x, %sw.bb1 ], [ %x, %sw.bb ]
+  ret i32 %retval.0
+}
+
+define i32 @goo(i32 noundef signext %x) {
+; CHECK-LABEL: goo:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    mflr r0
+; CHECK-NEXT:    stwu r1, -64(r1)
+; CHECK-NEXT:    stw r0, 72(r1)
+; CHECK-NEXT:    cmplwi r3, 12
+; CHECK-NEXT:    stw r31, 60(r1) # 4-byte Folded Spill
+; CHECK-NEXT:    mr r31, r3
+; CHECK-NEXT:    bgt cr0, L..BB1_7
+; CHECK-NEXT:  # %bb.1: # %entry
+; CHECK-NEXT:    li r3, 1
+; CHECK-NEXT:    slw r3, r3, r31
+; CHECK-NEXT:    andi. r4, r3, 5632
+; CHECK-NEXT:    bne cr0, L..BB1_4
+; CHECK-NEXT:  # %bb.2: # %entry
+; CHECK-NEXT:    andi. r3, r3, 2304
+; CHECK-NEXT:    beq cr0, L..BB1_5
+; CHECK-NEXT:  # %bb.3: # %sw.bb1
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    bl .foo2[PR]
+; CHECK-NEXT:    nop
+; CHECK-NEXT:    b L..BB1_9
+; CHECK-NEXT:  L..BB1_4: # %sw.bb2
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    bl .foo3[PR]
+; CHECK-NEXT:    nop
+; CHECK-NEXT:    b L..BB1_9
+; CHECK-NEXT:  L..BB1_5: # %entry
+; CHECK-NEXT:    cmplwi r31, 7
+; CHECK-NEXT:    bne cr0, L..BB1_7
+; CHECK-NEXT:  # %bb.6: # %sw.bb
+; CHECK-NEXT:    li r3, 7
+; CHECK-NEXT:    li r31, 7
+; CHECK-NEXT:    bl .foo1[PR]
+; CHECK-NEXT:    nop
+; CHECK-NEXT:    b L..BB1_9
+; CHECK-NEXT:  L..BB1_7: # %entry
+; CHECK-NEXT:    cmplwi r31, 4
+; CHECK-NEXT:    li r31, 0
+; CHECK-NEXT:    bne cr0, L..BB1_9
+; CHECK-NEXT:  # %bb.8: # %sw.bb3
+; CHECK-NEXT:    li r3, 4
+; CHECK-NEXT:    li r31, 4
+; CHECK-NEXT:    bl .foo4[PR]
+; CHECK-NEXT:    nop
+; CHECK-NEXT:  L..BB1_9: # %return
+; CHECK-NEXT:    mr r3, r31
+; CHECK-NEXT:    lwz r31, 60(r1) # 4-byte Folded Reload
+; CHECK-NEXT:    addi r1, r1, 64
+; CHECK-NEXT:    lwz r0, 8(r1)
+; CHECK-NEXT:    mtlr r0
+; CHECK-NEXT:    blr
+entry:
+  switch i32 %x, label %return [
+  i32 7, label %sw.bb
+  i32 8, label %sw.bb1
+  i32 11, label %sw.bb1
+  i32 9, label %sw.bb2
+  i32 10, label %sw.bb2
+  i32 12, label %sw.bb2
+  i32 4, label %sw.bb3
+  ]
+
+sw.bb:                                            ; preds = %entry
+  tail call void @foo1(i32 noundef signext 7)
+  br label %return
+
+sw.bb1:                                           ; preds = %entry, %entry
+  tail call void @foo2(i32 noundef signext %x)
+  br label %return
+
+sw.bb2:                                           ; preds = %entry, %entry, %entry
+  tail call void @foo3(i32 noundef signext %x)
+  br label %return
+
+sw.bb3:                                           ; preds = %entry
+  tail call void @foo4(i32 noundef signext 4)
+  br label %return
+
+return:                                           ; preds = %sw.bb, %sw.bb1, %sw.bb2, %sw.bb3, %entry
+  %retval.0 = phi i32 [ 0, %entry ], [ 4, %sw.bb3 ], [ %x, %sw.bb2 ], [ %x, %sw.bb1 ], [ 7, %sw.bb ]
+  ret i32 %retval.0
+}
+
+declare void @foo1(i32 noundef signext)
+
+declare void @foo2(i32 noundef signext)
+
+declare void @foo3(i32 noundef signext)
+
+declare void @foo4(i32 noundef signext)
diff --git a/llvm/test/CodeGen/PowerPC/combine-sext-and-shl-after-isel.ll b/llvm/test/CodeGen/PowerPC/combine-sext-and-shl-after-isel.ll
index 00a77f9..530169f 100644
--- a/llvm/test/CodeGen/PowerPC/combine-sext-and-shl-after-isel.ll
+++ b/llvm/test/CodeGen/PowerPC/combine-sext-and-shl-after-isel.ll
@@ -212,37 +212,33 @@ define hidden void @testCaller(i1 %incond) local_unnamed_addr align 2 nounwind {
 ; CHECK-NEXT:    std r30, 48(r1) # 8-byte Folded Spill
 ; CHECK-NEXT:    andi. r3, r3, 1
 ; CHECK-NEXT:    li r3, -1
+; CHECK-NEXT:    li r4, 0
 ; CHECK-NEXT:    li r30, 0
 ; CHECK-NEXT:    crmove 4*cr2+lt, gt
 ; CHECK-NEXT:    std r29, 40(r1) # 8-byte Folded Spill
 ; CHECK-NEXT:    b .LBB3_2
-; CHECK-NEXT:    .p2align 4
 ; CHECK-NEXT:  .LBB3_1: # %if.end116
 ; CHECK-NEXT:    #
 ; CHECK-NEXT:    bl callee
 ; CHECK-NEXT:    nop
 ; CHECK-NEXT:    mr r3, r29
-; CHECK-NEXT:  .LBB3_2: # %cond.end.i.i
-; CHECK-NEXT:    # =>This Loop Header: Depth=1
-; CHECK-NEXT:    # Child Loop BB3_3 Depth 2
-; CHECK-NEXT:    lwz r29, 0(r3)
-; CHECK-NEXT:    li r5, 0
-; CHECK-NEXT:    extsw r4, r29
-; CHECK-NEXT:    .p2align 5
-; CHECK-NEXT:  .LBB3_3: # %while.body5.i
-; CHECK-NEXT:    # Parent Loop BB3_2 Depth=1
-; CHECK-NEXT:    # => This Inner Loop Header: Depth=2
-; CHECK-NEXT:    addi r5, r5, -1
-; CHECK-NEXT:    cmpwi r5, 0
-; CHECK-NEXT:    bgt cr0, .LBB3_3
-; CHECK-NEXT:  # %bb.4: # %while.cond12.preheader.i
+; CHECK-NEXT:    li r4, 0
+; CHECK-NEXT:    .p2align 4
+; CHECK-NEXT:  .LBB3_2: # %while.body5.i
 ; CHECK-NEXT:    #
+; CHECK-NEXT:    addi r4, r4, -1
+; CHECK-NEXT:    cmpwi r4, 0
+; CHECK-NEXT:    bgt cr0, .LBB3_2
+; CHECK-NEXT:  # %bb.3: # %while.cond12.preheader.i
+; CHECK-NEXT:    #
+; CHECK-NEXT:    lwz r29, 0(r3)
 ; CHECK-NEXT:    bc 12, 4*cr2+lt, .LBB3_1
-; CHECK-NEXT:  # %bb.5: # %for.cond99.preheader
+; CHECK-NEXT:  # %bb.4: # %for.cond99.preheader
 ; CHECK-NEXT:    #
+; CHECK-NEXT:    extsw r4, r29
 ; CHECK-NEXT:    ld r5, 0(r3)
-; CHECK-NEXT:    sldi r4, r4, 2
 ; CHECK-NEXT:    stw r3, 0(r3)
+; CHECK-NEXT:    sldi r4, r4, 2
 ; CHECK-NEXT:    stwx r30, r5, r4
 ; CHECK-NEXT:    b .LBB3_1
 ;
@@ -256,37 +252,33 @@ define hidden void @testCaller(i1 %incond) local_unnamed_addr align 2 nounwind {
 ; CHECK-BE-NEXT:    std r30, 64(r1) # 8-byte Folded Spill
 ; CHECK-BE-NEXT:    andi. r3, r3, 1
 ; CHECK-BE-NEXT:    li r3, -1
+; CHECK-BE-NEXT:    li r4, 0
 ; CHECK-BE-NEXT:    li r30, 0
 ; CHECK-BE-NEXT:    crmove 4*cr2+lt, gt
 ; CHECK-BE-NEXT:    std r29, 56(r1) # 8-byte Folded Spill
 ; CHECK-BE-NEXT:    b .LBB3_2
-; CHECK-BE-NEXT:    .p2align 4
 ; CHECK-BE-NEXT:  .LBB3_1: # %if.end116
 ; CHECK-BE-NEXT:    #
 ; CHECK-BE-NEXT:    bl callee
 ; CHECK-BE-NEXT:    nop
 ; CHECK-BE-NEXT:    mr r3, r29
-; CHECK-BE-NEXT:  .LBB3_2: # %cond.end.i.i
-; CHECK-BE-NEXT:    # =>This Loop Header: Depth=1
-; CHECK-BE-NEXT:    # Child Loop BB3_3 Depth 2
-; CHECK-BE-NEXT:    lwz r29, 0(r3)
-; CHECK-BE-NEXT:    li r5, 0
-; CHECK-BE-NEXT:    extsw r4, r29
-; CHECK-BE-NEXT:    .p2align 5
-; CHECK-BE-NEXT:  .LBB3_3: # %while.body5.i
-; CHECK-BE-NEXT:    # Parent Loop BB3_2 Depth=1
-; CHECK-BE-NEXT:    # => This Inner Loop Header: Depth=2
-; CHECK-BE-NEXT:    addi r5, r5, -1
-; CHECK-BE-NEXT:    cmpwi r5, 0
-; CHECK-BE-NEXT:    bgt cr0, .LBB3_3
-; CHECK-BE-NEXT:  # %bb.4: # %while.cond12.preheader.i
+; CHECK-BE-NEXT:    li r4, 0
+; CHECK-BE-NEXT:    .p2align 4
+; CHECK-BE-NEXT:  .LBB3_2: # %while.body5.i
+; CHECK-BE-NEXT:    #
+; CHECK-BE-NEXT:    addi r4, r4, -1
+; CHECK-BE-NEXT:    cmpwi r4, 0
+; CHECK-BE-NEXT:    bgt cr0, .LBB3_2
+; CHECK-BE-NEXT:  # %bb.3: # %while.cond12.preheader.i
 ; CHECK-BE-NEXT:    #
+; CHECK-BE-NEXT:    lwz r29, 0(r3)
 ; CHECK-BE-NEXT:    bc 12, 4*cr2+lt, .LBB3_1
-; CHECK-BE-NEXT:  # %bb.5: # %for.cond99.preheader
+; CHECK-BE-NEXT:  # %bb.4: # %for.cond99.preheader
 ; CHECK-BE-NEXT:    #
+; CHECK-BE-NEXT:    extsw r4, r29
 ; CHECK-BE-NEXT:    ld r5, 0(r3)
-; CHECK-BE-NEXT:    sldi r4, r4, 2
 ; CHECK-BE-NEXT:    stw r3, 0(r3)
+; CHECK-BE-NEXT:    sldi r4, r4, 2
 ; CHECK-BE-NEXT:    stwx r30, r5, r4
 ; CHECK-BE-NEXT:    b .LBB3_1
 ;
@@ -300,32 +292,28 @@ define hidden void @testCaller(i1 %incond) local_unnamed_addr align 2 nounwind {
 ; CHECK-P9-NEXT:    std r0, 80(r1)
 ; CHECK-P9-NEXT:    std r30, 48(r1) # 8-byte Folded Spill
 ; CHECK-P9-NEXT:    li r3, -1
+; CHECK-P9-NEXT:    li r4, 0
 ; CHECK-P9-NEXT:    li r30, 0
 ; CHECK-P9-NEXT:    std r29, 40(r1) # 8-byte Folded Spill
 ; CHECK-P9-NEXT:    crmove 4*cr2+lt, gt
 ; CHECK-P9-NEXT:    b .LBB3_2
-; CHECK-P9-NEXT:    .p2align 4
 ; CHECK-P9-NEXT:  .LBB3_1: # %if.end116
 ; CHECK-P9-NEXT:    #
 ; CHECK-P9-NEXT:    bl callee
 ; CHECK-P9-NEXT:    nop
 ; CHECK-P9-NEXT:    mr r3, r29
-; CHECK-P9-NEXT:  .LBB3_2: # %cond.end.i.i
-; CHECK-P9-NEXT:    # =>This Loop Header: Depth=1
-; CHECK-P9-NEXT:    # Child Loop BB3_3 Depth 2
-; CHECK-P9-NEXT:    lwz r29, 0(r3)
 ; CHECK-P9-NEXT:    li r4, 0
-; CHECK-P9-NEXT:    .p2align 5
-; CHECK-P9-NEXT:  .LBB3_3: # %while.body5.i
-; CHECK-P9-NEXT:    # Parent Loop BB3_2 Depth=1
-; CHECK-P9-NEXT:    # => This Inner Loop Header: Depth=2
+; CHECK-P9-NEXT:    .p2align 4
+; CHECK-P9-NEXT:  .LBB3_2: # %while.body5.i
+; CHECK-P9-NEXT:    #
 ; CHECK-P9-NEXT:    addi r4, r4, -1
 ; CHECK-P9-NEXT:    cmpwi r4, 0
-; CHECK-P9-NEXT:    bgt cr0, .LBB3_3
-; CHECK-P9-NEXT:  # %bb.4: # %while.cond12.preheader.i
+; CHECK-P9-NEXT:    bgt cr0, .LBB3_2
+; CHECK-P9-NEXT:  # %bb.3: # %while.cond12.preheader.i
 ; CHECK-P9-NEXT:    #
+; CHECK-P9-NEXT:    lwz r29, 0(r3)
 ; CHECK-P9-NEXT:    bc 12, 4*cr2+lt, .LBB3_1
-; CHECK-P9-NEXT:  # %bb.5: # %for.cond99.preheader
+; CHECK-P9-NEXT:  # %bb.4: # %for.cond99.preheader
 ; CHECK-P9-NEXT:    #
 ; CHECK-P9-NEXT:    ld r4, 0(r3)
 ; CHECK-P9-NEXT:    extswsli r5, r29, 2
@@ -343,32 +331,28 @@ define hidden void @testCaller(i1 %incond) local_unnamed_addr align 2 nounwind {
 ; CHECK-P9-BE-NEXT:    std r0, 96(r1)
 ; CHECK-P9-BE-NEXT:    std r30, 64(r1) # 8-byte Folded Spill
 ; CHECK-P9-BE-NEXT:    li r3, -1
+; CHECK-P9-BE-NEXT:    li r4, 0
 ; CHECK-P9-BE-NEXT:    li r30, 0
 ; CHECK-P9-BE-NEXT:    std r29, 56(r1) # 8-byte Folded Spill
 ; CHECK-P9-BE-NEXT:    crmove 4*cr2+lt, gt
 ; CHECK-P9-BE-NEXT:    b .LBB3_2
-; CHECK-P9-BE-NEXT:    .p2align 4
 ; CHECK-P9-BE-NEXT:  .LBB3_1: # %if.end116
 ; CHECK-P9-BE-NEXT:    #
 ; CHECK-P9-BE-NEXT:    bl callee
 ; CHECK-P9-BE-NEXT:    nop
 ; CHECK-P9-BE-NEXT:    mr r3, r29
-; CHECK-P9-BE-NEXT:  .LBB3_2: # %cond.end.i.i
-; CHECK-P9-BE-NEXT:    # =>This Loop Header: Depth=1
-; CHECK-P9-BE-NEXT:    # Child Loop BB3_3 Depth 2
-; CHECK-P9-BE-NEXT:    lwz r29, 0(r3)
 ; CHECK-P9-BE-NEXT:    li r4, 0
-; CHECK-P9-BE-NEXT:    .p2align 5
-; CHECK-P9-BE-NEXT:  .LBB3_3: # %while.body5.i
-; CHECK-P9-BE-NEXT:    # Parent Loop BB3_2 Depth=1
-; CHECK-P9-BE-NEXT:    # => This Inner Loop Header: Depth=2
+; CHECK-P9-BE-NEXT:    .p2align 4
+; CHECK-P9-BE-NEXT:  .LBB3_2: # %while.body5.i
+; CHECK-P9-BE-NEXT:    #
 ; CHECK-P9-BE-NEXT:    addi r4, r4, -1
 ; CHECK-P9-BE-NEXT:    cmpwi r4, 0
-; CHECK-P9-BE-NEXT:    bgt cr0, .LBB3_3
-; CHECK-P9-BE-NEXT:  # %bb.4: # %while.cond12.preheader.i
+; CHECK-P9-BE-NEXT:    bgt cr0, .LBB3_2
+; CHECK-P9-BE-NEXT:  # %bb.3: # %while.cond12.preheader.i
 ; CHECK-P9-BE-NEXT:    #
+; CHECK-P9-BE-NEXT:    lwz r29, 0(r3)
 ; CHECK-P9-BE-NEXT:    bc 12, 4*cr2+lt, .LBB3_1
-; CHECK-P9-BE-NEXT:  # %bb.5: # %for.cond99.preheader
+; CHECK-P9-BE-NEXT:  # %bb.4: # %for.cond99.preheader
 ; CHECK-P9-BE-NEXT:    #
 ; CHECK-P9-BE-NEXT:    ld r4, 0(r3)
 ; CHECK-P9-BE-NEXT:    extswsli r5, r29, 2
diff --git a/llvm/test/CodeGen/PowerPC/vec_insert_elt.ll b/llvm/test/CodeGen/PowerPC/vec_insert_elt.ll
index 291a9c1..b006c78 100644
--- a/llvm/test/CodeGen/PowerPC/vec_insert_elt.ll
+++ b/llvm/test/CodeGen/PowerPC/vec_insert_elt.ll
@@ -242,17 +242,14 @@ define <2 x i64> @testDoubleword(<2 x i64> %a, i64 %b, i64 %idx) {
 ; AIX-P8-32-LABEL: testDoubleword:
 ; AIX-P8-32:       # %bb.0: # %entry
 ; AIX-P8-32-NEXT:    add r6, r6, r6
-; AIX-P8-32-NEXT:    addi r5, r1, -32
+; AIX-P8-32-NEXT:    addi r5, r1, -16
 ; AIX-P8-32-NEXT:    rlwinm r7, r6, 2, 28, 29
-; AIX-P8-32-NEXT:    stxvw4x v2, 0, r5
+; AIX-P8-32-NEXT:    stxvd2x v2, 0, r5
 ; AIX-P8-32-NEXT:    stwx r3, r5, r7
-; AIX-P8-32-NEXT:    addi r3, r1, -16
-; AIX-P8-32-NEXT:    lxvw4x vs0, 0, r5
-; AIX-P8-32-NEXT:    addi r5, r6, 1
-; AIX-P8-32-NEXT:    rlwinm r5, r5, 2, 28, 29
-; AIX-P8-32-NEXT:    stxvw4x vs0, 0, r3
-; AIX-P8-32-NEXT:    stwx r4, r3, r5
-; AIX-P8-32-NEXT:    lxvw4x v2, 0, r3
+; AIX-P8-32-NEXT:    addi r3, r6, 1
+; AIX-P8-32-NEXT:    rlwinm r3, r3, 2, 28, 29
+; AIX-P8-32-NEXT:    stwx r4, r5, r3
+; AIX-P8-32-NEXT:    lxvd2x v2, 0, r5
 ; AIX-P8-32-NEXT:    blr
 entry:
   %vecins = insertelement <2 x i64> %a, i64 %b, i64 %idx
@@ -426,17 +423,14 @@ define <4 x float> @testFloat2(<4 x float> %a, ptr %b, i32 zeroext %idx1, i32 ze
 ; AIX-P8-LABEL: testFloat2:
 ; AIX-P8:       # %bb.0: # %entry
 ; AIX-P8-NEXT:    lwz r6, 0(r3)
-; AIX-P8-NEXT:    rlwinm r4, r4, 2, 28, 29
-; AIX-P8-NEXT:    addi r7, r1, -32
+; AIX-P8-NEXT:    lwz r3, 1(r3)
+; AIX-P8-NEXT:    addi r7, r1, -16
 ; AIX-P8-NEXT:    stxvw4x v2, 0, r7
 ; AIX-P8-NEXT:    rlwinm r5, r5, 2, 28, 29
+; AIX-P8-NEXT:    rlwinm r4, r4, 2, 28, 29
 ; AIX-P8-NEXT:    stwx r6, r7, r4
-; AIX-P8-NEXT:    addi r4, r1, -16
-; AIX-P8-NEXT:    lxvw4x vs0, 0, r7
-; AIX-P8-NEXT:    lwz r3, 1(r3)
-; AIX-P8-NEXT:    stxvw4x vs0, 0, r4
-; AIX-P8-NEXT:    stwx r3, r4, r5
-; AIX-P8-NEXT:    lxvw4x v2, 0, r4
+; AIX-P8-NEXT:    stwx r3, r7, r5
+; AIX-P8-NEXT:    lxvw4x v2, 0, r7
 ; AIX-P8-NEXT:    blr
 entry:
   %add.ptr1 = getelementptr inbounds i8, ptr %b, i64 1
@@ -493,38 +487,32 @@ define <4 x float> @testFloat3(<4 x float> %a, ptr %b, i32 zeroext %idx1, i32 ze
 ;
 ; AIX-P8-64-LABEL: testFloat3:
 ; AIX-P8-64:       # %bb.0: # %entry
+; AIX-P8-64-NEXT:    li r7, 1
 ; AIX-P8-64-NEXT:    lis r6, 1
-; AIX-P8-64-NEXT:    rlwinm r4, r4, 2, 28, 29
-; AIX-P8-64-NEXT:    addi r7, r1, -32
 ; AIX-P8-64-NEXT:    rlwinm r5, r5, 2, 28, 29
+; AIX-P8-64-NEXT:    rlwinm r4, r4, 2, 28, 29
+; AIX-P8-64-NEXT:    rldic r7, r7, 36, 27
 ; AIX-P8-64-NEXT:    lwzx r6, r3, r6
+; AIX-P8-64-NEXT:    lwzx r3, r3, r7
+; AIX-P8-64-NEXT:    addi r7, r1, -16
 ; AIX-P8-64-NEXT:    stxvw4x v2, 0, r7
 ; AIX-P8-64-NEXT:    stwx r6, r7, r4
-; AIX-P8-64-NEXT:    li r4, 1
-; AIX-P8-64-NEXT:    lxvw4x vs0, 0, r7
-; AIX-P8-64-NEXT:    rldic r4, r4, 36, 27
-; AIX-P8-64-NEXT:    lwzx r3, r3, r4
-; AIX-P8-64-NEXT:    addi r4, r1, -16
-; AIX-P8-64-NEXT:    stxvw4x vs0, 0, r4
-; AIX-P8-64-NEXT:    stwx r3, r4, r5
-; AIX-P8-64-NEXT:    lxvw4x v2, 0, r4
+; AIX-P8-64-NEXT:    stwx r3, r7, r5
+; AIX-P8-64-NEXT:    lxvw4x v2, 0, r7
 ; AIX-P8-64-NEXT:    blr
 ;
 ; AIX-P8-32-LABEL: testFloat3:
 ; AIX-P8-32:       # %bb.0: # %entry
 ; AIX-P8-32-NEXT:    lis r6, 1
-; AIX-P8-32-NEXT:    rlwinm r4, r4, 2, 28, 29
-; AIX-P8-32-NEXT:    addi r7, r1, -32
 ; AIX-P8-32-NEXT:    rlwinm r5, r5, 2, 28, 29
+; AIX-P8-32-NEXT:    rlwinm r4, r4, 2, 28, 29
+; AIX-P8-32-NEXT:    addi r7, r1, -16
 ; AIX-P8-32-NEXT:    lwzx r6, r3, r6
+; AIX-P8-32-NEXT:    lwz r3, 0(r3)
 ; AIX-P8-32-NEXT:    stxvw4x v2, 0, r7
 ; AIX-P8-32-NEXT:    stwx r6, r7, r4
-; AIX-P8-32-NEXT:    addi r4, r1, -16
-; AIX-P8-32-NEXT:    lxvw4x vs0, 0, r7
-; AIX-P8-32-NEXT:    lwz r3, 0(r3)
-; AIX-P8-32-NEXT:    stxvw4x vs0, 0, r4
-; AIX-P8-32-NEXT:    stwx r3, r4, r5
-; AIX-P8-32-NEXT:    lxvw4x v2, 0, r4
+; AIX-P8-32-NEXT:    stwx r3, r7, r5
+; AIX-P8-32-NEXT:    lxvw4x v2, 0, r7
 ; AIX-P8-32-NEXT:    blr
 entry:
   %add.ptr = getelementptr inbounds i8, ptr %b, i64 65536