[PowerPC][NFC] Add test case for byval argument passing

Add a test case for byval argument passing where the argument size is more than
8 bytes and is not a factor of 8 bytes.

diff --git a/llvm/test/CodeGen/PowerPC/ppc64-byval-larger-struct.ll b/llvm/test/CodeGen/PowerPC/ppc64-byval-larger-struct.ll
new file mode 100644 (file)
index 0000000..d05ddf0
--- /dev/null
+++ b/llvm/test/CodeGen/PowerPC/ppc64-byval-larger-struct.ll
@@ -0,0 +1,1026 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -verify-machineinstrs --mtriple powerpc64le-unknown-linux-gnu \
+; RUN:   -mcpu=pwr8 -ppc-asm-full-reg-names < %s | FileCheck %s --check-prefix=P8LE
+; RUN: llc -verify-machineinstrs --mtriple powerpc64le-unknown-linux-gnu \
+; RUN:   -mcpu=pwr9 -ppc-asm-full-reg-names < %s | FileCheck %s --check-prefix=P9LE
+; RUN: llc -verify-machineinstrs --mtriple powerpc64le-unknown-linux-gnu \
+; RUN:   -mcpu=pwr10 -ppc-asm-full-reg-names < %s | FileCheck %s --check-prefix=P10LE
+; RUN: llc -verify-machineinstrs --mtriple powerpc64-unknown-linux-gnu \
+; RUN:   -mcpu=pwr8 -ppc-asm-full-reg-names < %s | FileCheck %s --check-prefix=P8BE
+; RUN: llc -verify-machineinstrs --mtriple powerpc64-unknown-linux-gnu \
+; RUN:   -mcpu=pwr9 -ppc-asm-full-reg-names < %s | FileCheck %s --check-prefix=P9BE
+; RUN: llc -verify-machineinstrs --mtriple powerpc64-unknown-linux-gnu \
+; RUN:   -mcpu=pwr10 -ppc-asm-full-reg-names < %s | FileCheck %s --check-prefix=P10BE
+
+define signext i8 @caller_9([9 x i8]* nocapture readonly byval([9 x i8]) %data) #0 {
+; P8LE-LABEL: caller_9:
+; P8LE:       # %bb.0: # %entry
+; P8LE-NEXT:    mflr r0
+; P8LE-NEXT:    std r0, 16(r1)
+; P8LE-NEXT:    stdu r1, -80(r1)
+; P8LE-NEXT:    addi r5, r1, 71
+; P8LE-NEXT:    std r3, 48(r1)
+; P8LE-NEXT:    std r4, 56(r1)
+; P8LE-NEXT:    stdx r3, 0, r5
+; P8LE-NEXT:    mr r3, r5
+; P8LE-NEXT:    stb r4, 79(r1)
+; P8LE-NEXT:    bl callee
+; P8LE-NEXT:    nop
+; P8LE-NEXT:    li r3, 0
+; P8LE-NEXT:    addi r1, r1, 80
+; P8LE-NEXT:    ld r0, 16(r1)
+; P8LE-NEXT:    mtlr r0
+; P8LE-NEXT:    blr
+;
+; P9LE-LABEL: caller_9:
+; P9LE:       # %bb.0: # %entry
+; P9LE-NEXT:    mflr r0
+; P9LE-NEXT:    std r0, 16(r1)
+; P9LE-NEXT:    stdu r1, -80(r1)
+; P9LE-NEXT:    addi r5, r1, 71
+; P9LE-NEXT:    std r3, 48(r1)
+; P9LE-NEXT:    std r4, 56(r1)
+; P9LE-NEXT:    stb r4, 79(r1)
+; P9LE-NEXT:    stdx r3, 0, r5
+; P9LE-NEXT:    mr r3, r5
+; P9LE-NEXT:    bl callee
+; P9LE-NEXT:    nop
+; P9LE-NEXT:    li r3, 0
+; P9LE-NEXT:    addi r1, r1, 80
+; P9LE-NEXT:    ld r0, 16(r1)
+; P9LE-NEXT:    mtlr r0
+; P9LE-NEXT:    blr
+;
+; P10LE-LABEL: caller_9:
+; P10LE:       # %bb.0: # %entry
+; P10LE-NEXT:    mflr r0
+; P10LE-NEXT:    std r0, 16(r1)
+; P10LE-NEXT:    stdu r1, -80(r1)
+; P10LE-NEXT:    addi r5, r1, 71
+; P10LE-NEXT:    std r3, 48(r1)
+; P10LE-NEXT:    std r4, 56(r1)
+; P10LE-NEXT:    stb r4, 79(r1)
+; P10LE-NEXT:    stdx r3, 0, r5
+; P10LE-NEXT:    mr r3, r5
+; P10LE-NEXT:    bl callee@notoc
+; P10LE-NEXT:    li r3, 0
+; P10LE-NEXT:    addi r1, r1, 80
+; P10LE-NEXT:    ld r0, 16(r1)
+; P10LE-NEXT:    mtlr r0
+; P10LE-NEXT:    blr
+;
+; P8BE-LABEL: caller_9:
+; P8BE:       # %bb.0: # %entry
+; P8BE-NEXT:    mflr r0
+; P8BE-NEXT:    std r0, 16(r1)
+; P8BE-NEXT:    stdu r1, -144(r1)
+; P8BE-NEXT:    lbz r6, 200(r1)
+; P8BE-NEXT:    addi r5, r1, 135
+; P8BE-NEXT:    std r3, 192(r1)
+; P8BE-NEXT:    std r4, 200(r1)
+; P8BE-NEXT:    stdx r3, 0, r5
+; P8BE-NEXT:    mr r3, r5
+; P8BE-NEXT:    stb r6, 143(r1)
+; P8BE-NEXT:    bl callee
+; P8BE-NEXT:    nop
+; P8BE-NEXT:    li r3, 0
+; P8BE-NEXT:    addi r1, r1, 144
+; P8BE-NEXT:    ld r0, 16(r1)
+; P8BE-NEXT:    mtlr r0
+; P8BE-NEXT:    blr
+;
+; P9BE-LABEL: caller_9:
+; P9BE:       # %bb.0: # %entry
+; P9BE-NEXT:    mflr r0
+; P9BE-NEXT:    std r0, 16(r1)
+; P9BE-NEXT:    stdu r1, -144(r1)
+; P9BE-NEXT:    std r4, 200(r1)
+; P9BE-NEXT:    addi r4, r1, 135
+; P9BE-NEXT:    lbz r5, 200(r1)
+; P9BE-NEXT:    std r3, 192(r1)
+; P9BE-NEXT:    stdx r3, 0, r4
+; P9BE-NEXT:    mr r3, r4
+; P9BE-NEXT:    stb r5, 143(r1)
+; P9BE-NEXT:    bl callee
+; P9BE-NEXT:    nop
+; P9BE-NEXT:    li r3, 0
+; P9BE-NEXT:    addi r1, r1, 144
+; P9BE-NEXT:    ld r0, 16(r1)
+; P9BE-NEXT:    mtlr r0
+; P9BE-NEXT:    blr
+;
+; P10BE-LABEL: caller_9:
+; P10BE:       # %bb.0: # %entry
+; P10BE-NEXT:    mflr r0
+; P10BE-NEXT:    std r0, 16(r1)
+; P10BE-NEXT:    stdu r1, -144(r1)
+; P10BE-NEXT:    lbz r5, 200(r1)
+; P10BE-NEXT:    std r4, 200(r1)
+; P10BE-NEXT:    std r3, 192(r1)
+; P10BE-NEXT:    addi r4, r1, 135
+; P10BE-NEXT:    stdx r3, 0, r4
+; P10BE-NEXT:    mr r3, r4
+; P10BE-NEXT:    stb r5, 143(r1)
+; P10BE-NEXT:    bl callee
+; P10BE-NEXT:    nop
+; P10BE-NEXT:    li r3, 0
+; P10BE-NEXT:    addi r1, r1, 144
+; P10BE-NEXT:    ld r0, 16(r1)
+; P10BE-NEXT:    mtlr r0
+; P10BE-NEXT:    blr
+entry:
+  %_param_data = alloca [9 x i8], align 1
+  %.elt0 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 0
+  %.unpack0 = load i8, i8* %.elt0, align 1
+  %.elt1 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 1
+  %.unpack1 = load i8, i8* %.elt1, align 1
+  %.elt2 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 2
+  %.unpack2 = load i8, i8* %.elt2, align 1
+  %.elt3 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 3
+  %.unpack3 = load i8, i8* %.elt3, align 1
+  %.elt4 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 4
+  %.unpack4 = load i8, i8* %.elt4, align 1
+  %.elt5 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 5
+  %.unpack5 = load i8, i8* %.elt5, align 1
+  %.elt6 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 6
+  %.unpack6 = load i8, i8* %.elt6, align 1
+  %.elt7 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 7
+  %.unpack7 = load i8, i8* %.elt7, align 1
+  %.elt8 = getelementptr inbounds [9 x i8], [9 x i8]* %data, i64 0, i64 8
+  %.unpack8 = load i8, i8* %.elt8, align 1
+  %.temp.0.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 0
+  store i8 %.unpack0, i8* %.temp.0.gep, align 1
+  %.temp.1.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 1
+  store i8 %.unpack1, i8* %.temp.1.gep, align 1
+  %.temp.2.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 2
+  store i8 %.unpack2, i8* %.temp.2.gep, align 1
+  %.temp.3.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 3
+  store i8 %.unpack3, i8* %.temp.3.gep, align 1
+  %.temp.4.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 4
+  store i8 %.unpack4, i8* %.temp.4.gep, align 1
+  %.temp.5.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 5
+  store i8 %.unpack5, i8* %.temp.5.gep, align 1
+  %.temp.6.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 6
+  store i8 %.unpack6, i8* %.temp.6.gep, align 1
+  %.temp.7.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 7
+  store i8 %.unpack7, i8* %.temp.7.gep, align 1
+  %.temp.8.gep = getelementptr inbounds [9 x i8], [9 x i8]* %_param_data, i64 0, i64 8
+  store i8 %.unpack8, i8* %.temp.8.gep, align 1
+  call void @callee(i8* nonnull %.temp.0.gep)
+  ret i8 0
+}
+
+define signext i8 @caller_10([10 x i8]* nocapture readonly byval([10 x i8]) %data) #0 {
+; P8LE-LABEL: caller_10:
+; P8LE:       # %bb.0: # %entry
+; P8LE-NEXT:    mflr r0
+; P8LE-NEXT:    std r0, 16(r1)
+; P8LE-NEXT:    stdu r1, -80(r1)
+; P8LE-NEXT:    addi r5, r1, 70
+; P8LE-NEXT:    std r3, 48(r1)
+; P8LE-NEXT:    std r4, 56(r1)
+; P8LE-NEXT:    stdx r3, 0, r5
+; P8LE-NEXT:    mr r3, r5
+; P8LE-NEXT:    sth r4, 78(r1)
+; P8LE-NEXT:    bl callee
+; P8LE-NEXT:    nop
+; P8LE-NEXT:    li r3, 0
+; P8LE-NEXT:    addi r1, r1, 80
+; P8LE-NEXT:    ld r0, 16(r1)
+; P8LE-NEXT:    mtlr r0
+; P8LE-NEXT:    blr
+;
+; P9LE-LABEL: caller_10:
+; P9LE:       # %bb.0: # %entry
+; P9LE-NEXT:    mflr r0
+; P9LE-NEXT:    std r0, 16(r1)
+; P9LE-NEXT:    stdu r1, -80(r1)
+; P9LE-NEXT:    addi r5, r1, 70
+; P9LE-NEXT:    std r3, 48(r1)
+; P9LE-NEXT:    std r4, 56(r1)
+; P9LE-NEXT:    sth r4, 78(r1)
+; P9LE-NEXT:    stdx r3, 0, r5
+; P9LE-NEXT:    mr r3, r5
+; P9LE-NEXT:    bl callee
+; P9LE-NEXT:    nop
+; P9LE-NEXT:    li r3, 0
+; P9LE-NEXT:    addi r1, r1, 80
+; P9LE-NEXT:    ld r0, 16(r1)
+; P9LE-NEXT:    mtlr r0
+; P9LE-NEXT:    blr
+;
+; P10LE-LABEL: caller_10:
+; P10LE:       # %bb.0: # %entry
+; P10LE-NEXT:    mflr r0
+; P10LE-NEXT:    std r0, 16(r1)
+; P10LE-NEXT:    stdu r1, -80(r1)
+; P10LE-NEXT:    addi r5, r1, 70
+; P10LE-NEXT:    std r3, 48(r1)
+; P10LE-NEXT:    std r4, 56(r1)
+; P10LE-NEXT:    sth r4, 78(r1)
+; P10LE-NEXT:    stdx r3, 0, r5
+; P10LE-NEXT:    mr r3, r5
+; P10LE-NEXT:    bl callee@notoc
+; P10LE-NEXT:    li r3, 0
+; P10LE-NEXT:    addi r1, r1, 80
+; P10LE-NEXT:    ld r0, 16(r1)
+; P10LE-NEXT:    mtlr r0
+; P10LE-NEXT:    blr
+;
+; P8BE-LABEL: caller_10:
+; P8BE:       # %bb.0: # %entry
+; P8BE-NEXT:    mflr r0
+; P8BE-NEXT:    std r0, 16(r1)
+; P8BE-NEXT:    stdu r1, -144(r1)
+; P8BE-NEXT:    lhz r6, 200(r1)
+; P8BE-NEXT:    addi r5, r1, 134
+; P8BE-NEXT:    std r3, 192(r1)
+; P8BE-NEXT:    std r4, 200(r1)
+; P8BE-NEXT:    stdx r3, 0, r5
+; P8BE-NEXT:    mr r3, r5
+; P8BE-NEXT:    sth r6, 142(r1)
+; P8BE-NEXT:    bl callee
+; P8BE-NEXT:    nop
+; P8BE-NEXT:    li r3, 0
+; P8BE-NEXT:    addi r1, r1, 144
+; P8BE-NEXT:    ld r0, 16(r1)
+; P8BE-NEXT:    mtlr r0
+; P8BE-NEXT:    blr
+;
+; P9BE-LABEL: caller_10:
+; P9BE:       # %bb.0: # %entry
+; P9BE-NEXT:    mflr r0
+; P9BE-NEXT:    std r0, 16(r1)
+; P9BE-NEXT:    stdu r1, -144(r1)
+; P9BE-NEXT:    std r4, 200(r1)
+; P9BE-NEXT:    addi r4, r1, 134
+; P9BE-NEXT:    lhz r5, 200(r1)
+; P9BE-NEXT:    std r3, 192(r1)
+; P9BE-NEXT:    stdx r3, 0, r4
+; P9BE-NEXT:    mr r3, r4
+; P9BE-NEXT:    sth r5, 142(r1)
+; P9BE-NEXT:    bl callee
+; P9BE-NEXT:    nop
+; P9BE-NEXT:    li r3, 0
+; P9BE-NEXT:    addi r1, r1, 144
+; P9BE-NEXT:    ld r0, 16(r1)
+; P9BE-NEXT:    mtlr r0
+; P9BE-NEXT:    blr
+;
+; P10BE-LABEL: caller_10:
+; P10BE:       # %bb.0: # %entry
+; P10BE-NEXT:    mflr r0
+; P10BE-NEXT:    std r0, 16(r1)
+; P10BE-NEXT:    stdu r1, -144(r1)
+; P10BE-NEXT:    lhz r5, 200(r1)
+; P10BE-NEXT:    std r4, 200(r1)
+; P10BE-NEXT:    std r3, 192(r1)
+; P10BE-NEXT:    addi r4, r1, 134
+; P10BE-NEXT:    stdx r3, 0, r4
+; P10BE-NEXT:    mr r3, r4
+; P10BE-NEXT:    sth r5, 142(r1)
+; P10BE-NEXT:    bl callee
+; P10BE-NEXT:    nop
+; P10BE-NEXT:    li r3, 0
+; P10BE-NEXT:    addi r1, r1, 144
+; P10BE-NEXT:    ld r0, 16(r1)
+; P10BE-NEXT:    mtlr r0
+; P10BE-NEXT:    blr
+entry:
+  %_param_data = alloca [10 x i8], align 1
+  %.elt0 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 0
+  %.unpack0 = load i8, i8* %.elt0, align 1
+  %.elt1 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 1
+  %.unpack1 = load i8, i8* %.elt1, align 1
+  %.elt2 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 2
+  %.unpack2 = load i8, i8* %.elt2, align 1
+  %.elt3 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 3
+  %.unpack3 = load i8, i8* %.elt3, align 1
+  %.elt4 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 4
+  %.unpack4 = load i8, i8* %.elt4, align 1
+  %.elt5 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 5
+  %.unpack5 = load i8, i8* %.elt5, align 1
+  %.elt6 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 6
+  %.unpack6 = load i8, i8* %.elt6, align 1
+  %.elt7 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 7
+  %.unpack7 = load i8, i8* %.elt7, align 1
+  %.elt8 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 8
+  %.unpack8 = load i8, i8* %.elt8, align 1
+  %.elt9 = getelementptr inbounds [10 x i8], [10 x i8]* %data, i64 0, i64 9
+  %.unpack9 = load i8, i8* %.elt9, align 1
+  %.temp.0.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 0
+  store i8 %.unpack0, i8* %.temp.0.gep, align 1
+  %.temp.1.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 1
+  store i8 %.unpack1, i8* %.temp.1.gep, align 1
+  %.temp.2.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 2
+  store i8 %.unpack2, i8* %.temp.2.gep, align 1
+  %.temp.3.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 3
+  store i8 %.unpack3, i8* %.temp.3.gep, align 1
+  %.temp.4.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 4
+  store i8 %.unpack4, i8* %.temp.4.gep, align 1
+  %.temp.5.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 5
+  store i8 %.unpack5, i8* %.temp.5.gep, align 1
+  %.temp.6.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 6
+  store i8 %.unpack6, i8* %.temp.6.gep, align 1
+  %.temp.7.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 7
+  store i8 %.unpack7, i8* %.temp.7.gep, align 1
+  %.temp.8.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 8
+  store i8 %.unpack8, i8* %.temp.8.gep, align 1
+  %.temp.9.gep = getelementptr inbounds [10 x i8], [10 x i8]* %_param_data, i64 0, i64 9
+  store i8 %.unpack9, i8* %.temp.9.gep, align 1
+  call void @callee(i8* nonnull %.temp.0.gep)
+  ret i8 0
+}
+
+define signext i8 @caller_12([12 x i8]* nocapture readonly byval([12 x i8]) %data) #0 {
+; P8LE-LABEL: caller_12:
+; P8LE:       # %bb.0: # %entry
+; P8LE-NEXT:    mflr r0
+; P8LE-NEXT:    std r0, 16(r1)
+; P8LE-NEXT:    stdu r1, -80(r1)
+; P8LE-NEXT:    addi r5, r1, 68
+; P8LE-NEXT:    std r3, 48(r1)
+; P8LE-NEXT:    std r4, 56(r1)
+; P8LE-NEXT:    std r3, 68(r1)
+; P8LE-NEXT:    mr r3, r5
+; P8LE-NEXT:    stw r4, 76(r1)
+; P8LE-NEXT:    bl callee
+; P8LE-NEXT:    nop
+; P8LE-NEXT:    li r3, 0
+; P8LE-NEXT:    addi r1, r1, 80
+; P8LE-NEXT:    ld r0, 16(r1)
+; P8LE-NEXT:    mtlr r0
+; P8LE-NEXT:    blr
+;
+; P9LE-LABEL: caller_12:
+; P9LE:       # %bb.0: # %entry
+; P9LE-NEXT:    mflr r0
+; P9LE-NEXT:    std r0, 16(r1)
+; P9LE-NEXT:    stdu r1, -80(r1)
+; P9LE-NEXT:    addi r5, r1, 68
+; P9LE-NEXT:    std r3, 48(r1)
+; P9LE-NEXT:    std r4, 56(r1)
+; P9LE-NEXT:    stw r4, 76(r1)
+; P9LE-NEXT:    std r3, 68(r1)
+; P9LE-NEXT:    mr r3, r5
+; P9LE-NEXT:    bl callee
+; P9LE-NEXT:    nop
+; P9LE-NEXT:    li r3, 0
+; P9LE-NEXT:    addi r1, r1, 80
+; P9LE-NEXT:    ld r0, 16(r1)
+; P9LE-NEXT:    mtlr r0
+; P9LE-NEXT:    blr
+;
+; P10LE-LABEL: caller_12:
+; P10LE:       # %bb.0: # %entry
+; P10LE-NEXT:    mflr r0
+; P10LE-NEXT:    std r0, 16(r1)
+; P10LE-NEXT:    stdu r1, -80(r1)
+; P10LE-NEXT:    addi r5, r1, 68
+; P10LE-NEXT:    std r3, 48(r1)
+; P10LE-NEXT:    std r4, 56(r1)
+; P10LE-NEXT:    stw r4, 76(r1)
+; P10LE-NEXT:    std r3, 68(r1)
+; P10LE-NEXT:    mr r3, r5
+; P10LE-NEXT:    bl callee@notoc
+; P10LE-NEXT:    li r3, 0
+; P10LE-NEXT:    addi r1, r1, 80
+; P10LE-NEXT:    ld r0, 16(r1)
+; P10LE-NEXT:    mtlr r0
+; P10LE-NEXT:    blr
+;
+; P8BE-LABEL: caller_12:
+; P8BE:       # %bb.0: # %entry
+; P8BE-NEXT:    mflr r0
+; P8BE-NEXT:    std r0, 16(r1)
+; P8BE-NEXT:    stdu r1, -144(r1)
+; P8BE-NEXT:    lwz r6, 200(r1)
+; P8BE-NEXT:    addi r5, r1, 132
+; P8BE-NEXT:    std r3, 192(r1)
+; P8BE-NEXT:    std r4, 200(r1)
+; P8BE-NEXT:    std r3, 132(r1)
+; P8BE-NEXT:    mr r3, r5
+; P8BE-NEXT:    stw r6, 140(r1)
+; P8BE-NEXT:    bl callee
+; P8BE-NEXT:    nop
+; P8BE-NEXT:    li r3, 0
+; P8BE-NEXT:    addi r1, r1, 144
+; P8BE-NEXT:    ld r0, 16(r1)
+; P8BE-NEXT:    mtlr r0
+; P8BE-NEXT:    blr
+;
+; P9BE-LABEL: caller_12:
+; P9BE:       # %bb.0: # %entry
+; P9BE-NEXT:    mflr r0
+; P9BE-NEXT:    std r0, 16(r1)
+; P9BE-NEXT:    stdu r1, -144(r1)
+; P9BE-NEXT:    std r4, 200(r1)
+; P9BE-NEXT:    addi r4, r1, 132
+; P9BE-NEXT:    lwz r5, 200(r1)
+; P9BE-NEXT:    std r3, 192(r1)
+; P9BE-NEXT:    std r3, 132(r1)
+; P9BE-NEXT:    mr r3, r4
+; P9BE-NEXT:    stw r5, 140(r1)
+; P9BE-NEXT:    bl callee
+; P9BE-NEXT:    nop
+; P9BE-NEXT:    li r3, 0
+; P9BE-NEXT:    addi r1, r1, 144
+; P9BE-NEXT:    ld r0, 16(r1)
+; P9BE-NEXT:    mtlr r0
+; P9BE-NEXT:    blr
+;
+; P10BE-LABEL: caller_12:
+; P10BE:       # %bb.0: # %entry
+; P10BE-NEXT:    mflr r0
+; P10BE-NEXT:    std r0, 16(r1)
+; P10BE-NEXT:    stdu r1, -144(r1)
+; P10BE-NEXT:    lwz r5, 200(r1)
+; P10BE-NEXT:    std r4, 200(r1)
+; P10BE-NEXT:    std r3, 192(r1)
+; P10BE-NEXT:    addi r4, r1, 132
+; P10BE-NEXT:    std r3, 132(r1)
+; P10BE-NEXT:    mr r3, r4
+; P10BE-NEXT:    stw r5, 140(r1)
+; P10BE-NEXT:    bl callee
+; P10BE-NEXT:    nop
+; P10BE-NEXT:    li r3, 0
+; P10BE-NEXT:    addi r1, r1, 144
+; P10BE-NEXT:    ld r0, 16(r1)
+; P10BE-NEXT:    mtlr r0
+; P10BE-NEXT:    blr
+entry:
+  %_param_data = alloca [12 x i8], align 1
+  %.elt0 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 0
+  %.unpack0 = load i8, i8* %.elt0, align 1
+  %.elt1 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 1
+  %.unpack1 = load i8, i8* %.elt1, align 1
+  %.elt2 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 2
+  %.unpack2 = load i8, i8* %.elt2, align 1
+  %.elt3 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 3
+  %.unpack3 = load i8, i8* %.elt3, align 1
+  %.elt4 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 4
+  %.unpack4 = load i8, i8* %.elt4, align 1
+  %.elt5 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 5
+  %.unpack5 = load i8, i8* %.elt5, align 1
+  %.elt6 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 6
+  %.unpack6 = load i8, i8* %.elt6, align 1
+  %.elt7 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 7
+  %.unpack7 = load i8, i8* %.elt7, align 1
+  %.elt8 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 8
+  %.unpack8 = load i8, i8* %.elt8, align 1
+  %.elt9 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 9
+  %.unpack9 = load i8, i8* %.elt9, align 1
+  %.elt10 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 10
+  %.unpack10 = load i8, i8* %.elt10, align 1
+  %.elt11 = getelementptr inbounds [12 x i8], [12 x i8]* %data, i64 0, i64 11
+  %.unpack11 = load i8, i8* %.elt11, align 1
+  %.temp.0.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 0
+  store i8 %.unpack0, i8* %.temp.0.gep, align 1
+  %.temp.1.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 1
+  store i8 %.unpack1, i8* %.temp.1.gep, align 1
+  %.temp.2.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 2
+  store i8 %.unpack2, i8* %.temp.2.gep, align 1
+  %.temp.3.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 3
+  store i8 %.unpack3, i8* %.temp.3.gep, align 1
+  %.temp.4.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 4
+  store i8 %.unpack4, i8* %.temp.4.gep, align 1
+  %.temp.5.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 5
+  store i8 %.unpack5, i8* %.temp.5.gep, align 1
+  %.temp.6.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 6
+  store i8 %.unpack6, i8* %.temp.6.gep, align 1
+  %.temp.7.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 7
+  store i8 %.unpack7, i8* %.temp.7.gep, align 1
+  %.temp.8.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 8
+  store i8 %.unpack8, i8* %.temp.8.gep, align 1
+  %.temp.9.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 9
+  store i8 %.unpack9, i8* %.temp.9.gep, align 1
+  %.temp.10.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 10
+  store i8 %.unpack10, i8* %.temp.10.gep, align 1
+  %.temp.11.gep = getelementptr inbounds [12 x i8], [12 x i8]* %_param_data, i64 0, i64 11
+  store i8 %.unpack11, i8* %.temp.11.gep, align 1
+  call void @callee(i8* nonnull %.temp.0.gep)
+  ret i8 0
+}
+
+define signext i8 @caller_14([14 x i8]* nocapture readonly byval([14 x i8]) %data) #0 {
+; P8LE-LABEL: caller_14:
+; P8LE:       # %bb.0: # %entry
+; P8LE-NEXT:    mflr r0
+; P8LE-NEXT:    std r0, 16(r1)
+; P8LE-NEXT:    stdu r1, -80(r1)
+; P8LE-NEXT:    addi r5, r1, 66
+; P8LE-NEXT:    std r3, 48(r1)
+; P8LE-NEXT:    std r4, 56(r1)
+; P8LE-NEXT:    stdx r3, 0, r5
+; P8LE-NEXT:    mr r3, r5
+; P8LE-NEXT:    stw r4, 74(r1)
+; P8LE-NEXT:    bl callee
+; P8LE-NEXT:    nop
+; P8LE-NEXT:    li r3, 0
+; P8LE-NEXT:    addi r1, r1, 80
+; P8LE-NEXT:    ld r0, 16(r1)
+; P8LE-NEXT:    mtlr r0
+; P8LE-NEXT:    blr
+;
+; P9LE-LABEL: caller_14:
+; P9LE:       # %bb.0: # %entry
+; P9LE-NEXT:    mflr r0
+; P9LE-NEXT:    std r0, 16(r1)
+; P9LE-NEXT:    stdu r1, -80(r1)
+; P9LE-NEXT:    addi r5, r1, 66
+; P9LE-NEXT:    std r3, 48(r1)
+; P9LE-NEXT:    std r4, 56(r1)
+; P9LE-NEXT:    stw r4, 74(r1)
+; P9LE-NEXT:    stdx r3, 0, r5
+; P9LE-NEXT:    mr r3, r5
+; P9LE-NEXT:    bl callee
+; P9LE-NEXT:    nop
+; P9LE-NEXT:    li r3, 0
+; P9LE-NEXT:    addi r1, r1, 80
+; P9LE-NEXT:    ld r0, 16(r1)
+; P9LE-NEXT:    mtlr r0
+; P9LE-NEXT:    blr
+;
+; P10LE-LABEL: caller_14:
+; P10LE:       # %bb.0: # %entry
+; P10LE-NEXT:    mflr r0
+; P10LE-NEXT:    std r0, 16(r1)
+; P10LE-NEXT:    stdu r1, -80(r1)
+; P10LE-NEXT:    addi r5, r1, 66
+; P10LE-NEXT:    std r3, 48(r1)
+; P10LE-NEXT:    std r4, 56(r1)
+; P10LE-NEXT:    stw r4, 74(r1)
+; P10LE-NEXT:    stdx r3, 0, r5
+; P10LE-NEXT:    mr r3, r5
+; P10LE-NEXT:    bl callee@notoc
+; P10LE-NEXT:    li r3, 0
+; P10LE-NEXT:    addi r1, r1, 80
+; P10LE-NEXT:    ld r0, 16(r1)
+; P10LE-NEXT:    mtlr r0
+; P10LE-NEXT:    blr
+;
+; P8BE-LABEL: caller_14:
+; P8BE:       # %bb.0: # %entry
+; P8BE-NEXT:    mflr r0
+; P8BE-NEXT:    std r0, 16(r1)
+; P8BE-NEXT:    stdu r1, -144(r1)
+; P8BE-NEXT:    lwz r6, 200(r1)
+; P8BE-NEXT:    addi r5, r1, 130
+; P8BE-NEXT:    std r3, 192(r1)
+; P8BE-NEXT:    std r4, 200(r1)
+; P8BE-NEXT:    stdx r3, 0, r5
+; P8BE-NEXT:    mr r3, r5
+; P8BE-NEXT:    stw r6, 138(r1)
+; P8BE-NEXT:    bl callee
+; P8BE-NEXT:    nop
+; P8BE-NEXT:    li r3, 0
+; P8BE-NEXT:    addi r1, r1, 144
+; P8BE-NEXT:    ld r0, 16(r1)
+; P8BE-NEXT:    mtlr r0
+; P8BE-NEXT:    blr
+;
+; P9BE-LABEL: caller_14:
+; P9BE:       # %bb.0: # %entry
+; P9BE-NEXT:    mflr r0
+; P9BE-NEXT:    std r0, 16(r1)
+; P9BE-NEXT:    stdu r1, -144(r1)
+; P9BE-NEXT:    std r4, 200(r1)
+; P9BE-NEXT:    addi r4, r1, 130
+; P9BE-NEXT:    lwz r5, 200(r1)
+; P9BE-NEXT:    std r3, 192(r1)
+; P9BE-NEXT:    stdx r3, 0, r4
+; P9BE-NEXT:    mr r3, r4
+; P9BE-NEXT:    stw r5, 138(r1)
+; P9BE-NEXT:    bl callee
+; P9BE-NEXT:    nop
+; P9BE-NEXT:    li r3, 0
+; P9BE-NEXT:    addi r1, r1, 144
+; P9BE-NEXT:    ld r0, 16(r1)
+; P9BE-NEXT:    mtlr r0
+; P9BE-NEXT:    blr
+;
+; P10BE-LABEL: caller_14:
+; P10BE:       # %bb.0: # %entry
+; P10BE-NEXT:    mflr r0
+; P10BE-NEXT:    std r0, 16(r1)
+; P10BE-NEXT:    stdu r1, -144(r1)
+; P10BE-NEXT:    lwz r5, 200(r1)
+; P10BE-NEXT:    std r4, 200(r1)
+; P10BE-NEXT:    std r3, 192(r1)
+; P10BE-NEXT:    addi r4, r1, 130
+; P10BE-NEXT:    stdx r3, 0, r4
+; P10BE-NEXT:    mr r3, r4
+; P10BE-NEXT:    stw r5, 138(r1)
+; P10BE-NEXT:    bl callee
+; P10BE-NEXT:    nop
+; P10BE-NEXT:    li r3, 0
+; P10BE-NEXT:    addi r1, r1, 144
+; P10BE-NEXT:    ld r0, 16(r1)
+; P10BE-NEXT:    mtlr r0
+; P10BE-NEXT:    blr
+entry:
+  %_param_data = alloca [14 x i8], align 1
+  %.elt0 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 0
+  %.unpack0 = load i8, i8* %.elt0, align 1
+  %.elt1 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 1
+  %.unpack1 = load i8, i8* %.elt1, align 1
+  %.elt2 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 2
+  %.unpack2 = load i8, i8* %.elt2, align 1
+  %.elt3 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 3
+  %.unpack3 = load i8, i8* %.elt3, align 1
+  %.elt4 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 4
+  %.unpack4 = load i8, i8* %.elt4, align 1
+  %.elt5 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 5
+  %.unpack5 = load i8, i8* %.elt5, align 1
+  %.elt6 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 6
+  %.unpack6 = load i8, i8* %.elt6, align 1
+  %.elt7 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 7
+  %.unpack7 = load i8, i8* %.elt7, align 1
+  %.elt8 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 8
+  %.unpack8 = load i8, i8* %.elt8, align 1
+  %.elt9 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 9
+  %.unpack9 = load i8, i8* %.elt9, align 1
+  %.elt10 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 10
+  %.unpack10 = load i8, i8* %.elt10, align 1
+  %.elt11 = getelementptr inbounds [14 x i8], [14 x i8]* %data, i64 0, i64 11
+  %.unpack11 = load i8, i8* %.elt11, align 1
+  %.temp.0.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 0
+  store i8 %.unpack0, i8* %.temp.0.gep, align 1
+  %.temp.1.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 1
+  store i8 %.unpack1, i8* %.temp.1.gep, align 1
+  %.temp.2.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 2
+  store i8 %.unpack2, i8* %.temp.2.gep, align 1
+  %.temp.3.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 3
+  store i8 %.unpack3, i8* %.temp.3.gep, align 1
+  %.temp.4.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 4
+  store i8 %.unpack4, i8* %.temp.4.gep, align 1
+  %.temp.5.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 5
+  store i8 %.unpack5, i8* %.temp.5.gep, align 1
+  %.temp.6.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 6
+  store i8 %.unpack6, i8* %.temp.6.gep, align 1
+  %.temp.7.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 7
+  store i8 %.unpack7, i8* %.temp.7.gep, align 1
+  %.temp.8.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 8
+  store i8 %.unpack8, i8* %.temp.8.gep, align 1
+  %.temp.9.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 9
+  store i8 %.unpack9, i8* %.temp.9.gep, align 1
+  %.temp.10.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 10
+  store i8 %.unpack10, i8* %.temp.10.gep, align 1
+  %.temp.11.gep = getelementptr inbounds [14 x i8], [14 x i8]* %_param_data, i64 0, i64 11
+  store i8 %.unpack11, i8* %.temp.11.gep, align 1
+  call void @callee(i8* nonnull %.temp.0.gep)
+  ret i8 0
+}
+
+define signext i8 @caller_16([16 x i8]* nocapture readonly byval([16 x i8]) %data) #0 {
+; P8LE-LABEL: caller_16:
+; P8LE:       # %bb.0: # %entry
+; P8LE-NEXT:    mflr r0
+; P8LE-NEXT:    std r0, 16(r1)
+; P8LE-NEXT:    stdu r1, -80(r1)
+; P8LE-NEXT:    addi r5, r1, 64
+; P8LE-NEXT:    std r3, 48(r1)
+; P8LE-NEXT:    std r4, 56(r1)
+; P8LE-NEXT:    std r3, 64(r1)
+; P8LE-NEXT:    mr r3, r5
+; P8LE-NEXT:    stw r4, 72(r1)
+; P8LE-NEXT:    bl callee
+; P8LE-NEXT:    nop
+; P8LE-NEXT:    li r3, 0
+; P8LE-NEXT:    addi r1, r1, 80
+; P8LE-NEXT:    ld r0, 16(r1)
+; P8LE-NEXT:    mtlr r0
+; P8LE-NEXT:    blr
+;
+; P9LE-LABEL: caller_16:
+; P9LE:       # %bb.0: # %entry
+; P9LE-NEXT:    mflr r0
+; P9LE-NEXT:    std r0, 16(r1)
+; P9LE-NEXT:    stdu r1, -80(r1)
+; P9LE-NEXT:    addi r5, r1, 64
+; P9LE-NEXT:    std r3, 48(r1)
+; P9LE-NEXT:    std r4, 56(r1)
+; P9LE-NEXT:    stw r4, 72(r1)
+; P9LE-NEXT:    std r3, 64(r1)
+; P9LE-NEXT:    mr r3, r5
+; P9LE-NEXT:    bl callee
+; P9LE-NEXT:    nop
+; P9LE-NEXT:    li r3, 0
+; P9LE-NEXT:    addi r1, r1, 80
+; P9LE-NEXT:    ld r0, 16(r1)
+; P9LE-NEXT:    mtlr r0
+; P9LE-NEXT:    blr
+;
+; P10LE-LABEL: caller_16:
+; P10LE:       # %bb.0: # %entry
+; P10LE-NEXT:    mflr r0
+; P10LE-NEXT:    std r0, 16(r1)
+; P10LE-NEXT:    stdu r1, -80(r1)
+; P10LE-NEXT:    addi r5, r1, 64
+; P10LE-NEXT:    std r3, 48(r1)
+; P10LE-NEXT:    std r4, 56(r1)
+; P10LE-NEXT:    stw r4, 72(r1)
+; P10LE-NEXT:    std r3, 64(r1)
+; P10LE-NEXT:    mr r3, r5
+; P10LE-NEXT:    bl callee@notoc
+; P10LE-NEXT:    li r3, 0
+; P10LE-NEXT:    addi r1, r1, 80
+; P10LE-NEXT:    ld r0, 16(r1)
+; P10LE-NEXT:    mtlr r0
+; P10LE-NEXT:    blr
+;
+; P8BE-LABEL: caller_16:
+; P8BE:       # %bb.0: # %entry
+; P8BE-NEXT:    mflr r0
+; P8BE-NEXT:    std r0, 16(r1)
+; P8BE-NEXT:    stdu r1, -144(r1)
+; P8BE-NEXT:    std r4, 200(r1)
+; P8BE-NEXT:    addi r5, r1, 128
+; P8BE-NEXT:    std r3, 192(r1)
+; P8BE-NEXT:    lwz r4, 200(r1)
+; P8BE-NEXT:    std r3, 128(r1)
+; P8BE-NEXT:    mr r3, r5
+; P8BE-NEXT:    stw r4, 136(r1)
+; P8BE-NEXT:    bl callee
+; P8BE-NEXT:    nop
+; P8BE-NEXT:    li r3, 0
+; P8BE-NEXT:    addi r1, r1, 144
+; P8BE-NEXT:    ld r0, 16(r1)
+; P8BE-NEXT:    mtlr r0
+; P8BE-NEXT:    blr
+;
+; P9BE-LABEL: caller_16:
+; P9BE:       # %bb.0: # %entry
+; P9BE-NEXT:    mflr r0
+; P9BE-NEXT:    std r0, 16(r1)
+; P9BE-NEXT:    stdu r1, -144(r1)
+; P9BE-NEXT:    std r4, 200(r1)
+; P9BE-NEXT:    addi r4, r1, 128
+; P9BE-NEXT:    std r3, 192(r1)
+; P9BE-NEXT:    lwz r5, 200(r1)
+; P9BE-NEXT:    std r3, 128(r1)
+; P9BE-NEXT:    mr r3, r4
+; P9BE-NEXT:    stw r5, 136(r1)
+; P9BE-NEXT:    bl callee
+; P9BE-NEXT:    nop
+; P9BE-NEXT:    li r3, 0
+; P9BE-NEXT:    addi r1, r1, 144
+; P9BE-NEXT:    ld r0, 16(r1)
+; P9BE-NEXT:    mtlr r0
+; P9BE-NEXT:    blr
+;
+; P10BE-LABEL: caller_16:
+; P10BE:       # %bb.0: # %entry
+; P10BE-NEXT:    mflr r0
+; P10BE-NEXT:    std r0, 16(r1)
+; P10BE-NEXT:    stdu r1, -144(r1)
+; P10BE-NEXT:    std r4, 200(r1)
+; P10BE-NEXT:    std r3, 192(r1)
+; P10BE-NEXT:    addi r4, r1, 128
+; P10BE-NEXT:    lwz r5, 200(r1)
+; P10BE-NEXT:    std r3, 128(r1)
+; P10BE-NEXT:    mr r3, r4
+; P10BE-NEXT:    stw r5, 136(r1)
+; P10BE-NEXT:    bl callee
+; P10BE-NEXT:    nop
+; P10BE-NEXT:    li r3, 0
+; P10BE-NEXT:    addi r1, r1, 144
+; P10BE-NEXT:    ld r0, 16(r1)
+; P10BE-NEXT:    mtlr r0
+; P10BE-NEXT:    blr
+entry:
+  %_param_data = alloca [16 x i8], align 1
+  %.elt0 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 0
+  %.unpack0 = load i8, i8* %.elt0, align 1
+  %.elt1 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 1
+  %.unpack1 = load i8, i8* %.elt1, align 1
+  %.elt2 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 2
+  %.unpack2 = load i8, i8* %.elt2, align 1
+  %.elt3 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 3
+  %.unpack3 = load i8, i8* %.elt3, align 1
+  %.elt4 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 4
+  %.unpack4 = load i8, i8* %.elt4, align 1
+  %.elt5 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 5
+  %.unpack5 = load i8, i8* %.elt5, align 1
+  %.elt6 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 6
+  %.unpack6 = load i8, i8* %.elt6, align 1
+  %.elt7 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 7
+  %.unpack7 = load i8, i8* %.elt7, align 1
+  %.elt8 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 8
+  %.unpack8 = load i8, i8* %.elt8, align 1
+  %.elt9 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 9
+  %.unpack9 = load i8, i8* %.elt9, align 1
+  %.elt10 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 10
+  %.unpack10 = load i8, i8* %.elt10, align 1
+  %.elt11 = getelementptr inbounds [16 x i8], [16 x i8]* %data, i64 0, i64 11
+  %.unpack11 = load i8, i8* %.elt11, align 1
+  %.temp.0.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 0
+  store i8 %.unpack0, i8* %.temp.0.gep, align 1
+  %.temp.1.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 1
+  store i8 %.unpack1, i8* %.temp.1.gep, align 1
+  %.temp.2.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 2
+  store i8 %.unpack2, i8* %.temp.2.gep, align 1
+  %.temp.3.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 3
+  store i8 %.unpack3, i8* %.temp.3.gep, align 1
+  %.temp.4.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 4
+  store i8 %.unpack4, i8* %.temp.4.gep, align 1
+  %.temp.5.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 5
+  store i8 %.unpack5, i8* %.temp.5.gep, align 1
+  %.temp.6.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 6
+  store i8 %.unpack6, i8* %.temp.6.gep, align 1
+  %.temp.7.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 7
+  store i8 %.unpack7, i8* %.temp.7.gep, align 1
+  %.temp.8.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 8
+  store i8 %.unpack8, i8* %.temp.8.gep, align 1
+  %.temp.9.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 9
+  store i8 %.unpack9, i8* %.temp.9.gep, align 1
+  %.temp.10.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 10
+  store i8 %.unpack10, i8* %.temp.10.gep, align 1
+  %.temp.11.gep = getelementptr inbounds [16 x i8], [16 x i8]* %_param_data, i64 0, i64 11
+  store i8 %.unpack11, i8* %.temp.11.gep, align 1
+  call void @callee(i8* nonnull %.temp.0.gep)
+  ret i8 0
+}
+
+define signext i8 @caller_18([18 x i8]* nocapture readonly byval([18 x i8]) %data) #0 {
+; P8LE-LABEL: caller_18:
+; P8LE:       # %bb.0: # %entry
+; P8LE-NEXT:    mflr r0
+; P8LE-NEXT:    std r0, 16(r1)
+; P8LE-NEXT:    stdu r1, -96(r1)
+; P8LE-NEXT:    addi r6, r1, 78
+; P8LE-NEXT:    std r3, 48(r1)
+; P8LE-NEXT:    std r5, 64(r1)
+; P8LE-NEXT:    std r4, 56(r1)
+; P8LE-NEXT:    stdx r3, 0, r6
+; P8LE-NEXT:    mr r3, r6
+; P8LE-NEXT:    stw r4, 86(r1)
+; P8LE-NEXT:    bl callee
+; P8LE-NEXT:    nop
+; P8LE-NEXT:    li r3, 0
+; P8LE-NEXT:    addi r1, r1, 96
+; P8LE-NEXT:    ld r0, 16(r1)
+; P8LE-NEXT:    mtlr r0
+; P8LE-NEXT:    blr
+;
+; P9LE-LABEL: caller_18:
+; P9LE:       # %bb.0: # %entry
+; P9LE-NEXT:    mflr r0
+; P9LE-NEXT:    std r0, 16(r1)
+; P9LE-NEXT:    stdu r1, -96(r1)
+; P9LE-NEXT:    std r5, 64(r1)
+; P9LE-NEXT:    addi r5, r1, 78
+; P9LE-NEXT:    std r3, 48(r1)
+; P9LE-NEXT:    std r4, 56(r1)
+; P9LE-NEXT:    stw r4, 86(r1)
+; P9LE-NEXT:    stdx r3, 0, r5
+; P9LE-NEXT:    mr r3, r5
+; P9LE-NEXT:    bl callee
+; P9LE-NEXT:    nop
+; P9LE-NEXT:    li r3, 0
+; P9LE-NEXT:    addi r1, r1, 96
+; P9LE-NEXT:    ld r0, 16(r1)
+; P9LE-NEXT:    mtlr r0
+; P9LE-NEXT:    blr
+;
+; P10LE-LABEL: caller_18:
+; P10LE:       # %bb.0: # %entry
+; P10LE-NEXT:    mflr r0
+; P10LE-NEXT:    std r0, 16(r1)
+; P10LE-NEXT:    stdu r1, -96(r1)
+; P10LE-NEXT:    std r5, 64(r1)
+; P10LE-NEXT:    addi r5, r1, 78
+; P10LE-NEXT:    std r3, 48(r1)
+; P10LE-NEXT:    std r4, 56(r1)
+; P10LE-NEXT:    stw r4, 86(r1)
+; P10LE-NEXT:    stdx r3, 0, r5
+; P10LE-NEXT:    mr r3, r5
+; P10LE-NEXT:    bl callee@notoc
+; P10LE-NEXT:    li r3, 0
+; P10LE-NEXT:    addi r1, r1, 96
+; P10LE-NEXT:    ld r0, 16(r1)
+; P10LE-NEXT:    mtlr r0
+; P10LE-NEXT:    blr
+;
+; P8BE-LABEL: caller_18:
+; P8BE:       # %bb.0: # %entry
+; P8BE-NEXT:    mflr r0
+; P8BE-NEXT:    std r0, 16(r1)
+; P8BE-NEXT:    stdu r1, -144(r1)
+; P8BE-NEXT:    std r4, 200(r1)
+; P8BE-NEXT:    addi r6, r1, 126
+; P8BE-NEXT:    std r3, 192(r1)
+; P8BE-NEXT:    std r5, 208(r1)
+; P8BE-NEXT:    lwz r4, 200(r1)
+; P8BE-NEXT:    stdx r3, 0, r6
+; P8BE-NEXT:    mr r3, r6
+; P8BE-NEXT:    stw r4, 134(r1)
+; P8BE-NEXT:    bl callee
+; P8BE-NEXT:    nop
+; P8BE-NEXT:    li r3, 0
+; P8BE-NEXT:    addi r1, r1, 144
+; P8BE-NEXT:    ld r0, 16(r1)
+; P8BE-NEXT:    mtlr r0
+; P8BE-NEXT:    blr
+;
+; P9BE-LABEL: caller_18:
+; P9BE:       # %bb.0: # %entry
+; P9BE-NEXT:    mflr r0
+; P9BE-NEXT:    std r0, 16(r1)
+; P9BE-NEXT:    stdu r1, -144(r1)
+; P9BE-NEXT:    std r4, 200(r1)
+; P9BE-NEXT:    addi r4, r1, 126
+; P9BE-NEXT:    std r5, 208(r1)
+; P9BE-NEXT:    lwz r5, 200(r1)
+; P9BE-NEXT:    std r3, 192(r1)
+; P9BE-NEXT:    stdx r3, 0, r4
+; P9BE-NEXT:    mr r3, r4
+; P9BE-NEXT:    stw r5, 134(r1)
+; P9BE-NEXT:    bl callee
+; P9BE-NEXT:    nop
+; P9BE-NEXT:    li r3, 0
+; P9BE-NEXT:    addi r1, r1, 144
+; P9BE-NEXT:    ld r0, 16(r1)
+; P9BE-NEXT:    mtlr r0
+; P9BE-NEXT:    blr
+;
+; P10BE-LABEL: caller_18:
+; P10BE:       # %bb.0: # %entry
+; P10BE-NEXT:    mflr r0
+; P10BE-NEXT:    std r0, 16(r1)
+; P10BE-NEXT:    stdu r1, -144(r1)
+; P10BE-NEXT:    std r4, 200(r1)
+; P10BE-NEXT:    std r3, 192(r1)
+; P10BE-NEXT:    addi r4, r1, 126
+; P10BE-NEXT:    std r5, 208(r1)
+; P10BE-NEXT:    lwz r5, 200(r1)
+; P10BE-NEXT:    stdx r3, 0, r4
+; P10BE-NEXT:    mr r3, r4
+; P10BE-NEXT:    stw r5, 134(r1)
+; P10BE-NEXT:    bl callee
+; P10BE-NEXT:    nop
+; P10BE-NEXT:    li r3, 0
+; P10BE-NEXT:    addi r1, r1, 144
+; P10BE-NEXT:    ld r0, 16(r1)
+; P10BE-NEXT:    mtlr r0
+; P10BE-NEXT:    blr
+entry:
+  %_param_data = alloca [18 x i8], align 1
+  %.elt0 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 0
+  %.unpack0 = load i8, i8* %.elt0, align 1
+  %.elt1 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 1
+  %.unpack1 = load i8, i8* %.elt1, align 1
+  %.elt2 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 2
+  %.unpack2 = load i8, i8* %.elt2, align 1
+  %.elt3 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 3
+  %.unpack3 = load i8, i8* %.elt3, align 1
+  %.elt4 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 4
+  %.unpack4 = load i8, i8* %.elt4, align 1
+  %.elt5 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 5
+  %.unpack5 = load i8, i8* %.elt5, align 1
+  %.elt6 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 6
+  %.unpack6 = load i8, i8* %.elt6, align 1
+  %.elt7 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 7
+  %.unpack7 = load i8, i8* %.elt7, align 1
+  %.elt8 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 8
+  %.unpack8 = load i8, i8* %.elt8, align 1
+  %.elt9 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 9
+  %.unpack9 = load i8, i8* %.elt9, align 1
+  %.elt10 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 10
+  %.unpack10 = load i8, i8* %.elt10, align 1
+  %.elt11 = getelementptr inbounds [18 x i8], [18 x i8]* %data, i64 0, i64 11
+  %.unpack11 = load i8, i8* %.elt11, align 1
+  %.temp.0.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 0
+  store i8 %.unpack0, i8* %.temp.0.gep, align 1
+  %.temp.1.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 1
+  store i8 %.unpack1, i8* %.temp.1.gep, align 1
+  %.temp.2.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 2
+  store i8 %.unpack2, i8* %.temp.2.gep, align 1
+  %.temp.3.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 3
+  store i8 %.unpack3, i8* %.temp.3.gep, align 1
+  %.temp.4.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 4
+  store i8 %.unpack4, i8* %.temp.4.gep, align 1
+  %.temp.5.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 5
+  store i8 %.unpack5, i8* %.temp.5.gep, align 1
+  %.temp.6.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 6
+  store i8 %.unpack6, i8* %.temp.6.gep, align 1
+  %.temp.7.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 7
+  store i8 %.unpack7, i8* %.temp.7.gep, align 1
+  %.temp.8.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 8
+  store i8 %.unpack8, i8* %.temp.8.gep, align 1
+  %.temp.9.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 9
+  store i8 %.unpack9, i8* %.temp.9.gep, align 1
+  %.temp.10.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 10
+  store i8 %.unpack10, i8* %.temp.10.gep, align 1
+  %.temp.11.gep = getelementptr inbounds [18 x i8], [18 x i8]* %_param_data, i64 0, i64 11
+  store i8 %.unpack11, i8* %.temp.11.gep, align 1
+  call void @callee(i8* nonnull %.temp.0.gep)
+  ret i8 0
+}
+
+
+
+declare void @callee(i8*) local_unnamed_addr #0
+
+attributes #0 = { nounwind }
+