[VE] Add comprehensive stackframe tests

[kaz7]

Summary

Add comprehensive stackframe regression tests as a preparation of
VEFrameLowering.cpp optimizations.

llvm/test/CodeGen/VE/Scalar/stackframe_align.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=ve | FileCheck %s
+; RUN: llc < %s -mtriple=ve --frame-pointer=all \
+; RUN:     | FileCheck %s --check-prefix=CHECKFP
+
+;;; Check stack frame allocation with static and dynamic stack object with
+;;; alignments as a test of getFrameIndexReference().
+
+;; Allocated buffer places from 9 to 15 bytes in 16 bytes local vars area.
+
+; Function Attrs: nounwind
+define i8* @test_frame7(i8* %0) {
+; CHECK-LABEL: test_frame7:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -192
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB0_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB0_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 185(, %s11)
+; CHECK-NEXT:    st1b %s1, 185(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+;
+; CHECKFP-LABEL: test_frame7:
+; CHECKFP:       # %bb.0:
+; CHECKFP-NEXT:    st %s9, (, %s11)
+; CHECKFP-NEXT:    st %s10, 8(, %s11)
+; CHECKFP-NEXT:    st %s15, 24(, %s11)
+; CHECKFP-NEXT:    st %s16, 32(, %s11)
+; CHECKFP-NEXT:    or %s9, 0, %s11
+; CHECKFP-NEXT:    lea %s13, -192
+; CHECKFP-NEXT:    and %s13, %s13, (32)0
+; CHECKFP-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB0_2
+; CHECKFP-NEXT:  # %bb.1:
+; CHECKFP-NEXT:    ld %s61, 24(, %s14)
+; CHECKFP-NEXT:    or %s62, 0, %s0
+; CHECKFP-NEXT:    lea %s63, 315
+; CHECKFP-NEXT:    shm.l %s63, (%s61)
+; CHECKFP-NEXT:    shm.l %s8, 8(%s61)
+; CHECKFP-NEXT:    shm.l %s11, 16(%s61)
+; CHECKFP-NEXT:    monc
+; CHECKFP-NEXT:    or %s0, 0, %s62
+; CHECKFP-NEXT:  .LBB0_2:
+; CHECKFP-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECKFP-NEXT:    lea %s0, -7(, %s9)
+; CHECKFP-NEXT:    st1b %s1, -7(, %s9)
+; CHECKFP-NEXT:    or %s11, 0, %s9
+; CHECKFP-NEXT:    ld %s16, 32(, %s11)
+; CHECKFP-NEXT:    ld %s15, 24(, %s11)
+; CHECKFP-NEXT:    ld %s10, 8(, %s11)
+; CHECKFP-NEXT:    ld %s9, (, %s11)
+; CHECKFP-NEXT:    b.l.t (, %s10)
+  %2 = alloca [7 x i8], align 1
+  %3 = getelementptr inbounds [7 x i8], [7 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+;; Allocated buffer is aligned by 8, so it places from 8 to 14 bytes in 16
+;; bytes local vars area.
+
+; Function Attrs: nounwind
+define i8* @test_frame7_align8(i8* %0) {
+; CHECK-LABEL: test_frame7_align8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -192
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB1_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB1_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 184(, %s11)
+; CHECK-NEXT:    st1b %s1, 184(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+;
+; CHECKFP-LABEL: test_frame7_align8:
+; CHECKFP:       # %bb.0:
+; CHECKFP-NEXT:    st %s9, (, %s11)
+; CHECKFP-NEXT:    st %s10, 8(, %s11)
+; CHECKFP-NEXT:    st %s15, 24(, %s11)
+; CHECKFP-NEXT:    st %s16, 32(, %s11)
+; CHECKFP-NEXT:    or %s9, 0, %s11
+; CHECKFP-NEXT:    lea %s13, -192
+; CHECKFP-NEXT:    and %s13, %s13, (32)0
+; CHECKFP-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB1_2
+; CHECKFP-NEXT:  # %bb.1:
+; CHECKFP-NEXT:    ld %s61, 24(, %s14)
+; CHECKFP-NEXT:    or %s62, 0, %s0
+; CHECKFP-NEXT:    lea %s63, 315
+; CHECKFP-NEXT:    shm.l %s63, (%s61)
+; CHECKFP-NEXT:    shm.l %s8, 8(%s61)
+; CHECKFP-NEXT:    shm.l %s11, 16(%s61)
+; CHECKFP-NEXT:    monc
+; CHECKFP-NEXT:    or %s0, 0, %s62
+; CHECKFP-NEXT:  .LBB1_2:
+; CHECKFP-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECKFP-NEXT:    lea %s0, -8(, %s9)
+; CHECKFP-NEXT:    st1b %s1, -8(, %s9)
+; CHECKFP-NEXT:    or %s11, 0, %s9
+; CHECKFP-NEXT:    ld %s16, 32(, %s11)
+; CHECKFP-NEXT:    ld %s15, 24(, %s11)
+; CHECKFP-NEXT:    ld %s10, 8(, %s11)
+; CHECKFP-NEXT:    ld %s9, (, %s11)
+; CHECKFP-NEXT:    b.l.t (, %s10)
+  %2 = alloca [7 x i8], align 8
+  %3 = getelementptr inbounds [7 x i8], [7 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+;; Allocated buffer is aligned by 16, so it places from 0 to 15 bytes in 16
+;; bytes local vars area.
+
+; Function Attrs: nounwind
+define i8* @test_frame16_align16(i8* %0) {
+; CHECK-LABEL: test_frame16_align16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -192
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB2_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB2_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 176(, %s11)
+; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+;
+; CHECKFP-LABEL: test_frame16_align16:
+; CHECKFP:       # %bb.0:
+; CHECKFP-NEXT:    st %s9, (, %s11)
+; CHECKFP-NEXT:    st %s10, 8(, %s11)
+; CHECKFP-NEXT:    st %s15, 24(, %s11)
+; CHECKFP-NEXT:    st %s16, 32(, %s11)
+; CHECKFP-NEXT:    or %s9, 0, %s11
+; CHECKFP-NEXT:    lea %s13, -192
+; CHECKFP-NEXT:    and %s13, %s13, (32)0
+; CHECKFP-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB2_2
+; CHECKFP-NEXT:  # %bb.1:
+; CHECKFP-NEXT:    ld %s61, 24(, %s14)
+; CHECKFP-NEXT:    or %s62, 0, %s0
+; CHECKFP-NEXT:    lea %s63, 315
+; CHECKFP-NEXT:    shm.l %s63, (%s61)
+; CHECKFP-NEXT:    shm.l %s8, 8(%s61)
+; CHECKFP-NEXT:    shm.l %s11, 16(%s61)
+; CHECKFP-NEXT:    monc
+; CHECKFP-NEXT:    or %s0, 0, %s62
+; CHECKFP-NEXT:  .LBB2_2:
+; CHECKFP-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECKFP-NEXT:    lea %s0, -16(, %s9)
+; CHECKFP-NEXT:    st1b %s1, -16(, %s9)
+; CHECKFP-NEXT:    or %s11, 0, %s9
+; CHECKFP-NEXT:    ld %s16, 32(, %s11)
+; CHECKFP-NEXT:    ld %s15, 24(, %s11)
+; CHECKFP-NEXT:    ld %s10, 8(, %s11)
+; CHECKFP-NEXT:    ld %s9, (, %s11)
+; CHECKFP-NEXT:    b.l.t (, %s10)
+  %2 = alloca [16 x i8], align 16
+  %3 = getelementptr inbounds [16 x i8], [16 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+;; Allocated buffer is aligned by 32, so it places from 0 to 15 bytes in 48
+;; bytes local vars area.  Or it places from 192 (aligned to 32 bytes) to
+;; 207 bytes in 224 + alpha allocated local vars area.
+
+; Function Attrs: nounwind
+define i8* @test_frame16_align32(i8* %0) {
+; CHECK-LABEL: test_frame16_align32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -224
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    and %s11, %s11, (59)1
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB3_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB3_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 192(, %s11)
+; CHECK-NEXT:    st1b %s1, 192(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+;
+; CHECKFP-LABEL: test_frame16_align32:
+; CHECKFP:       # %bb.0:
+; CHECKFP-NEXT:    st %s9, (, %s11)
+; CHECKFP-NEXT:    st %s10, 8(, %s11)
+; CHECKFP-NEXT:    st %s15, 24(, %s11)
+; CHECKFP-NEXT:    st %s16, 32(, %s11)
+; CHECKFP-NEXT:    or %s9, 0, %s11
+; CHECKFP-NEXT:    lea %s13, -224
+; CHECKFP-NEXT:    and %s13, %s13, (32)0
+; CHECKFP-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECKFP-NEXT:    and %s11, %s11, (59)1
+; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB3_2
+; CHECKFP-NEXT:  # %bb.1:
+; CHECKFP-NEXT:    ld %s61, 24(, %s14)
+; CHECKFP-NEXT:    or %s62, 0, %s0
+; CHECKFP-NEXT:    lea %s63, 315
+; CHECKFP-NEXT:    shm.l %s63, (%s61)
+; CHECKFP-NEXT:    shm.l %s8, 8(%s61)
+; CHECKFP-NEXT:    shm.l %s11, 16(%s61)
+; CHECKFP-NEXT:    monc
+; CHECKFP-NEXT:    or %s0, 0, %s62
+; CHECKFP-NEXT:  .LBB3_2:
+; CHECKFP-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECKFP-NEXT:    lea %s0, 192(, %s11)
+; CHECKFP-NEXT:    st1b %s1, 192(, %s11)
+; CHECKFP-NEXT:    or %s11, 0, %s9
+; CHECKFP-NEXT:    ld %s16, 32(, %s11)
+; CHECKFP-NEXT:    ld %s15, 24(, %s11)
+; CHECKFP-NEXT:    ld %s10, 8(, %s11)
+; CHECKFP-NEXT:    ld %s9, (, %s11)
+; CHECKFP-NEXT:    b.l.t (, %s10)
+  %2 = alloca [16 x i8], align 32
+  %3 = getelementptr inbounds [16 x i8], [16 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+;; Allocated buffer is aligned by 32, so it places from 0 to 31 bytes in 48
+;; + alpha bytes local vars area, or it places from 192 (32 bytes aligned 176)
+;; to 223 in 224 + alpha bytes local vars area..
+
+; Function Attrs: nounwind
+define i8* @test_frame32_align32(i8* %0) {
+; CHECK-LABEL: test_frame32_align32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -224
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    and %s11, %s11, (59)1
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB4_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB4_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 192(, %s11)
+; CHECK-NEXT:    st1b %s1, 192(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+;
+; CHECKFP-LABEL: test_frame32_align32:
+; CHECKFP:       # %bb.0:
+; CHECKFP-NEXT:    st %s9, (, %s11)
+; CHECKFP-NEXT:    st %s10, 8(, %s11)
+; CHECKFP-NEXT:    st %s15, 24(, %s11)
+; CHECKFP-NEXT:    st %s16, 32(, %s11)
+; CHECKFP-NEXT:    or %s9, 0, %s11
+; CHECKFP-NEXT:    lea %s13, -224
+; CHECKFP-NEXT:    and %s13, %s13, (32)0
+; CHECKFP-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECKFP-NEXT:    and %s11, %s11, (59)1
+; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB4_2
+; CHECKFP-NEXT:  # %bb.1:
+; CHECKFP-NEXT:    ld %s61, 24(, %s14)
+; CHECKFP-NEXT:    or %s62, 0, %s0
+; CHECKFP-NEXT:    lea %s63, 315
+; CHECKFP-NEXT:    shm.l %s63, (%s61)
+; CHECKFP-NEXT:    shm.l %s8, 8(%s61)
+; CHECKFP-NEXT:    shm.l %s11, 16(%s61)
+; CHECKFP-NEXT:    monc
+; CHECKFP-NEXT:    or %s0, 0, %s62
+; CHECKFP-NEXT:  .LBB4_2:
+; CHECKFP-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECKFP-NEXT:    lea %s0, 192(, %s11)
+; CHECKFP-NEXT:    st1b %s1, 192(, %s11)
+; CHECKFP-NEXT:    or %s11, 0, %s9
+; CHECKFP-NEXT:    ld %s16, 32(, %s11)
+; CHECKFP-NEXT:    ld %s15, 24(, %s11)
+; CHECKFP-NEXT:    ld %s10, 8(, %s11)
+; CHECKFP-NEXT:    ld %s9, (, %s11)
+; CHECKFP-NEXT:    b.l.t (, %s10)
+  %2 = alloca [32 x i8], align 32
+  %3 = getelementptr inbounds [32 x i8], [32 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+;; Dynamically allocated buffer is aligned by 16, so it places from 0 to 31
+;; bytes in allocated area, or it places from 240 (32 bytes aligned 176+64)
+;; to 271 in allocated area (actually it places not newly allocated area
+;; but in somewhere between newly allocated area and allocated area at the
+;; prologue since VE ABI requires the reserved area at the top of stack).
+
+;; FIXME: (size+15)/16*16 is not enough.
+
+; Function Attrs: nounwind
+define i8* @test_frame_dynalign16(i8* %0, i64 %1) {
+; CHECK-LABEL: test_frame_dynalign16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -240
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB5_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB5_2:
+; CHECK-NEXT:    or %s2, 0, %s0
+; CHECK-NEXT:    lea %s0, 15(, %s1)
+; CHECK-NEXT:    and %s0, -16, %s0
+; CHECK-NEXT:    lea %s1, __ve_grow_stack@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s12, __ve_grow_stack@hi(, %s1)
+; CHECK-NEXT:    bsic %s10, (, %s12)
+; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    ld1b.zx %s1, (, %s2)
+; CHECK-NEXT:    st1b %s1, (, %s0)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+;
+; CHECKFP-LABEL: test_frame_dynalign16:
+; CHECKFP:       # %bb.0:
+; CHECKFP-NEXT:    st %s9, (, %s11)
+; CHECKFP-NEXT:    st %s10, 8(, %s11)
+; CHECKFP-NEXT:    st %s15, 24(, %s11)
+; CHECKFP-NEXT:    st %s16, 32(, %s11)
+; CHECKFP-NEXT:    or %s9, 0, %s11
+; CHECKFP-NEXT:    lea %s13, -240
+; CHECKFP-NEXT:    and %s13, %s13, (32)0
+; CHECKFP-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB5_2
+; CHECKFP-NEXT:  # %bb.1:
+; CHECKFP-NEXT:    ld %s61, 24(, %s14)
+; CHECKFP-NEXT:    or %s62, 0, %s0
+; CHECKFP-NEXT:    lea %s63, 315
+; CHECKFP-NEXT:    shm.l %s63, (%s61)
+; CHECKFP-NEXT:    shm.l %s8, 8(%s61)
+; CHECKFP-NEXT:    shm.l %s11, 16(%s61)
+; CHECKFP-NEXT:    monc
+; CHECKFP-NEXT:    or %s0, 0, %s62
+; CHECKFP-NEXT:  .LBB5_2:
+; CHECKFP-NEXT:    or %s2, 0, %s0
+; CHECKFP-NEXT:    lea %s0, 15(, %s1)
+; CHECKFP-NEXT:    and %s0, -16, %s0
+; CHECKFP-NEXT:    lea %s1, __ve_grow_stack@lo
+; CHECKFP-NEXT:    and %s1, %s1, (32)0
+; CHECKFP-NEXT:    lea.sl %s12, __ve_grow_stack@hi(, %s1)
+; CHECKFP-NEXT:    bsic %s10, (, %s12)
+; CHECKFP-NEXT:    lea %s0, 240(, %s11)
+; CHECKFP-NEXT:    ld1b.zx %s1, (, %s2)
+; CHECKFP-NEXT:    st1b %s1, (, %s0)
+; CHECKFP-NEXT:    or %s11, 0, %s9
+; CHECKFP-NEXT:    ld %s16, 32(, %s11)
+; CHECKFP-NEXT:    ld %s15, 24(, %s11)
+; CHECKFP-NEXT:    ld %s10, 8(, %s11)
+; CHECKFP-NEXT:    ld %s9, (, %s11)
+; CHECKFP-NEXT:    b.l.t (, %s10)
+  %3 = alloca i8, i64 %1, align 16
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+;; This test allocates static buffer with 16 bytes align and dynamic buffer
+;; with 32 bytes align.  In LLVM, stack frame is always aligned to 32 bytes
+;; (bigger one).  So, LLVM allocates 176 (RSA) + 64 (call site) + 32 (32 bytes
+;; aligned 16 bytes data) + 16 (pad to align) if FP is not eliminated.
+;; Statically allocated buffer is aligned to 16, so it places from 16 to 31
+;; bytes from BP in 32 + alpha bytes local vars area, or it places from 272
+;; to 287 bytes from BP in 288 + alpha bytes local vars area.
+;; Dynamically allocated buffer is aligned to 32, so it places from aligned
+;; address between 240 and 271 from SP.
+
+; Function Attrs: nounwind
+define i8* @test_frame16_align16_dynalign32(i8* %0, i64 %n) {
+; CHECK-LABEL: test_frame16_align16_dynalign32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    st %s17, 40(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -288
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    and %s11, %s11, (59)1
+; CHECK-NEXT:    or %s17, 0, %s11
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB6_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB6_2:
+; CHECK-NEXT:    ld1b.zx %s0, (, %s0)
+; CHECK-NEXT:    st1b %s0, 272(, %s17)
+; CHECK-NEXT:    lea %s0, 15(, %s1)
+; CHECK-NEXT:    and %s0, -16, %s0
+; CHECK-NEXT:    lea %s1, __ve_grow_stack_align@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s12, __ve_grow_stack_align@hi(, %s1)
+; CHECK-NEXT:    or %s1, -32, (0)1
+; CHECK-NEXT:    bsic %s10, (, %s12)
+; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    ld1b.zx %s1, 272(, %s17)
+; CHECK-NEXT:    lea %s0, 31(, %s0)
+; CHECK-NEXT:    and %s0, -32, %s0
+; CHECK-NEXT:    st1b %s1, (, %s0)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s17, 40(, %s11)
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+;
+; CHECKFP-LABEL: test_frame16_align16_dynalign32:
+; CHECKFP:       # %bb.0:
+; CHECKFP-NEXT:    st %s9, (, %s11)
+; CHECKFP-NEXT:    st %s10, 8(, %s11)
+; CHECKFP-NEXT:    st %s15, 24(, %s11)
+; CHECKFP-NEXT:    st %s16, 32(, %s11)
+; CHECKFP-NEXT:    st %s17, 40(, %s11)
+; CHECKFP-NEXT:    or %s9, 0, %s11
+; CHECKFP-NEXT:    lea %s13, -288
+; CHECKFP-NEXT:    and %s13, %s13, (32)0
+; CHECKFP-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECKFP-NEXT:    and %s11, %s11, (59)1
+; CHECKFP-NEXT:    or %s17, 0, %s11
+; CHECKFP-NEXT:    brge.l.t %s11, %s8, .LBB6_2
+; CHECKFP-NEXT:  # %bb.1:
+; CHECKFP-NEXT:    ld %s61, 24(, %s14)
+; CHECKFP-NEXT:    or %s62, 0, %s0
+; CHECKFP-NEXT:    lea %s63, 315
+; CHECKFP-NEXT:    shm.l %s63, (%s61)
+; CHECKFP-NEXT:    shm.l %s8, 8(%s61)
+; CHECKFP-NEXT:    shm.l %s11, 16(%s61)
+; CHECKFP-NEXT:    monc
+; CHECKFP-NEXT:    or %s0, 0, %s62
+; CHECKFP-NEXT:  .LBB6_2:
+; CHECKFP-NEXT:    ld1b.zx %s0, (, %s0)
+; CHECKFP-NEXT:    st1b %s0, 272(, %s17)
+; CHECKFP-NEXT:    lea %s0, 15(, %s1)
+; CHECKFP-NEXT:    and %s0, -16, %s0
+; CHECKFP-NEXT:    lea %s1, __ve_grow_stack_align@lo
+; CHECKFP-NEXT:    and %s1, %s1, (32)0
+; CHECKFP-NEXT:    lea.sl %s12, __ve_grow_stack_align@hi(, %s1)
+; CHECKFP-NEXT:    or %s1, -32, (0)1
+; CHECKFP-NEXT:    bsic %s10, (, %s12)
+; CHECKFP-NEXT:    lea %s0, 240(, %s11)
+; CHECKFP-NEXT:    ld1b.zx %s1, 272(, %s17)
+; CHECKFP-NEXT:    lea %s0, 31(, %s0)
+; CHECKFP-NEXT:    and %s0, -32, %s0
+; CHECKFP-NEXT:    st1b %s1, (, %s0)
+; CHECKFP-NEXT:    or %s11, 0, %s9
+; CHECKFP-NEXT:    ld %s17, 40(, %s11)
+; CHECKFP-NEXT:    ld %s16, 32(, %s11)
+; CHECKFP-NEXT:    ld %s15, 24(, %s11)
+; CHECKFP-NEXT:    ld %s10, 8(, %s11)
+; CHECKFP-NEXT:    ld %s9, (, %s11)
+; CHECKFP-NEXT:    b.l.t (, %s10)
+  %2 = alloca [16 x i8], align 16
+  %3 = getelementptr inbounds [16 x i8], [16 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  %5 = alloca i8, i64 %n, align 32
+  %6 = load i8, i8* %3, align 1
+  store i8 %6, i8* %5, align 1
+  ret i8* %5
+}
+

llvm/test/CodeGen/VE/Scalar/stackframe_call.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=ve | FileCheck %s
-; RUN: llc < %s -mtriple=ve -relocation-model=pic | FileCheck %s --check-prefix=PIC
+; RUN: llc < %s -mtriple=ve -relocation-model=pic \
+; RUN:     | FileCheck %s --check-prefix=PIC
 
-;; Check stack frame allocation of a function which calls other functions
+;;; Check stack frame allocation of a function which calls other functions
+;;; under following conditions and combinations of them:
+;;;   - access variable or not
+;;;   - no stack object, a stack object using BP, or a stack object not using BP
+;;;   - isPositionIndependent or not
 
-; Function Attrs: norecurse nounwind readnone
-define signext i32 @test_frame0(i32 signext %0) {
+@data = external global i8, align 1
+
+; Function Attrs: nounwind
+define i8* @test_frame0(i8* %0, i8* %1) {
 ; CHECK-LABEL: test_frame0:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    adds.w.sx %s0, 3, %s0
-; CHECK-NEXT:    adds.w.sx %s0, %s0, (0)1
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -240
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB0_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB0_2:
+; CHECK-NEXT:    lea %s2, fun@lo
+; CHECK-NEXT:    and %s2, %s2, (32)0
+; CHECK-NEXT:    lea.sl %s12, fun@hi(, %s2)
+; CHECK-NEXT:    bsic %s10, (, %s12)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
 ; PIC-LABEL: test_frame0:
 ; PIC:       # %bb.0:
-; PIC-NEXT:    adds.w.sx %s0, 3, %s0
-; PIC-NEXT:    adds.w.sx %s0, %s0, (0)1
+; PIC-NEXT:    st %s9, (, %s11)
+; PIC-NEXT:    st %s10, 8(, %s11)
+; PIC-NEXT:    st %s15, 24(, %s11)
+; PIC-NEXT:    st %s16, 32(, %s11)
+; PIC-NEXT:    or %s9, 0, %s11
+; PIC-NEXT:    lea %s13, -240
+; PIC-NEXT:    and %s13, %s13, (32)0
+; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; PIC-NEXT:    brge.l.t %s11, %s8, .LBB0_2
+; PIC-NEXT:  # %bb.1:
+; PIC-NEXT:    ld %s61, 24(, %s14)
+; PIC-NEXT:    or %s62, 0, %s0
+; PIC-NEXT:    lea %s63, 315
+; PIC-NEXT:    shm.l %s63, (%s61)
+; PIC-NEXT:    shm.l %s8, 8(%s61)
+; PIC-NEXT:    shm.l %s11, 16(%s61)
+; PIC-NEXT:    monc
+; PIC-NEXT:    or %s0, 0, %s62
+; PIC-NEXT:  .LBB0_2:
+; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
+; PIC-NEXT:    and %s15, %s15, (32)0
+; PIC-NEXT:    sic %s16
+; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
+; PIC-NEXT:    lea %s12, fun@plt_lo(-24)
+; PIC-NEXT:    and %s12, %s12, (32)0
+; PIC-NEXT:    sic %s16
+; PIC-NEXT:    lea.sl %s12, fun@plt_hi(%s16, %s12)
+; PIC-NEXT:    bsic %s10, (, %s12)
+; PIC-NEXT:    or %s11, 0, %s9
+; PIC-NEXT:    ld %s16, 32(, %s11)
+; PIC-NEXT:    ld %s15, 24(, %s11)
+; PIC-NEXT:    ld %s10, 8(, %s11)
+; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = add nsw i32 %0, 3
-  ret i32 %2
+  %3 = tail call i8* @fun(i8* %0, i8* %1)
+  ret i8* %3
 }
 
+declare i8* @fun(i8*, i8*)
+
 ; Function Attrs: nounwind
 define i8* @test_frame32(i8* %0) {
 ; CHECK-LABEL: test_frame32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
; PIC-NEXT:    b.l.t (, %s10)
   %4 = call i8* @fun(i8* nonnull %3, i8* %0)
   call void @llvm.lifetime.end.p0i8(i64 32, i8* nonnull %3)
   ret i8* %4
 }
 
 ; Function Attrs: argmemonly nofree nosync nounwind willreturn
 declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture)
 
-declare i8* @fun(i8*, i8*)
-
 ; Function Attrs: argmemonly nofree nosync nounwind willreturn
 declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture)
 
 ; Function Attrs: nounwind
-define i8* @test_frame64(i8* %0) {
-; CHECK-LABEL: test_frame64:
+define i8* @test_align32(i32 signext %0, i8* nocapture readnone %1) {
+; CHECK-LABEL: test_align32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -304
+; CHECK-NEXT:    lea %s13, -288
 ; CHECK-NEXT:    and %s13, %s13, (32)0
 ; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    and %s11, %s11, (59)1
+; CHECK-NEXT:    or %s17, 0, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB2_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
 ; CHECK-NEXT:    or %s62, 0, %s0
 ; CHECK-NEXT:    lea %s63, 315
 ; CHECK-NEXT:    shm.l %s63, (%s61)
 ; CHECK-NEXT:    shm.l %s8, 8(%s61)
 ; CHECK-NEXT:    shm.l %s11, 16(%s61)
 ; CHECK-NEXT:    monc
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB2_2:
-; CHECK-NEXT:    or %s1, 0, %s0
+; CHECK-NEXT:    lea %s0, 15(, %s0)
+; CHECK-NEXT:    and %s0, -16, %s0
+; CHECK-NEXT:    lea %s1, __ve_grow_stack_align@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s12, __ve_grow_stack_align@hi(, %s1)
+; CHECK-NEXT:    or %s1, -32, (0)1
+; CHECK-NEXT:    bsic %s10, (, %s12)
+; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    lea %s0, 31(, %s0)
+; CHECK-NEXT:    and %s1, -32, %s0
 ; CHECK-NEXT:    lea %s0, fun@lo
 ; CHECK-NEXT:    and %s0, %s0, (32)0
 ; CHECK-NEXT:    lea.sl %s12, fun@hi(, %s0)
-; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    lea %s0, 256(, %s17)
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s17, 40(, %s11)
 ; CHECK-NEXT:    ld %s16, 32(, %s11)
 ; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
-; PIC-LABEL: test_frame64:
+; PIC-LABEL: test_align32:
 ; PIC:       # %bb.0:
 ; PIC-NEXT:    st %s9, (, %s11)
 ; PIC-NEXT:    st %s10, 8(, %s11)
 ; PIC-NEXT:    st %s15, 24(, %s11)
 ; PIC-NEXT:    st %s16, 32(, %s11)
+; PIC-NEXT:    st %s17, 40(, %s11)
 ; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s13, -304
+; PIC-NEXT:    lea %s13, -288
 ; PIC-NEXT:    and %s13, %s13, (32)0
 ; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; PIC-NEXT:    and %s11, %s11, (59)1
+; PIC-NEXT:    or %s17, 0, %s11
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB2_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)
 ; PIC-NEXT:    shm.l %s8, 8(%s61)
 ; PIC-NEXT:    shm.l %s11, 16(%s61)
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB2_2:
-; PIC-NEXT:    or %s1, 0, %s0
 ; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
 ; PIC-NEXT:    and %s15, %s15, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
+; PIC-NEXT:    lea %s0, 15(, %s0)
+; PIC-NEXT:    and %s0, -16, %s0
+; PIC-NEXT:    lea %s12, __ve_grow_stack_align@plt_lo(-24)
+; PIC-NEXT:    and %s12, %s12, (32)0
+; PIC-NEXT:    sic %s16
+; PIC-NEXT:    lea.sl %s12, __ve_grow_stack_align@plt_hi(%s16, %s12)
+; PIC-NEXT:    or %s1, -32, (0)1
+; PIC-NEXT:    bsic %s10, (, %s12)
+; PIC-NEXT:    lea %s0, 240(, %s11)
+; PIC-NEXT:    lea %s0, 31(, %s0)
+; PIC-NEXT:    and %s1, -32, %s0
 ; PIC-NEXT:    lea %s12, fun@plt_lo(-24)
 ; PIC-NEXT:    and %s12, %s12, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s12, fun@plt_hi(%s16, %s12)
-; PIC-NEXT:    lea %s0, 240(, %s11)
+; PIC-NEXT:    lea %s0, 256(, %s17)
 ; PIC-NEXT:    bsic %s10, (, %s12)
 ; PIC-NEXT:    or %s11, 0, %s9
+; PIC-NEXT:    ld %s17, 40(, %s11)
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = alloca [64 x i8], align 1
-  %3 = getelementptr inbounds [64 x i8], [64 x i8]* %2, i64 0, i64 0
-  call void @llvm.lifetime.start.p0i8(i64 64, i8* nonnull %3)
-  %4 = call i8* @fun(i8* nonnull %3, i8* %0)
-  call void @llvm.lifetime.end.p0i8(i64 64, i8* nonnull %3)
-  ret i8* %4
+  %3 = alloca [32 x i8], align 32
+  %4 = getelementptr inbounds [32 x i8], [32 x i8]* %3, i64 0, i64 0
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* nonnull %4)
+  %5 = sext i32 %0 to i64
+  %6 = alloca i8, i64 %5, align 32
+  %7 = call i8* @fun(i8* nonnull %4, i8* nonnull %6)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* nonnull %4)
+  ret i8* %7
 }
 
 ; Function Attrs: nounwind
-define i8* @test_frame128(i8* %0) {
-; CHECK-LABEL: test_frame128:
+define i8* @test_frame0_var(i8* %0, i8* %1) {
+; CHECK-LABEL: test_frame0_var:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -368
+; CHECK-NEXT:    lea %s13, -240
 ; CHECK-NEXT:    and %s13, %s13, (32)0
 ; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB3_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
 ; CHECK-NEXT:    or %s62, 0, %s0
 ; CHECK-NEXT:    lea %s63, 315
 ; CHECK-NEXT:    shm.l %s63, (%s61)
 ; CHECK-NEXT:    shm.l %s8, 8(%s61)
 ; CHECK-NEXT:    shm.l %s11, 16(%s61)
 ; CHECK-NEXT:    monc
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB3_2:
-; CHECK-NEXT:    or %s1, 0, %s0
-; CHECK-NEXT:    lea %s0, fun@lo
-; CHECK-NEXT:    and %s0, %s0, (32)0
-; CHECK-NEXT:    lea.sl %s12, fun@hi(, %s0)
-; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    lea %s2, data@lo
+; CHECK-NEXT:    and %s2, %s2, (32)0
+; CHECK-NEXT:    lea.sl %s2, data@hi(, %s2)
+; CHECK-NEXT:    ld1b.zx %s2, (, %s2)
+; CHECK-NEXT:    st1b %s2, (, %s0)
+; CHECK-NEXT:    lea %s2, fun@lo
+; CHECK-NEXT:    and %s2, %s2, (32)0
+; CHECK-NEXT:    lea.sl %s12, fun@hi(, %s2)
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s16, 32(, %s11)
 ; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
-; PIC-LABEL: test_frame128:
+; PIC-LABEL: test_frame0_var:
 ; PIC:       # %bb.0:
 ; PIC-NEXT:    st %s9, (, %s11)
 ; PIC-NEXT:    st %s10, 8(, %s11)
 ; PIC-NEXT:    st %s15, 24(, %s11)
 ; PIC-NEXT:    st %s16, 32(, %s11)
 ; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s13, -368
+; PIC-NEXT:    lea %s13, -240
 ; PIC-NEXT:    and %s13, %s13, (32)0
 ; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB3_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)
 ; PIC-NEXT:    shm.l %s8, 8(%s61)
 ; PIC-NEXT:    shm.l %s11, 16(%s61)
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB3_2:
-; PIC-NEXT:    or %s1, 0, %s0
 ; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
 ; PIC-NEXT:    and %s15, %s15, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
+; PIC-NEXT:    lea %s2, data@got_lo
+; PIC-NEXT:    and %s2, %s2, (32)0
+; PIC-NEXT:    lea.sl %s2, data@got_hi(, %s2)
+; PIC-NEXT:    ld %s2, (%s2, %s15)
+; PIC-NEXT:    ld1b.zx %s2, (, %s2)
+; PIC-NEXT:    st1b %s2, (, %s0)
 ; PIC-NEXT:    lea %s12, fun@plt_lo(-24)
 ; PIC-NEXT:    and %s12, %s12, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s12, fun@plt_hi(%s16, %s12)
-; PIC-NEXT:    lea %s0, 240(, %s11)
 ; PIC-NEXT:    bsic %s10, (, %s12)
 ; PIC-NEXT:    or %s11, 0, %s9
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = alloca [128 x i8], align 1
-  %3 = getelementptr inbounds [128 x i8], [128 x i8]* %2, i64 0, i64 0
-  call void @llvm.lifetime.start.p0i8(i64 128, i8* nonnull %3)
-  %4 = call i8* @fun(i8* nonnull %3, i8* %0)
-  call void @llvm.lifetime.end.p0i8(i64 128, i8* nonnull %3)
+  %3 = load i8, i8* @data, align 1
+  store i8 %3, i8* %0, align 1
+  %4 = tail call i8* @fun(i8* nonnull %0, i8* %1)
   ret i8* %4
 }
 
 ; Function Attrs: nounwind
-define i8* @test_frame65536(i8* %0) {
-; CHECK-LABEL: test_frame65536:
+define i8* @test_frame32_var(i8* %0) {
+; CHECK-LABEL: test_frame32_var:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -65776
+; CHECK-NEXT:    lea %s13, -272
 ; CHECK-NEXT:    and %s13, %s13, (32)0
 ; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB4_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
 ; CHECK-NEXT:    or %s62, 0, %s0
 ; CHECK-NEXT:    lea %s63, 315
 ; CHECK-NEXT:    shm.l %s63, (%s61)
 ; CHECK-NEXT:    shm.l %s8, 8(%s61)
 ; CHECK-NEXT:    shm.l %s11, 16(%s61)
 ; CHECK-NEXT:    monc
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB4_2:
+; CHECK-NEXT:    lea %s1, data@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s1, data@hi(, %s1)
+; CHECK-NEXT:    ld1b.zx %s2, (, %s1)
 ; CHECK-NEXT:    or %s1, 0, %s0
+; CHECK-NEXT:    st1b %s2, 240(, %s11)
 ; CHECK-NEXT:    lea %s0, fun@lo
 ; CHECK-NEXT:    and %s0, %s0, (32)0
 ; CHECK-NEXT:    lea.sl %s12, fun@hi(, %s0)
 ; CHECK-NEXT:    lea %s0, 240(, %s11)
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s16, 32(, %s11)
 ; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
-; PIC-LABEL: test_frame65536:
+; PIC-LABEL: test_frame32_var:
 ; PIC:       # %bb.0:
 ; PIC-NEXT:    st %s9, (, %s11)
 ; PIC-NEXT:    st %s10, 8(, %s11)
 ; PIC-NEXT:    st %s15, 24(, %s11)
 ; PIC-NEXT:    st %s16, 32(, %s11)
 ; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s13, -65776
+; PIC-NEXT:    lea %s13, -272
 ; PIC-NEXT:    and %s13, %s13, (32)0
 ; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB4_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)
 ; PIC-NEXT:    shm.l %s8, 8(%s61)
 ; PIC-NEXT:    shm.l %s11, 16(%s61)
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB4_2:
-; PIC-NEXT:    or %s1, 0, %s0
 ; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
 ; PIC-NEXT:    and %s15, %s15, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
+; PIC-NEXT:    lea %s1, data@got_lo
+; PIC-NEXT:    and %s1, %s1, (32)0
+; PIC-NEXT:    lea.sl %s1, data@got_hi(, %s1)
+; PIC-NEXT:    ld %s1, (%s1, %s15)
+; PIC-NEXT:    ld1b.zx %s2, (, %s1)
+; PIC-NEXT:    or %s1, 0, %s0
+; PIC-NEXT:    st1b %s2, 240(, %s11)
 ; PIC-NEXT:    lea %s12, fun@plt_lo(-24)
 ; PIC-NEXT:    and %s12, %s12, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s12, fun@plt_hi(%s16, %s12)
 ; PIC-NEXT:    lea %s0, 240(, %s11)
 ; PIC-NEXT:    bsic %s10, (, %s12)
 ; PIC-NEXT:    or %s11, 0, %s9
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = alloca [65536 x i8], align 1
-  %3 = getelementptr inbounds [65536 x i8], [65536 x i8]* %2, i64 0, i64 0
-  call void @llvm.lifetime.start.p0i8(i64 65536, i8* nonnull %3)
-  %4 = call i8* @fun(i8* nonnull %3, i8* %0)
-  call void @llvm.lifetime.end.p0i8(i64 65536, i8* nonnull %3)
-  ret i8* %4
+  %2 = alloca [32 x i8], align 1
+  %3 = getelementptr inbounds [32 x i8], [32 x i8]* %2, i64 0, i64 0
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* nonnull %3)
+  %4 = load i8, i8* @data, align 1
+  store i8 %4, i8* %3, align 1
+  %5 = call i8* @fun(i8* nonnull %3, i8* %0)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* nonnull %3)
+  ret i8* %5
 }
 
 ; Function Attrs: nounwind
-define i8* @test_frame4294967296(i8* %0) {
-; CHECK-LABEL: test_frame4294967296:
+define i8* @test_align32_var(i32 signext %0, i8* nocapture readnone %1) {
+; CHECK-LABEL: test_align32_var:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -240
+; CHECK-NEXT:    lea %s13, -288
 ; CHECK-NEXT:    and %s13, %s13, (32)0
-; CHECK-NEXT:    lea.sl %s11, -2(%s13, %s11)
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    and %s11, %s11, (59)1
+; CHECK-NEXT:    or %s17, 0, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB5_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
 ; CHECK-NEXT:    or %s62, 0, %s0
 ; CHECK-NEXT:    lea %s63, 315
 ; CHECK-NEXT:    shm.l %s63, (%s61)
 ; CHECK-NEXT:    shm.l %s8, 8(%s61)
 ; CHECK-NEXT:    shm.l %s11, 16(%s61)
 ; CHECK-NEXT:    monc
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB5_2:
-; CHECK-NEXT:    or %s1, 0, %s0
+; CHECK-NEXT:    lea %s0, 15(, %s0)
+; CHECK-NEXT:    and %s0, -16, %s0
+; CHECK-NEXT:    lea %s1, __ve_grow_stack_align@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s12, __ve_grow_stack_align@hi(, %s1)
+; CHECK-NEXT:    or %s1, -32, (0)1
+; CHECK-NEXT:    bsic %s10, (, %s12)
+; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    lea %s1, data@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s1, data@hi(, %s1)
+; CHECK-NEXT:    ld1b.zx %s2, (, %s1)
+; CHECK-NEXT:    lea %s0, 31(, %s0)
+; CHECK-NEXT:    and %s1, -32, %s0
+; CHECK-NEXT:    st1b %s2, (, %s1)
 ; CHECK-NEXT:    lea %s0, fun@lo
 ; CHECK-NEXT:    and %s0, %s0, (32)0
 ; CHECK-NEXT:    lea.sl %s12, fun@hi(, %s0)
-; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    lea %s0, 256(, %s17)
 ; CHECK-NEXT:    bsic %s10, (, %s12)
 ; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s17, 40(, %s11)
 ; CHECK-NEXT:    ld %s16, 32(, %s11)
 ; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
-; PIC-LABEL: test_frame4294967296:
+; PIC-LABEL: test_align32_var:
 ; PIC:       # %bb.0:
 ; PIC-NEXT:    st %s9, (, %s11)
 ; PIC-NEXT:    st %s10, 8(, %s11)
 ; PIC-NEXT:    st %s15, 24(, %s11)
 ; PIC-NEXT:    st %s16, 32(, %s11)
+; PIC-NEXT:    st %s17, 40(, %s11)
 ; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s13, -240
+; PIC-NEXT:    lea %s13, -288
 ; PIC-NEXT:    and %s13, %s13, (32)0
-; PIC-NEXT:    lea.sl %s11, -2(%s13, %s11)
+; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; PIC-NEXT:    and %s11, %s11, (59)1
+; PIC-NEXT:    or %s17, 0, %s11
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB5_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)
 ; PIC-NEXT:    shm.l %s8, 8(%s61)
 ; PIC-NEXT:    shm.l %s11, 16(%s61)
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB5_2:
-; PIC-NEXT:    or %s1, 0, %s0
 ; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
 ; PIC-NEXT:    and %s15, %s15, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
+; PIC-NEXT:    lea %s0, 15(, %s0)
+; PIC-NEXT:    and %s0, -16, %s0
+; PIC-NEXT:    lea %s12, __ve_grow_stack_align@plt_lo(-24)
+; PIC-NEXT:    and %s12, %s12, (32)0
+; PIC-NEXT:    sic %s16
+; PIC-NEXT:    lea.sl %s12, __ve_grow_stack_align@plt_hi(%s16, %s12)
+; PIC-NEXT:    or %s1, -32, (0)1
+; PIC-NEXT:    bsic %s10, (, %s12)
+; PIC-NEXT:    lea %s0, data@got_lo
+; PIC-NEXT:    and %s0, %s0, (32)0
+; PIC-NEXT:    lea.sl %s0, data@got_hi(, %s0)
+; PIC-NEXT:    ld %s0, (%s0, %s15)
+; PIC-NEXT:    lea %s1, 240(, %s11)
+; PIC-NEXT:    ld1b.zx %s0, (, %s0)
+; PIC-NEXT:    lea %s1, 31(, %s1)
+; PIC-NEXT:    and %s1, -32, %s1
+; PIC-NEXT:    st1b %s0, (, %s1)
 ; PIC-NEXT:    lea %s12, fun@plt_lo(-24)
 ; PIC-NEXT:    and %s12, %s12, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s12, fun@plt_hi(%s16, %s12)
-; PIC-NEXT:    lea %s0, 240(, %s11)
+; PIC-NEXT:    lea %s0, 256(, %s17)
 ; PIC-NEXT:    bsic %s10, (, %s12)
 ; PIC-NEXT:    or %s11, 0, %s9
+; PIC-NEXT:    ld %s17, 40(, %s11)
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = alloca [4294967296 x i8], align 1
-  %3 = getelementptr inbounds [4294967296 x i8], [4294967296 x i8]* %2, i64 0, i64 0
-  call void @llvm.lifetime.start.p0i8(i64 4294967296, i8* nonnull %3)
-  %4 = call i8* @fun(i8* nonnull %3, i8* %0)
-  call void @llvm.lifetime.end.p0i8(i64 4294967296, i8* nonnull %3)
-  ret i8* %4
+  %3 = alloca [32 x i8], align 32
+  %4 = getelementptr inbounds [32 x i8], [32 x i8]* %3, i64 0, i64 0
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* nonnull %4)
+  %5 = sext i32 %0 to i64
+  %6 = alloca i8, i64 %5, align 32
+  %7 = load i8, i8* @data, align 1
+  store i8 %7, i8* %6, align 32
+  %8 = call i8* @fun(i8* nonnull %4, i8* nonnull %6)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* nonnull %4)
+  ret i8* %8
 }

llvm/test/CodeGen/VE/Scalar/stackframe_nocall.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=ve | FileCheck %s
-; RUN: llc < %s -mtriple=ve -relocation-model=pic | FileCheck %s --check-prefix=PIC
+; RUN: llc < %s -mtriple=ve -relocation-model=pic \
+; RUN:     | FileCheck %s --check-prefix=PIC
 
-;; Check stack frame allocation of a function which doesn't call other functions
+;;; Check stack frame allocation of a function which does not calls other
+;;; functions under following conditions and combinations of them:
+;;;   - access variable or not
+;;;   - no stack object, a stack object using BP, or a stack object not using BP
+;;;   - isPositionIndependent or not
 
 @data = external global i8, align 1
 
 ; Function Attrs: norecurse nounwind readnone
-define signext i32 @test_frame0(i32 signext %0) {
+define i8* @test_frame0(i8* nocapture readnone %0, i8* readnone returned %1) {
 ; CHECK-LABEL: test_frame0:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    adds.w.sx %s0, 3, %s0
-; CHECK-NEXT:    adds.w.sx %s0, %s0, (0)1
+; CHECK-NEXT:    or %s0, 0, %s1
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
 ; PIC-LABEL: test_frame0:
 ; PIC:       # %bb.0:
-; PIC-NEXT:    adds.w.sx %s0, 3, %s0
-; PIC-NEXT:    adds.w.sx %s0, %s0, (0)1
+; PIC-NEXT:    or %s0, 0, %s1
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = add nsw i32 %0, 3
-  ret i32 %2
+  ret i8* %1
 }
 
-; Function Attrs: nounwind
-define i8* @test_frame32(i8* %0) {
+; Function Attrs: nofree nounwind
+define nonnull i8* @test_frame32(i8* nocapture readonly %0) {
 ; CHECK-LABEL: test_frame32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
 ; CHECK-NEXT:    lea %s13, -208
 ; CHECK-NEXT:    and %s13, %s13, (32)0
 ; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB1_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
 ; CHECK-NEXT:    or %s62, 0, %s0
 ; CHECK-NEXT:    lea %s63, 315
 ; CHECK-NEXT:    shm.l %s63, (%s61)
 ; CHECK-NEXT:    shm.l %s8, 8(%s61)
 ; CHECK-NEXT:    shm.l %s11, 16(%s61)
 ; CHECK-NEXT:    monc
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB1_2:
-; CHECK-NEXT:    lea %s0, data@lo
-; CHECK-NEXT:    and %s0, %s0, (32)0
-; CHECK-NEXT:    lea.sl %s0, data@hi(, %s0)
 ; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
 ; CHECK-NEXT:    lea %s0, 176(, %s11)
 ; CHECK-NEXT:    st1b %s1, 176(, %s11)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s16, 32(, %s11)
 ; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)

 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)
 ; PIC-NEXT:    shm.l %s8, 8(%s61)
 ; PIC-NEXT:    shm.l %s11, 16(%s61)
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB1_2:
-; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
-; PIC-NEXT:    and %s15, %s15, (32)0
-; PIC-NEXT:    sic %s16
-; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
-; PIC-NEXT:    lea %s0, data@got_lo
-; PIC-NEXT:    and %s0, %s0, (32)0
-; PIC-NEXT:    lea.sl %s0, data@got_hi(, %s0)
-; PIC-NEXT:    ld %s0, (%s0, %s15)
 ; PIC-NEXT:    ld1b.zx %s1, (, %s0)
 ; PIC-NEXT:    lea %s0, 176(, %s11)
 ; PIC-NEXT:    st1b %s1, 176(, %s11)
 ; PIC-NEXT:    or %s11, 0, %s9
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
   %2 = alloca [32 x i8], align 1
   %3 = getelementptr inbounds [32 x i8], [32 x i8]* %2, i64 0, i64 0
-  %4 = load i8, i8* @data, align 1
-  store i8 %4, i8* %3, align 1
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* nonnull %3)
+  %4 = load i8, i8* %0, align 1
+  store volatile i8 %4, i8* %3, align 1
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* nonnull %3)
   ret i8* %3
 }
 
 ; Function Attrs: argmemonly nofree nosync nounwind willreturn
 declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture)
 
-declare i8* @fun(i8*, i8*)
-
 ; Function Attrs: argmemonly nofree nosync nounwind willreturn
 declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture)
 
-; Function Attrs: nounwind
-define i8* @test_frame64(i8* %0) {
-; CHECK-LABEL: test_frame64:
+; Function Attrs: nofree nounwind
+define noalias nonnull i8* @test_align32(i32 signext %0, i8* nocapture readonly %1) {
+; CHECK-LABEL: test_align32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -240
+; CHECK-NEXT:    lea %s13, -288
 ; CHECK-NEXT:    and %s13, %s13, (32)0
 ; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    and %s11, %s11, (59)1
+; CHECK-NEXT:    or %s17, 0, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB2_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
 ; CHECK-NEXT:    or %s62, 0, %s0
 ; CHECK-NEXT:    lea %s63, 315
 ; CHECK-NEXT:    shm.l %s63, (%s61)
 ; CHECK-NEXT:    shm.l %s8, 8(%s61)
 ; CHECK-NEXT:    shm.l %s11, 16(%s61)
 ; CHECK-NEXT:    monc
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB2_2:
-; CHECK-NEXT:    lea %s0, data@lo
-; CHECK-NEXT:    and %s0, %s0, (32)0
-; CHECK-NEXT:    lea.sl %s0, data@hi(, %s0)
-; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s2, 0, %s1
+; CHECK-NEXT:    lea %s0, 15(, %s0)
+; CHECK-NEXT:    and %s0, -16, %s0
+; CHECK-NEXT:    lea %s1, __ve_grow_stack_align@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s12, __ve_grow_stack_align@hi(, %s1)
+; CHECK-NEXT:    or %s1, -32, (0)1
+; CHECK-NEXT:    bsic %s10, (, %s12)
+; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    ld1b.zx %s1, (, %s2)
+; CHECK-NEXT:    lea %s0, 31(, %s0)
+; CHECK-NEXT:    and %s2, -32, %s0
+; CHECK-NEXT:    lea %s0, 256(, %s17)
+; CHECK-NEXT:    st1b %s1, (, %s2)
 ; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s17, 40(, %s11)
 ; CHECK-NEXT:    ld %s16, 32(, %s11)
 ; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
-; PIC-LABEL: test_frame64:
+; PIC-LABEL: test_align32:
 ; PIC:       # %bb.0:
 ; PIC-NEXT:    st %s9, (, %s11)
 ; PIC-NEXT:    st %s10, 8(, %s11)
 ; PIC-NEXT:    st %s15, 24(, %s11)
 ; PIC-NEXT:    st %s16, 32(, %s11)
+; PIC-NEXT:    st %s17, 40(, %s11)
 ; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s13, -240
+; PIC-NEXT:    lea %s13, -288
 ; PIC-NEXT:    and %s13, %s13, (32)0
 ; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; PIC-NEXT:    and %s11, %s11, (59)1
+; PIC-NEXT:    or %s17, 0, %s11
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB2_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)
 ; PIC-NEXT:    shm.l %s8, 8(%s61)
 ; PIC-NEXT:    shm.l %s11, 16(%s61)
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB2_2:
+; PIC-NEXT:    or %s2, 0, %s1
 ; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
 ; PIC-NEXT:    and %s15, %s15, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
-; PIC-NEXT:    lea %s0, data@got_lo
-; PIC-NEXT:    and %s0, %s0, (32)0
-; PIC-NEXT:    lea.sl %s0, data@got_hi(, %s0)
-; PIC-NEXT:    ld %s0, (%s0, %s15)
-; PIC-NEXT:    ld1b.zx %s1, (, %s0)
-; PIC-NEXT:    lea %s0, 176(, %s11)
-; PIC-NEXT:    st1b %s1, 176(, %s11)
+; PIC-NEXT:    lea %s0, 15(, %s0)
+; PIC-NEXT:    and %s0, -16, %s0
+; PIC-NEXT:    lea %s12, __ve_grow_stack_align@plt_lo(-24)
+; PIC-NEXT:    and %s12, %s12, (32)0
+; PIC-NEXT:    sic %s16
+; PIC-NEXT:    lea.sl %s12, __ve_grow_stack_align@plt_hi(%s16, %s12)
+; PIC-NEXT:    or %s1, -32, (0)1
+; PIC-NEXT:    bsic %s10, (, %s12)
+; PIC-NEXT:    lea %s0, 240(, %s11)
+; PIC-NEXT:    ld1b.zx %s1, (, %s2)
+; PIC-NEXT:    lea %s0, 31(, %s0)
+; PIC-NEXT:    and %s2, -32, %s0
+; PIC-NEXT:    lea %s0, 256(, %s17)
+; PIC-NEXT:    st1b %s1, (, %s2)
 ; PIC-NEXT:    or %s11, 0, %s9
+; PIC-NEXT:    ld %s17, 40(, %s11)
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = alloca [64 x i8], align 1
-  %3 = getelementptr inbounds [64 x i8], [64 x i8]* %2, i64 0, i64 0
-  %4 = load i8, i8* @data, align 1
-  store i8 %4, i8* %3, align 1
-  ret i8* %3
+  %3 = alloca [32 x i8], align 32
+  %4 = getelementptr inbounds [32 x i8], [32 x i8]* %3, i64 0, i64 0
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* nonnull %4)
+  %5 = sext i32 %0 to i64
+  %6 = alloca i8, i64 %5, align 32
+  %7 = load i8, i8* %1, align 1
+  store volatile i8 %7, i8* %6, align 32
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* nonnull %4)
+  ret i8* %4
 }
 
-; Function Attrs: nounwind
-define i8* @test_frame128(i8* %0) {
-; CHECK-LABEL: test_frame128:
+; Function Attrs: nofree norecurse nounwind
+define i8* @test_frame0_var(i8* returned %0, i8* nocapture readnone %1) {
+; CHECK-LABEL: test_frame0_var:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    st %s9, (, %s11)
-; CHECK-NEXT:    st %s10, 8(, %s11)
-; CHECK-NEXT:    st %s15, 24(, %s11)
-; CHECK-NEXT:    st %s16, 32(, %s11)
-; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -304
-; CHECK-NEXT:    and %s13, %s13, (32)0
-; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
-; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB3_2
-; CHECK-NEXT:  # %bb.1:
-; CHECK-NEXT:    ld %s61, 24(, %s14)
-; CHECK-NEXT:    or %s62, 0, %s0
-; CHECK-NEXT:    lea %s63, 315
-; CHECK-NEXT:    shm.l %s63, (%s61)
-; CHECK-NEXT:    shm.l %s8, 8(%s61)
-; CHECK-NEXT:    shm.l %s11, 16(%s61)
-; CHECK-NEXT:    monc
-; CHECK-NEXT:    or %s0, 0, %s62
-; CHECK-NEXT:  .LBB3_2:
-; CHECK-NEXT:    lea %s0, data@lo
-; CHECK-NEXT:    and %s0, %s0, (32)0
-; CHECK-NEXT:    lea.sl %s0, data@hi(, %s0)
-; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
-; CHECK-NEXT:    or %s11, 0, %s9
-; CHECK-NEXT:    ld %s16, 32(, %s11)
-; CHECK-NEXT:    ld %s15, 24(, %s11)
-; CHECK-NEXT:    ld %s10, 8(, %s11)
-; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    lea %s1, data@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s1, data@hi(, %s1)
+; CHECK-NEXT:    ld1b.zx %s1, (, %s1)
+; CHECK-NEXT:    st1b %s1, (, %s0)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
-; PIC-LABEL: test_frame128:
+; PIC-LABEL: test_frame0_var:
 ; PIC:       # %bb.0:
 ; PIC-NEXT:    st %s9, (, %s11)
 ; PIC-NEXT:    st %s10, 8(, %s11)
 ; PIC-NEXT:    st %s15, 24(, %s11)
 ; PIC-NEXT:    st %s16, 32(, %s11)
 ; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s13, -304
+; PIC-NEXT:    lea %s13, -176
 ; PIC-NEXT:    and %s13, %s13, (32)0
 ; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB3_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)
 ; PIC-NEXT:    shm.l %s8, 8(%s61)
 ; PIC-NEXT:    shm.l %s11, 16(%s61)
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB3_2:
 ; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
 ; PIC-NEXT:    and %s15, %s15, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
-; PIC-NEXT:    lea %s0, data@got_lo
-; PIC-NEXT:    and %s0, %s0, (32)0
-; PIC-NEXT:    lea.sl %s0, data@got_hi(, %s0)
-; PIC-NEXT:    ld %s0, (%s0, %s15)
-; PIC-NEXT:    ld1b.zx %s1, (, %s0)
-; PIC-NEXT:    lea %s0, 176(, %s11)
-; PIC-NEXT:    st1b %s1, 176(, %s11)
+; PIC-NEXT:    lea %s1, data@got_lo
+; PIC-NEXT:    and %s1, %s1, (32)0
+; PIC-NEXT:    lea.sl %s1, data@got_hi(, %s1)
+; PIC-NEXT:    ld %s1, (%s1, %s15)
+; PIC-NEXT:    ld1b.zx %s1, (, %s1)
+; PIC-NEXT:    st1b %s1, (, %s0)
 ; PIC-NEXT:    or %s11, 0, %s9
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = alloca [128 x i8], align 1
-  %3 = getelementptr inbounds [128 x i8], [128 x i8]* %2, i64 0, i64 0
-  %4 = load i8, i8* @data, align 1
-  store i8 %4, i8* %3, align 1
-  ret i8* %3
+  %3 = load i8, i8* @data, align 1
+  store i8 %3, i8* %0, align 1
+  ret i8* %0
 }
 
-; Function Attrs: nounwind
-define i8* @test_frame65536(i8* %0) {
-; CHECK-LABEL: test_frame65536:
+; Function Attrs: nofree nounwind
+define nonnull i8* @test_frame32_var(i8* nocapture readnone %0) {
+; CHECK-LABEL: test_frame32_var:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -65712
+; CHECK-NEXT:    lea %s13, -208
 ; CHECK-NEXT:    and %s13, %s13, (32)0
 ; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB4_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
 ; CHECK-NEXT:    or %s62, 0, %s0
 ; CHECK-NEXT:    lea %s63, 315
 ; CHECK-NEXT:    shm.l %s63, (%s61)

 ; CHECK-NEXT:    st1b %s1, 176(, %s11)
 ; CHECK-NEXT:    or %s11, 0, %s9
 ; CHECK-NEXT:    ld %s16, 32(, %s11)
 ; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
-; PIC-LABEL: test_frame65536:
+; PIC-LABEL: test_frame32_var:
 ; PIC:       # %bb.0:
 ; PIC-NEXT:    st %s9, (, %s11)
 ; PIC-NEXT:    st %s10, 8(, %s11)
 ; PIC-NEXT:    st %s15, 24(, %s11)
 ; PIC-NEXT:    st %s16, 32(, %s11)
 ; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s13, -65712
+; PIC-NEXT:    lea %s13, -208
 ; PIC-NEXT:    and %s13, %s13, (32)0
 ; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB4_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)

 ; PIC-NEXT:    lea %s0, 176(, %s11)
 ; PIC-NEXT:    st1b %s1, 176(, %s11)
 ; PIC-NEXT:    or %s11, 0, %s9
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = alloca [65536 x i8], align 1
-  %3 = getelementptr inbounds [65536 x i8], [65536 x i8]* %2, i64 0, i64 0
+  %2 = alloca [32 x i8], align 1
+  %3 = getelementptr inbounds [32 x i8], [32 x i8]* %2, i64 0, i64 0
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* nonnull %3)
   %4 = load i8, i8* @data, align 1
-  store i8 %4, i8* %3, align 1
+  store volatile i8 %4, i8* %3, align 1
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* nonnull %3)
   ret i8* %3
 }
 
-; Function Attrs: nounwind
-define i8* @test_frame4294967296(i8* %0) {
-; CHECK-LABEL: test_frame4294967296:
+; Function Attrs: nofree nounwind
+define noalias nonnull i8* @test_align32_var(i32 signext %0, i8* nocapture readonly %1) {
+; CHECK-LABEL: test_align32_var:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    st %s9, (, %s11)
 ; CHECK-NEXT:    st %s10, 8(, %s11)
 ; CHECK-NEXT:    st %s15, 24(, %s11)
 ; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    st %s17, 40(, %s11)
 ; CHECK-NEXT:    or %s9, 0, %s11
-; CHECK-NEXT:    lea %s13, -176
+; CHECK-NEXT:    lea %s13, -288
 ; CHECK-NEXT:    and %s13, %s13, (32)0
-; CHECK-NEXT:    lea.sl %s11, -2(%s13, %s11)
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    and %s11, %s11, (59)1
+; CHECK-NEXT:    or %s17, 0, %s11
 ; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB5_2
 ; CHECK-NEXT:  # %bb.1:
 ; CHECK-NEXT:    ld %s61, 24(, %s14)
 ; CHECK-NEXT:    or %s62, 0, %s0
 ; CHECK-NEXT:    lea %s63, 315
 ; CHECK-NEXT:    shm.l %s63, (%s61)
 ; CHECK-NEXT:    shm.l %s8, 8(%s61)
 ; CHECK-NEXT:    shm.l %s11, 16(%s61)
 ; CHECK-NEXT:    monc
 ; CHECK-NEXT:    or %s0, 0, %s62
 ; CHECK-NEXT:  .LBB5_2:
-; CHECK-NEXT:    lea %s0, data@lo
-; CHECK-NEXT:    and %s0, %s0, (32)0
-; CHECK-NEXT:    lea.sl %s0, data@hi(, %s0)
-; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
-; CHECK-NEXT:    lea %s0, 176(, %s11)
-; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s2, 0, %s1
+; CHECK-NEXT:    lea %s0, 15(, %s0)
+; CHECK-NEXT:    and %s0, -16, %s0
+; CHECK-NEXT:    lea %s1, __ve_grow_stack_align@lo
+; CHECK-NEXT:    and %s1, %s1, (32)0
+; CHECK-NEXT:    lea.sl %s12, __ve_grow_stack_align@hi(, %s1)
+; CHECK-NEXT:    or %s1, -32, (0)1
+; CHECK-NEXT:    bsic %s10, (, %s12)
+; CHECK-NEXT:    lea %s0, 240(, %s11)
+; CHECK-NEXT:    ld1b.zx %s1, (, %s2)
+; CHECK-NEXT:    lea %s0, 31(, %s0)
+; CHECK-NEXT:    and %s2, -32, %s0
+; CHECK-NEXT:    lea %s0, 256(, %s17)
+; CHECK-NEXT:    st1b %s1, (, %s2)
 ; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s17, 40(, %s11)
 ; CHECK-NEXT:    ld %s16, 32(, %s11)
 ; CHECK-NEXT:    ld %s15, 24(, %s11)
 ; CHECK-NEXT:    ld %s10, 8(, %s11)
 ; CHECK-NEXT:    ld %s9, (, %s11)
 ; CHECK-NEXT:    b.l.t (, %s10)
 ;
-; PIC-LABEL: test_frame4294967296:
+; PIC-LABEL: test_align32_var:
 ; PIC:       # %bb.0:
 ; PIC-NEXT:    st %s9, (, %s11)
 ; PIC-NEXT:    st %s10, 8(, %s11)
 ; PIC-NEXT:    st %s15, 24(, %s11)
 ; PIC-NEXT:    st %s16, 32(, %s11)
+; PIC-NEXT:    st %s17, 40(, %s11)
 ; PIC-NEXT:    or %s9, 0, %s11
-; PIC-NEXT:    lea %s13, -176
+; PIC-NEXT:    lea %s13, -288
 ; PIC-NEXT:    and %s13, %s13, (32)0
-; PIC-NEXT:    lea.sl %s11, -2(%s13, %s11)
+; PIC-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; PIC-NEXT:    and %s11, %s11, (59)1
+; PIC-NEXT:    or %s17, 0, %s11
 ; PIC-NEXT:    brge.l.t %s11, %s8, .LBB5_2
 ; PIC-NEXT:  # %bb.1:
 ; PIC-NEXT:    ld %s61, 24(, %s14)
 ; PIC-NEXT:    or %s62, 0, %s0
 ; PIC-NEXT:    lea %s63, 315
 ; PIC-NEXT:    shm.l %s63, (%s61)
 ; PIC-NEXT:    shm.l %s8, 8(%s61)
 ; PIC-NEXT:    shm.l %s11, 16(%s61)
 ; PIC-NEXT:    monc
 ; PIC-NEXT:    or %s0, 0, %s62
 ; PIC-NEXT:  .LBB5_2:
+; PIC-NEXT:    or %s2, 0, %s1
 ; PIC-NEXT:    lea %s15, _GLOBAL_OFFSET_TABLE_@pc_lo(-24)
 ; PIC-NEXT:    and %s15, %s15, (32)0
 ; PIC-NEXT:    sic %s16
 ; PIC-NEXT:    lea.sl %s15, _GLOBAL_OFFSET_TABLE_@pc_hi(%s16, %s15)
-; PIC-NEXT:    lea %s0, data@got_lo
-; PIC-NEXT:    and %s0, %s0, (32)0
-; PIC-NEXT:    lea.sl %s0, data@got_hi(, %s0)
-; PIC-NEXT:    ld %s0, (%s0, %s15)
-; PIC-NEXT:    ld1b.zx %s1, (, %s0)
-; PIC-NEXT:    lea %s0, 176(, %s11)
-; PIC-NEXT:    st1b %s1, 176(, %s11)
+; PIC-NEXT:    lea %s0, 15(, %s0)
+; PIC-NEXT:    and %s0, -16, %s0
+; PIC-NEXT:    lea %s12, __ve_grow_stack_align@plt_lo(-24)
+; PIC-NEXT:    and %s12, %s12, (32)0
+; PIC-NEXT:    sic %s16
+; PIC-NEXT:    lea.sl %s12, __ve_grow_stack_align@plt_hi(%s16, %s12)
+; PIC-NEXT:    or %s1, -32, (0)1
+; PIC-NEXT:    bsic %s10, (, %s12)
+; PIC-NEXT:    lea %s0, 240(, %s11)
+; PIC-NEXT:    ld1b.zx %s1, (, %s2)
+; PIC-NEXT:    lea %s0, 31(, %s0)
+; PIC-NEXT:    and %s2, -32, %s0
+; PIC-NEXT:    lea %s0, 256(, %s17)
+; PIC-NEXT:    st1b %s1, (, %s2)
 ; PIC-NEXT:    or %s11, 0, %s9
+; PIC-NEXT:    ld %s17, 40(, %s11)
 ; PIC-NEXT:    ld %s16, 32(, %s11)
 ; PIC-NEXT:    ld %s15, 24(, %s11)
 ; PIC-NEXT:    ld %s10, 8(, %s11)
 ; PIC-NEXT:    ld %s9, (, %s11)
 ; PIC-NEXT:    b.l.t (, %s10)
-  %2 = alloca [4294967296 x i8], align 1
-  %3 = getelementptr inbounds [4294967296 x i8], [4294967296 x i8]* %2, i64 0, i64 0
-  %4 = load i8, i8* @data, align 1
-  store i8 %4, i8* %3, align 1
-  ret i8* %3
+  %3 = alloca [32 x i8], align 32
+  %4 = getelementptr inbounds [32 x i8], [32 x i8]* %3, i64 0, i64 0
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* nonnull %4)
+  %5 = sext i32 %0 to i64
+  %6 = alloca i8, i64 %5, align 32
+  %7 = load i8, i8* %1, align 1
+  store volatile i8 %7, i8* %6, align 32
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* nonnull %4)
+  ret i8* %4
 }

llvm/test/CodeGen/VE/Scalar/stackframe_size.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=ve | FileCheck %s
+
+;;; Check stack frame allocation especially about the new SP calculation
+;;; using multiple sizes as a test of emitSPAdjustment().
+
+; Function Attrs: norecurse nounwind readnone
+define signext i32 @test_frame0(i32 signext %0) {
+; CHECK-LABEL: test_frame0:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    adds.w.sx %s0, 3, %s0
+; CHECK-NEXT:    adds.w.sx %s0, %s0, (0)1
+; CHECK-NEXT:    b.l.t (, %s10)
+  %2 = add nsw i32 %0, 3
+  ret i32 %2
+}
+
+; Function Attrs: nounwind
+define i8* @test_frame8(i8* %0) {
+; CHECK-LABEL: test_frame8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -192
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB1_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB1_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 184(, %s11)
+; CHECK-NEXT:    st1b %s1, 184(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+  %2 = alloca [8 x i8], align 1
+  %3 = getelementptr inbounds [8 x i8], [8 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+; Function Attrs: nounwind
+define i8* @test_frame16(i8* %0) {
+; CHECK-LABEL: test_frame16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -192
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB2_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB2_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 176(, %s11)
+; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+  %2 = alloca [16 x i8], align 1
+  %3 = getelementptr inbounds [16 x i8], [16 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+; Function Attrs: nounwind
+define i8* @test_frame32(i8* %0) {
+; CHECK-LABEL: test_frame32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -208
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB3_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB3_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 176(, %s11)
+; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+  %2 = alloca [32 x i8], align 1
+  %3 = getelementptr inbounds [32 x i8], [32 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+; Function Attrs: nounwind
+define i8* @test_frame64(i8* %0) {
+; CHECK-LABEL: test_frame64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -240
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB4_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB4_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 176(, %s11)
+; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+  %2 = alloca [64 x i8], align 1
+  %3 = getelementptr inbounds [64 x i8], [64 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+; Function Attrs: nounwind
+define i8* @test_frame128(i8* %0) {
+; CHECK-LABEL: test_frame128:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -304
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB5_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB5_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 176(, %s11)
+; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+  %2 = alloca [128 x i8], align 1
+  %3 = getelementptr inbounds [128 x i8], [128 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+; Function Attrs: nounwind
+define i8* @test_frame65536(i8* %0) {
+; CHECK-LABEL: test_frame65536:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -65712
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -1(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB6_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB6_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 176(, %s11)
+; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+  %2 = alloca [65536 x i8], align 1
+  %3 = getelementptr inbounds [65536 x i8], [65536 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}
+
+; Function Attrs: nounwind
+define i8* @test_frame4294967296(i8* %0) {
+; CHECK-LABEL: test_frame4294967296:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    st %s9, (, %s11)
+; CHECK-NEXT:    st %s10, 8(, %s11)
+; CHECK-NEXT:    st %s15, 24(, %s11)
+; CHECK-NEXT:    st %s16, 32(, %s11)
+; CHECK-NEXT:    or %s9, 0, %s11
+; CHECK-NEXT:    lea %s13, -176
+; CHECK-NEXT:    and %s13, %s13, (32)0
+; CHECK-NEXT:    lea.sl %s11, -2(%s13, %s11)
+; CHECK-NEXT:    brge.l.t %s11, %s8, .LBB7_2
+; CHECK-NEXT:  # %bb.1:
+; CHECK-NEXT:    ld %s61, 24(, %s14)
+; CHECK-NEXT:    or %s62, 0, %s0
+; CHECK-NEXT:    lea %s63, 315
+; CHECK-NEXT:    shm.l %s63, (%s61)
+; CHECK-NEXT:    shm.l %s8, 8(%s61)
+; CHECK-NEXT:    shm.l %s11, 16(%s61)
+; CHECK-NEXT:    monc
+; CHECK-NEXT:    or %s0, 0, %s62
+; CHECK-NEXT:  .LBB7_2:
+; CHECK-NEXT:    ld1b.zx %s1, (, %s0)
+; CHECK-NEXT:    lea %s0, 176(, %s11)
+; CHECK-NEXT:    st1b %s1, 176(, %s11)
+; CHECK-NEXT:    or %s11, 0, %s9
+; CHECK-NEXT:    ld %s16, 32(, %s11)
+; CHECK-NEXT:    ld %s15, 24(, %s11)
+; CHECK-NEXT:    ld %s10, 8(, %s11)
+; CHECK-NEXT:    ld %s9, (, %s11)
+; CHECK-NEXT:    b.l.t (, %s10)
+  %2 = alloca [4294967296 x i8], align 1
+  %3 = getelementptr inbounds [4294967296 x i8], [4294967296 x i8]* %2, i64 0, i64 0
+  %4 = load i8, i8* %0, align 1
+  store i8 %4, i8* %3, align 1
+  ret i8* %3
+}