test/CodeGen/LoongArch/stack-probing-dynamic.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=loongarch64 -O2 < %s | FileCheck %s -check-prefix=LA64
; RUN: llc -mtriple=loongarch32 -O2 < %s | FileCheck %s -check-prefix=LA32

;; From llvm/test/CodeGen/AArch64/stack-probing-dynamic.ll

;; Dynamically-sized allocation, needs a loop which can handle any size at
;; runtime. The final iteration of the loop will temporarily put SP below the
;; target address, but this doesn't break any of the ABI constraints on the
;; stack, and also doesn't probe below the target SP value.
define void @dynamic(i64 %size, ptr %out) #0 {
;
; LA64-LABEL: dynamic:
; LA64:       # %bb.0:
; LA64-NEXT:    addi.d $sp, $sp, -16
; LA64-NEXT:    .cfi_def_cfa_offset 16
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    st.d $ra, $sp, 8 # 8-byte Folded Spill
; LA64-NEXT:    st.d $fp, $sp, 0 # 8-byte Folded Spill
; LA64-NEXT:    .cfi_offset 1, -8
; LA64-NEXT:    .cfi_offset 22, -16
; LA64-NEXT:    addi.d $fp, $sp, 16
; LA64-NEXT:    .cfi_def_cfa 22, 0
; LA64-NEXT:    addi.d $a0, $a0, 15
; LA64-NEXT:    bstrins.d $a0, $zero, 3, 0
; LA64-NEXT:    sub.d $a0, $sp, $a0
; LA64-NEXT:    lu12i.w $a2, 1
; LA64-NEXT:  .LBB0_1: # =>This Inner Loop Header: Depth=1
; LA64-NEXT:    sub.d $sp, $sp, $a2
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    bltu $a0, $sp, .LBB0_1
; LA64-NEXT:  # %bb.2:
; LA64-NEXT:    move $sp, $a0
; LA64-NEXT:    st.d $a0, $a1, 0
; LA64-NEXT:    addi.d $sp, $fp, -16
; LA64-NEXT:    ld.d $fp, $sp, 0 # 8-byte Folded Reload
; LA64-NEXT:    ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64-NEXT:    addi.d $sp, $sp, 16
; LA64-NEXT:    ret
;
; LA32-LABEL: dynamic:
; LA32:       # %bb.0:
; LA32-NEXT:    addi.w $sp, $sp, -16
; LA32-NEXT:    .cfi_def_cfa_offset 16
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
; LA32-NEXT:    .cfi_offset 1, -4
; LA32-NEXT:    .cfi_offset 22, -8
; LA32-NEXT:    addi.w $fp, $sp, 16
; LA32-NEXT:    .cfi_def_cfa 22, 0
; LA32-NEXT:    addi.w $a0, $a0, 15
; LA32-NEXT:    addi.w $a1, $zero, -16
; LA32-NEXT:    and $a0, $a0, $a1
; LA32-NEXT:    sub.w $a0, $sp, $a0
; LA32-NEXT:    lu12i.w $a1, 1
; LA32-NEXT:  .LBB0_1: # =>This Inner Loop Header: Depth=1
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bltu $a0, $sp, .LBB0_1
; LA32-NEXT:  # %bb.2:
; LA32-NEXT:    move $sp, $a0
; LA32-NEXT:    st.w $a0, $a2, 0
; LA32-NEXT:    addi.w $sp, $fp, -16
; LA32-NEXT:    ld.w $fp, $sp, 8 # 4-byte Folded Reload
; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload
; LA32-NEXT:    addi.w $sp, $sp, 16
; LA32-NEXT:    ret
  %v = alloca i8, i64 %size, align 1
  store ptr %v, ptr %out, align 8
  ret void
}

;; This function has a fixed-size stack slot and a dynamic one. The fixed size
;; slot isn't large enough that we would normally probe it, but we need to do so
;; here otherwise the gap between the CSR save and the first probe of the
;; dynamic allocation could be too far apart when the size of the dynamic
;; allocation is close to the guard size.
define void @dynamic_fixed(i64 %size, ptr %out1, ptr %out2) #0 {
;
; LA64-LABEL: dynamic_fixed:
; LA64:       # %bb.0:
; LA64-NEXT:    addi.d $sp, $sp, -96
; LA64-NEXT:    .cfi_def_cfa_offset 96
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    st.d $ra, $sp, 88 # 8-byte Folded Spill
; LA64-NEXT:    st.d $fp, $sp, 80 # 8-byte Folded Spill
; LA64-NEXT:    .cfi_offset 1, -8
; LA64-NEXT:    .cfi_offset 22, -16
; LA64-NEXT:    addi.d $fp, $sp, 96
; LA64-NEXT:    .cfi_def_cfa 22, 0
; LA64-NEXT:    addi.d $a3, $fp, -88
; LA64-NEXT:    st.d $a3, $a1, 0
; LA64-NEXT:    addi.d $a0, $a0, 15
; LA64-NEXT:    bstrins.d $a0, $zero, 3, 0
; LA64-NEXT:    sub.d $a0, $sp, $a0
; LA64-NEXT:    lu12i.w $a1, 1
; LA64-NEXT:  .LBB1_1: # =>This Inner Loop Header: Depth=1
; LA64-NEXT:    sub.d $sp, $sp, $a1
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    bltu $a0, $sp, .LBB1_1
; LA64-NEXT:  # %bb.2:
; LA64-NEXT:    move $sp, $a0
; LA64-NEXT:    st.d $a0, $a2, 0
; LA64-NEXT:    addi.d $sp, $fp, -96
; LA64-NEXT:    ld.d $fp, $sp, 80 # 8-byte Folded Reload
; LA64-NEXT:    ld.d $ra, $sp, 88 # 8-byte Folded Reload
; LA64-NEXT:    addi.d $sp, $sp, 96
; LA64-NEXT:    ret
;
; LA32-LABEL: dynamic_fixed:
; LA32:       # %bb.0:
; LA32-NEXT:    addi.w $sp, $sp, -80
; LA32-NEXT:    .cfi_def_cfa_offset 80
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    st.w $ra, $sp, 76 # 4-byte Folded Spill
; LA32-NEXT:    st.w $fp, $sp, 72 # 4-byte Folded Spill
; LA32-NEXT:    .cfi_offset 1, -4
; LA32-NEXT:    .cfi_offset 22, -8
; LA32-NEXT:    addi.w $fp, $sp, 80
; LA32-NEXT:    .cfi_def_cfa 22, 0
; LA32-NEXT:    addi.w $a1, $fp, -72
; LA32-NEXT:    st.w $a1, $a2, 0
; LA32-NEXT:    addi.w $a0, $a0, 15
; LA32-NEXT:    addi.w $a1, $zero, -16
; LA32-NEXT:    and $a0, $a0, $a1
; LA32-NEXT:    sub.w $a0, $sp, $a0
; LA32-NEXT:    lu12i.w $a1, 1
; LA32-NEXT:  .LBB1_1: # =>This Inner Loop Header: Depth=1
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bltu $a0, $sp, .LBB1_1
; LA32-NEXT:  # %bb.2:
; LA32-NEXT:    move $sp, $a0
; LA32-NEXT:    st.w $a0, $a3, 0
; LA32-NEXT:    addi.w $sp, $fp, -80
; LA32-NEXT:    ld.w $fp, $sp, 72 # 4-byte Folded Reload
; LA32-NEXT:    ld.w $ra, $sp, 76 # 4-byte Folded Reload
; LA32-NEXT:    addi.w $sp, $sp, 80
; LA32-NEXT:    ret
  %v1 = alloca i8, i64 64, align 1
  store ptr %v1, ptr %out1, align 8
  %v2 = alloca i8, i64 %size, align 1
  store ptr %v2, ptr %out2, align 8
  ret void
}

;; Dynamic allocation, with an alignment requirement greater than the alignment
;; of SP. Done by ANDing the target SP with a constant to align it down, then
;; doing the loop as normal. Note that we also re-align the stack in the prolog,
;; which isn't actually needed because the only aligned allocations are dynamic,
;; this is done even without stack probing.
define void @dynamic_align_64(i64 %size, ptr %out) #0 {
;
; LA64-LABEL: dynamic_align_64:
; LA64:       # %bb.0:
; LA64-NEXT:    addi.d $sp, $sp, -128
; LA64-NEXT:    .cfi_def_cfa_offset 128
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    st.d $ra, $sp, 120 # 8-byte Folded Spill
; LA64-NEXT:    st.d $fp, $sp, 112 # 8-byte Folded Spill
; LA64-NEXT:    st.d $s8, $sp, 104 # 8-byte Folded Spill
; LA64-NEXT:    .cfi_offset 1, -8
; LA64-NEXT:    .cfi_offset 22, -16
; LA64-NEXT:    .cfi_offset 31, -24
; LA64-NEXT:    addi.d $fp, $sp, 128
; LA64-NEXT:    .cfi_def_cfa 22, 0
; LA64-NEXT:    bstrins.d $sp, $zero, 5, 0
; LA64-NEXT:    move $s8, $sp
; LA64-NEXT:    addi.d $a0, $a0, 15
; LA64-NEXT:    bstrins.d $a0, $zero, 3, 0
; LA64-NEXT:    sub.d $a0, $sp, $a0
; LA64-NEXT:    bstrins.d $a0, $zero, 5, 0
; LA64-NEXT:    lu12i.w $a2, 1
; LA64-NEXT:  .LBB2_1: # =>This Inner Loop Header: Depth=1
; LA64-NEXT:    sub.d $sp, $sp, $a2
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    bltu $a0, $sp, .LBB2_1
; LA64-NEXT:  # %bb.2:
; LA64-NEXT:    move $sp, $a0
; LA64-NEXT:    st.d $a0, $a1, 0
; LA64-NEXT:    addi.d $sp, $fp, -128
; LA64-NEXT:    ld.d $s8, $sp, 104 # 8-byte Folded Reload
; LA64-NEXT:    ld.d $fp, $sp, 112 # 8-byte Folded Reload
; LA64-NEXT:    ld.d $ra, $sp, 120 # 8-byte Folded Reload
; LA64-NEXT:    addi.d $sp, $sp, 128
; LA64-NEXT:    ret
;
; LA32-LABEL: dynamic_align_64:
; LA32:       # %bb.0:
; LA32-NEXT:    addi.w $sp, $sp, -64
; LA32-NEXT:    .cfi_def_cfa_offset 64
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    st.w $ra, $sp, 60 # 4-byte Folded Spill
; LA32-NEXT:    st.w $fp, $sp, 56 # 4-byte Folded Spill
; LA32-NEXT:    st.w $s8, $sp, 52 # 4-byte Folded Spill
; LA32-NEXT:    .cfi_offset 1, -4
; LA32-NEXT:    .cfi_offset 22, -8
; LA32-NEXT:    .cfi_offset 31, -12
; LA32-NEXT:    addi.w $fp, $sp, 64
; LA32-NEXT:    .cfi_def_cfa 22, 0
; LA32-NEXT:    bstrins.w $sp, $zero, 5, 0
; LA32-NEXT:    move $s8, $sp
; LA32-NEXT:    addi.w $a0, $a0, 15
; LA32-NEXT:    addi.w $a1, $zero, -16
; LA32-NEXT:    and $a0, $a0, $a1
; LA32-NEXT:    sub.w $a0, $sp, $a0
; LA32-NEXT:    addi.w $a1, $zero, -64
; LA32-NEXT:    and $a0, $a0, $a1
; LA32-NEXT:    lu12i.w $a1, 1
; LA32-NEXT:  .LBB2_1: # =>This Inner Loop Header: Depth=1
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bltu $a0, $sp, .LBB2_1
; LA32-NEXT:  # %bb.2:
; LA32-NEXT:    move $sp, $a0
; LA32-NEXT:    st.w $a0, $a2, 0
; LA32-NEXT:    addi.w $sp, $fp, -64
; LA32-NEXT:    ld.w $s8, $sp, 52 # 4-byte Folded Reload
; LA32-NEXT:    ld.w $fp, $sp, 56 # 4-byte Folded Reload
; LA32-NEXT:    ld.w $ra, $sp, 60 # 4-byte Folded Reload
; LA32-NEXT:    addi.w $sp, $sp, 64
; LA32-NEXT:    ret
  %v = alloca i8, i64 %size, align 64
  store ptr %v, ptr %out, align 8
  ret void
}

;; Dynamic allocation, with an alignment greater than the stack guard size. The
;; only difference to the dynamic allocation is the constant used for aligning
;; the target SP, the loop will probe the whole allocation without needing to
;; know about the alignment padding.
define void @dynamic_align_8192(i64 %size, ptr %out) #0 {
;
; LA64-LABEL: dynamic_align_8192:
; LA64:       # %bb.0:
; LA64-NEXT:    addi.d $sp, $sp, -2032
; LA64-NEXT:    .cfi_def_cfa_offset 2032
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    st.d $ra, $sp, 2024 # 8-byte Folded Spill
; LA64-NEXT:    st.d $fp, $sp, 2016 # 8-byte Folded Spill
; LA64-NEXT:    st.d $s8, $sp, 2008 # 8-byte Folded Spill
; LA64-NEXT:    .cfi_offset 1, -8
; LA64-NEXT:    .cfi_offset 22, -16
; LA64-NEXT:    .cfi_offset 31, -24
; LA64-NEXT:    addi.d $fp, $sp, 2032
; LA64-NEXT:    .cfi_def_cfa 22, 0
; LA64-NEXT:    lu12i.w $a2, 1
; LA64-NEXT:    sub.d $sp, $sp, $a2
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    sub.d $sp, $sp, $a2
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    sub.d $sp, $sp, $a2
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    addi.d $sp, $sp, -2048
; LA64-NEXT:    addi.d $sp, $sp, -16
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    bstrins.d $sp, $zero, 12, 0
; LA64-NEXT:    move $s8, $sp
; LA64-NEXT:    addi.d $a0, $a0, 15
; LA64-NEXT:    bstrins.d $a0, $zero, 3, 0
; LA64-NEXT:    sub.d $a0, $sp, $a0
; LA64-NEXT:    bstrins.d $a0, $zero, 12, 0
; LA64-NEXT:  .LBB3_1: # =>This Inner Loop Header: Depth=1
; LA64-NEXT:    sub.d $sp, $sp, $a2
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    bltu $a0, $sp, .LBB3_1
; LA64-NEXT:  # %bb.2:
; LA64-NEXT:    move $sp, $a0
; LA64-NEXT:    st.d $a0, $a1, 0
; LA64-NEXT:    lu12i.w $a0, 4
; LA64-NEXT:    sub.d $sp, $fp, $a0
; LA64-NEXT:    lu12i.w $a0, 3
; LA64-NEXT:    ori $a0, $a0, 2064
; LA64-NEXT:    add.d $sp, $sp, $a0
; LA64-NEXT:    ld.d $s8, $sp, 2008 # 8-byte Folded Reload
; LA64-NEXT:    ld.d $fp, $sp, 2016 # 8-byte Folded Reload
; LA64-NEXT:    ld.d $ra, $sp, 2024 # 8-byte Folded Reload
; LA64-NEXT:    addi.d $sp, $sp, 2032
; LA64-NEXT:    ret
;
; LA32-LABEL: dynamic_align_8192:
; LA32:       # %bb.0:
; LA32-NEXT:    addi.w $sp, $sp, -2032
; LA32-NEXT:    .cfi_def_cfa_offset 2032
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    st.w $ra, $sp, 2028 # 4-byte Folded Spill
; LA32-NEXT:    st.w $fp, $sp, 2024 # 4-byte Folded Spill
; LA32-NEXT:    st.w $s8, $sp, 2020 # 4-byte Folded Spill
; LA32-NEXT:    .cfi_offset 1, -4
; LA32-NEXT:    .cfi_offset 22, -8
; LA32-NEXT:    .cfi_offset 31, -12
; LA32-NEXT:    addi.w $fp, $sp, 2032
; LA32-NEXT:    .cfi_def_cfa 22, 0
; LA32-NEXT:    lu12i.w $a1, 1
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    addi.w $sp, $sp, -2048
; LA32-NEXT:    addi.w $sp, $sp, -16
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bstrins.w $sp, $zero, 12, 0
; LA32-NEXT:    move $s8, $sp
; LA32-NEXT:    addi.w $a0, $a0, 15
; LA32-NEXT:    addi.w $a1, $zero, -16
; LA32-NEXT:    and $a0, $a0, $a1
; LA32-NEXT:    sub.w $a0, $sp, $a0
; LA32-NEXT:    lu12i.w $a1, -2
; LA32-NEXT:    and $a0, $a0, $a1
; LA32-NEXT:    lu12i.w $a1, 1
; LA32-NEXT:  .LBB3_1: # =>This Inner Loop Header: Depth=1
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bltu $a0, $sp, .LBB3_1
; LA32-NEXT:  # %bb.2:
; LA32-NEXT:    move $sp, $a0
; LA32-NEXT:    st.w $a0, $a2, 0
; LA32-NEXT:    lu12i.w $a0, 4
; LA32-NEXT:    sub.w $sp, $fp, $a0
; LA32-NEXT:    lu12i.w $a0, 3
; LA32-NEXT:    ori $a0, $a0, 2064
; LA32-NEXT:    add.w $sp, $sp, $a0
; LA32-NEXT:    ld.w $s8, $sp, 2020 # 4-byte Folded Reload
; LA32-NEXT:    ld.w $fp, $sp, 2024 # 4-byte Folded Reload
; LA32-NEXT:    ld.w $ra, $sp, 2028 # 4-byte Folded Reload
; LA32-NEXT:    addi.w $sp, $sp, 2032
; LA32-NEXT:    ret
  %v = alloca i8, i64 %size, align 8192
  store ptr %v, ptr %out, align 8
  ret void
}

;; If a function has variable-sized stack objects, then any function calls which
;; need to pass arguments on the stack must allocate the stack space for them
;; dynamically, to ensure they are at the bottom of the frame.
define void @no_reserved_call_frame(i64 %n) #0 {
;
; LA64-LABEL: no_reserved_call_frame:
; LA64:       # %bb.0: # %entry
; LA64-NEXT:    addi.d $sp, $sp, -16
; LA64-NEXT:    .cfi_def_cfa_offset 16
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    st.d $ra, $sp, 8 # 8-byte Folded Spill
; LA64-NEXT:    st.d $fp, $sp, 0 # 8-byte Folded Spill
; LA64-NEXT:    .cfi_offset 1, -8
; LA64-NEXT:    .cfi_offset 22, -16
; LA64-NEXT:    addi.d $fp, $sp, 16
; LA64-NEXT:    .cfi_def_cfa 22, 0
; LA64-NEXT:    slli.d $a0, $a0, 2
; LA64-NEXT:    addi.d $a0, $a0, 15
; LA64-NEXT:    bstrins.d $a0, $zero, 3, 0
; LA64-NEXT:    sub.d $a0, $sp, $a0
; LA64-NEXT:    lu12i.w $a1, 1
; LA64-NEXT:  .LBB4_1: # %entry
; LA64-NEXT:    # =>This Inner Loop Header: Depth=1
; LA64-NEXT:    sub.d $sp, $sp, $a1
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    bltu $a0, $sp, .LBB4_1
; LA64-NEXT:  # %bb.2: # %entry
; LA64-NEXT:    move $sp, $a0
; LA64-NEXT:    lu12i.w $a1, 1
; LA64-NEXT:    sub.d $sp, $sp, $a1
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    pcaddu18i $ra, %call36(callee_stack_args)
; LA64-NEXT:    jirl $ra, $ra, 0
; LA64-NEXT:    lu12i.w $a0, 1
; LA64-NEXT:    add.d $sp, $sp, $a0
; LA64-NEXT:    addi.d $sp, $fp, -16
; LA64-NEXT:    ld.d $fp, $sp, 0 # 8-byte Folded Reload
; LA64-NEXT:    ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64-NEXT:    addi.d $sp, $sp, 16
; LA64-NEXT:    ret
;
; LA32-LABEL: no_reserved_call_frame:
; LA32:       # %bb.0: # %entry
; LA32-NEXT:    addi.w $sp, $sp, -16
; LA32-NEXT:    .cfi_def_cfa_offset 16
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
; LA32-NEXT:    .cfi_offset 1, -4
; LA32-NEXT:    .cfi_offset 22, -8
; LA32-NEXT:    addi.w $fp, $sp, 16
; LA32-NEXT:    .cfi_def_cfa 22, 0
; LA32-NEXT:    slli.w $a0, $a0, 2
; LA32-NEXT:    addi.w $a0, $a0, 15
; LA32-NEXT:    addi.w $a1, $zero, -16
; LA32-NEXT:    and $a0, $a0, $a1
; LA32-NEXT:    sub.w $a0, $sp, $a0
; LA32-NEXT:    lu12i.w $a1, 1
; LA32-NEXT:  .LBB4_1: # %entry
; LA32-NEXT:    # =>This Inner Loop Header: Depth=1
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bltu $a0, $sp, .LBB4_1
; LA32-NEXT:  # %bb.2: # %entry
; LA32-NEXT:    move $sp, $a0
; LA32-NEXT:    lu12i.w $a1, 1
; LA32-NEXT:    sub.w $sp, $sp, $a1
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    addi.w $sp, $sp, -32
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bl callee_stack_args
; LA32-NEXT:    lu12i.w $a0, 1
; LA32-NEXT:    ori $a0, $a0, 32
; LA32-NEXT:    add.w $sp, $sp, $a0
; LA32-NEXT:    addi.w $sp, $fp, -16
; LA32-NEXT:    ld.w $fp, $sp, 8 # 4-byte Folded Reload
; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload
; LA32-NEXT:    addi.w $sp, $sp, 16
; LA32-NEXT:    ret
entry:
  %v = alloca i32, i64 %n
  call void @callee_stack_args(ptr %v, [518 x i64] poison)
  ret void
}

;; Same as above but without a variable-sized allocation, so the reserved call
;; frame can be folded into the fixed-size allocation in the prologue.
define void @reserved_call_frame(i64 %n) #0 {
;
; LA64-LABEL: reserved_call_frame:
; LA64:       # %bb.0: # %entry
; LA64-NEXT:    addi.d $sp, $sp, -2032
; LA64-NEXT:    .cfi_def_cfa_offset 2032
; LA64-NEXT:    st.d $ra, $sp, 2024 # 8-byte Folded Spill
; LA64-NEXT:    .cfi_offset 1, -8
; LA64-NEXT:    lu12i.w $a0, 1
; LA64-NEXT:    sub.d $sp, $sp, $a0
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    .cfi_def_cfa_offset 6128
; LA64-NEXT:    addi.d $sp, $sp, -48
; LA64-NEXT:    .cfi_def_cfa_offset 6176
; LA64-NEXT:    lu12i.w $a0, 1
; LA64-NEXT:    add.d $a0, $sp, $a0
; LA64-NEXT:    pcaddu18i $ra, %call36(callee_stack_args)
; LA64-NEXT:    jirl $ra, $ra, 0
; LA64-NEXT:    lu12i.w $a0, 1
; LA64-NEXT:    ori $a0, $a0, 48
; LA64-NEXT:    add.d $sp, $sp, $a0
; LA64-NEXT:    ld.d $ra, $sp, 2024 # 8-byte Folded Reload
; LA64-NEXT:    addi.d $sp, $sp, 2032
; LA64-NEXT:    ret
;
; LA32-LABEL: reserved_call_frame:
; LA32:       # %bb.0: # %entry
; LA32-NEXT:    addi.w $sp, $sp, -2032
; LA32-NEXT:    .cfi_def_cfa_offset 2032
; LA32-NEXT:    st.w $ra, $sp, 2028 # 4-byte Folded Spill
; LA32-NEXT:    .cfi_offset 1, -4
; LA32-NEXT:    lu12i.w $a0, 1
; LA32-NEXT:    sub.w $sp, $sp, $a0
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    .cfi_def_cfa_offset 6128
; LA32-NEXT:    addi.w $sp, $sp, -80
; LA32-NEXT:    .cfi_def_cfa_offset 6208
; LA32-NEXT:    lu12i.w $a0, 1
; LA32-NEXT:    ori $a0, $a0, 36
; LA32-NEXT:    add.w $a0, $sp, $a0
; LA32-NEXT:    bl callee_stack_args
; LA32-NEXT:    lu12i.w $a0, 1
; LA32-NEXT:    ori $a0, $a0, 80
; LA32-NEXT:    add.w $sp, $sp, $a0
; LA32-NEXT:    ld.w $ra, $sp, 2028 # 4-byte Folded Reload
; LA32-NEXT:    addi.w $sp, $sp, 2032
; LA32-NEXT:    ret
entry:
  %v = alloca i32, i64 518
  call void @callee_stack_args(ptr %v, [518 x i64] poison)
  ret void
}

declare void @callee_stack_args(ptr, [518 x i64])


attributes #0 = { uwtable(async) "probe-stack"="inline-asm" "frame-pointer"="none" }