; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=loongarch64 -O2 < %s -verify-machineinstrs | FileCheck %s -check-prefix=LA64
; RUN: llc -mtriple=loongarch32 -O2 < %s -verify-machineinstrs | FileCheck %s -check-prefix=LA32

;; Test that very large outgoing call frames in functions with variable-sized
;; objects get proper stack probing. The outgoing args are large enough to force
;; the PROBED_STACKALLOC path, which must be expanded in a non-entry block.

define void @f(i64 %n) #0 {
; LA64-LABEL: f:
; 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:  .LBB0_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, .LBB0_1
; LA64-NEXT:  # %bb.2: # %entry
; LA64-NEXT:    move $sp, $a0
; LA64-NEXT:    lu12i.w $a1, 5
; LA64-NEXT:    sub.d $t1, $sp, $a1
; LA64-NEXT:    lu12i.w $t2, 1
; LA64-NEXT:  .LBB0_3: # %entry
; LA64-NEXT:    # =>This Inner Loop Header: Depth=1
; LA64-NEXT:    sub.d $sp, $sp, $t2
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    bne $sp, $t1, .LBB0_3
; LA64-NEXT:  # %bb.4: # %entry
; LA64-NEXT:    addi.d $sp, $sp, -2048
; LA64-NEXT:    addi.d $sp, $sp, -1424
; LA64-NEXT:    st.d $zero, $sp, 0
; LA64-NEXT:    pcaddu18i $ra, %call36(g)
; LA64-NEXT:    jirl $ra, $ra, 0
; LA64-NEXT:    lu12i.w $a0, 5
; LA64-NEXT:    ori $a0, $a0, 3472
; 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: f:
; 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:  .LBB0_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, .LBB0_1
; LA32-NEXT:  # %bb.2: # %entry
; LA32-NEXT:    move $sp, $a0
; LA32-NEXT:    lu12i.w $a1, 5
; LA32-NEXT:    sub.w $t1, $sp, $a1
; LA32-NEXT:    lu12i.w $t2, 1
; LA32-NEXT:  .LBB0_3: # %entry
; LA32-NEXT:    # =>This Inner Loop Header: Depth=1
; LA32-NEXT:    sub.w $sp, $sp, $t2
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bne $sp, $t1, .LBB0_3
; LA32-NEXT:  # %bb.4: # %entry
; LA32-NEXT:    addi.w $sp, $sp, -2048
; LA32-NEXT:    addi.w $sp, $sp, -1456
; LA32-NEXT:    st.w $zero, $sp, 0
; LA32-NEXT:    bl g
; LA32-NEXT:    lu12i.w $a0, 5
; LA32-NEXT:    ori $a0, $a0, 3504
; 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 @g(ptr %v, [3000 x i64] poison)
  ret void
}

declare void @g(ptr, [3000 x i64])

attributes #0 = { "probe-stack"="inline-asm" }