[PowerPC][NFC] adding an option to keep the test point

[shchenz]

Summary

Passes before hardware loop insertion change the loop to a form which
is not a hardware loop candidate (return early before checking the ctr clobbers).
And the PHI in the loop exit block is also optimized away. This breaks the
previous test point when the case was committed. Fixing this by running this
case just before hardware loop insertion pass.

[shchenz]

gentle ping

[nemanjai]

LGTM.

llvm/test/CodeGen/PowerPC/pr48527.ll

2

Please add a note such as:

; NOTE: this needs to run starting at HW loops to ensure that
; the original loop does not end up producing a HW loop. If
; other passes are run prior, the loop will be transformed.

llvm/test/CodeGen/PowerPC/pr48527.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -relocation-model=pic -verify-machineinstrs < %s \
+; RUN: llc -relocation-model=pic -verify-machineinstrs -start-before=hardware-loops < %s \
 ; RUN:   -mtriple=powerpc64le-unknown-linux-gnu -mcpu=pwr8 | FileCheck %s
+
+; NOTE: this needs to run starting at HW loops to ensure that the original loop
+; does not end up producing a HW loop. If other passes are run prior, the loop
+; will be transformed.
+
 %struct.e.0.12.28.44.104.108.112.188 = type { i32 }
 %struct.t.1.13.29.45.105.109.113.189 = type { i64, i64 }
 
 @g = external local_unnamed_addr global %struct.e.0.12.28.44.104.108.112.188, align 4
 @aj = external thread_local local_unnamed_addr global %struct.t.1.13.29.45.105.109.113.189, align 8
 
 define void @_ZNK1q1rEv() local_unnamed_addr #0 align 2 {
 ; CHECK-LABEL: _ZNK1q1rEv:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    mflr 0
+; CHECK-NEXT:    std 28, -32(1) # 8-byte Folded Spill
 ; CHECK-NEXT:    std 29, -24(1) # 8-byte Folded Spill
 ; CHECK-NEXT:    std 30, -16(1) # 8-byte Folded Spill
 ; CHECK-NEXT:    stdu 1, -64(1)
 ; CHECK-NEXT:    std 0, 80(1)
+; CHECK-NEXT:    li 29, 0
 ; CHECK-NEXT:    lwz 30, 0(3)
 ; CHECK-NEXT:    addis 3, 2, .LC0@toc@ha
-; CHECK-NEXT:    ld 29, .LC0@toc@l(3)
+; CHECK-NEXT:    ld 28, .LC0@toc@l(3)
 ; CHECK-NEXT:    addis 3, 2, aj@got@tlsgd@ha
 ; CHECK-NEXT:    addi 3, 3, aj@got@tlsgd@l
 ; CHECK-NEXT:    bl __tls_get_addr(aj@tlsgd)
 ; CHECK-NEXT:    nop
 ; CHECK-NEXT:    addi 4, 3, 8
 ; CHECK-NEXT:    .p2align 5
 ; CHECK-NEXT:  .LBB0_1: # %monotonic.i
 ; CHECK-NEXT:    #
-; CHECK-NEXT:    lwz 5, 0(29)
+; CHECK-NEXT:    lwz 5, 0(28)
 ; CHECK-NEXT:    andi. 5, 5, 255
 ; CHECK-NEXT:    bne 0, .LBB0_4
 ; CHECK-NEXT:  # %bb.2: # %for.cond.i
 ; CHECK-NEXT:    #
-; CHECK-NEXT:    addi 30, 30, -1
-; CHECK-NEXT:    cmplwi 30, 0
+; CHECK-NEXT:    addi 29, 29, 1
+; CHECK-NEXT:    cmplw 29, 30
 ; CHECK-NEXT:    bne 0, .LBB0_1
 ; CHECK-NEXT:  # %bb.3:
 ; CHECK-NEXT:    mr 4, 3
 ; CHECK-NEXT:  .LBB0_4: # %if.end
 ; CHECK-NEXT:    li 3, 1
 ; CHECK-NEXT:    std 3, 0(4)
 ; CHECK-NEXT:    addi 1, 1, 64
 ; CHECK-NEXT:    ld 0, 16(1)
 ; CHECK-NEXT:    ld 30, -16(1) # 8-byte Folded Reload
 ; CHECK-NEXT:    ld 29, -24(1) # 8-byte Folded Reload
+; CHECK-NEXT:    ld 28, -32(1) # 8-byte Folded Reload
 ; CHECK-NEXT:    mtlr 0
 ; CHECK-NEXT:    blr
 entry:
   %0 = load i32, ptr undef, align 4
   br label %monotonic.i
 
 for.cond.i:                                       ; preds = %monotonic.i
   %exitcond.not = icmp eq i32 %inc.i, %0