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

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.

Reviewed By: nemanjai

Differential revision: https://reviews.llvm.org/D140366

diff --git a/llvm/test/CodeGen/PowerPC/pr48527.ll b/llvm/test/CodeGen/PowerPC/pr48527.ll
index 262f7a9805fdc04613108bd130b7daaba6487fd5..f0beead0e4b651f6014c43bbbdbc330845150f33 100644 (file)
--- a/llvm/test/CodeGen/PowerPC/pr48527.ll
+++ b/llvm/test/CodeGen/PowerPC/pr48527.ll
@@ -1,6 +1,11 @@
 ; 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 }
 
@@ -11,13 +16,15 @@ 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)
@@ -26,13 +33,13 @@ define void @_ZNK1q1rEv() local_unnamed_addr #0 align 2 {
 ; 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
@@ -43,6 +50,7 @@ define void @_ZNK1q1rEv() local_unnamed_addr #0 align 2 {
 ; 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: