[X86] Check immediate before get it.

[LuoYuanke]

Summary

For CMP imm instruction, when the operand 1 is symbol address we should
check if it is immediate first. Here is the example code.
CMP64mi32 $noreg, 8, killed renamable $rcx, @d, $noreg, @a, implicit-def $eflags
However I am not able to create a small test case to expose this issue.

[lebedev.ri]

llvm-reduce should get you 99% of the way there.

[LuoYuanke]

Thank @lebedev.ri. This is triggered by our internal code, the same test case passes with llvm trunk code. BTW, do you think we need to check immediate before getting it?

[pengfei]

llvm/lib/Target/X86/X86InstrInfo.cpp

6410

But CMP64mi32 shouldn't go to here.

[craig.topper]

It would be nice to have a test, but this change seems ok.

llvm/lib/Target/X86/X86InstrInfo.cpp

6410

The CMPmi32 will be expanded to a load and CMPri32. This switches use the opcode after it's expanded.

[pengfei]

LGTM but let's wait one or two days to see opinions from others.

llvm/lib/Target/X86/X86InstrInfo.cpp

6410

Thanks Craig.
Discussed with Yuanke offline, I understand the fix here is to handle relocImm, i.e., @a in the example.
I think it makes sense to skip it here since we don't know it's value at this point.

[lebedev.ri]

It isn't immediately obvious to me how that can happen.
I don't think this should land without a test.

[pengfei]

Maybe we can use an MIR test?

[craig.topper]

I don't what Intel's original failure looked like, but here's a test that should reproduce this with -run-pass=machinelicm https://reviews.llvm.org/P8267 needs more cleanup.

I hacked the MIR just before machinelicm by sinking the CMP64mi32 and SETCCr into the loop. That makes MachineLICM want to unfold it since the load part is invariant being from a constant global.

[LuoYuanke]

Apply Craig's test case. Many thanks to Craig. :)

[LuoYuanke]

In D104037#2813726, @craig.topper wrote:

I don't what Intel's original failure looked like, but here's a test that should reproduce this with -run-pass=machinelicm https://reviews.llvm.org/P8267 needs more cleanup.

I hacked the MIR just before machinelicm by sinking the CMP64mi32 and SETCCr into the loop. That makes MachineLICM want to unfold it since the load part is invariant being from a constant global.

Hi Craig,

Awesome. Many thanks for the test case. I applied your test case to this patch.

[LuoYuanke]

Update test case to pass expensive check.

llvm/lib/Target/X86/X86InstrInfo.cpp

bool X86InstrInfo::unfoldMemoryOperand(
   default: break;
   case X86::CMP64ri32:
   case X86::CMP64ri8:
   case X86::CMP32ri:
   case X86::CMP32ri8:
   case X86::CMP16ri:
   case X86::CMP16ri8:
   case X86::CMP8ri: {
     MachineOperand &MO0 = DataMI->getOperand(0);
     MachineOperand &MO1 = DataMI->getOperand(1);
-    if (MO1.getImm() == 0) {
+    if (MO1.isImm() && MO1.getImm() == 0) {
       unsigned NewOpc;
       switch (DataMI->getOpcode()) {
       default: llvm_unreachable("Unreachable!");
       case X86::CMP64ri8:
       case X86::CMP64ri32: NewOpc = X86::TEST64rr; break;
       case X86::CMP32ri8:
       case X86::CMP32ri:   NewOpc = X86::TEST32rr; break;
       case X86::CMP16ri8:

llvm/test/CodeGen/X86/unfoldMemoryOperand.mir

+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=x86_64-- -run-pass machinelicm -mcpu=skx -verify-machineinstrs -o - %s | FileCheck %s
+--- |
+  @x = dso_local global i32 0, align 4
+  @z = dso_local local_unnamed_addr global [1024 x i32] zeroinitializer, align 16
+  @y = dso_local local_unnamed_addr constant i32* null, align 8
+
+  ; Function Attrs: nofree norecurse nosync nounwind uwtable writeonly mustprogress
+  define dso_local void @_Z3foov() local_unnamed_addr #0 {
+    %1 = load i32*, i32** @y, align 8, !tbaa !3
+    %2 = icmp eq i32* %1, @x
+    %3 = zext i1 %2 to i32
+    br label %5
+  4:                                                ; preds = %5
+    ret void
+  5:                                                ; preds = %5, %0
+    %lsr.iv = phi i64 [ %lsr.iv.next, %5 ], [ -4096, %0 ]
+    %uglygep = getelementptr i8, i8* bitcast ([1024 x i32]* @z to i8*), i64 %lsr.iv
+    %uglygep2 = bitcast i8* %uglygep to i32*
+    %scevgep = getelementptr i32, i32* %uglygep2, i64 1024
+    store i32 %3, i32* %scevgep, align 4, !tbaa !7
+    %lsr.iv.next = add nsw i64 %lsr.iv, 4
+    %6 = icmp eq i64 %lsr.iv.next, 0
+    br i1 %6, label %4, label %5, !llvm.loop !9
+  }
+
+  attributes #0 = { nofree norecurse nosync nounwind uwtable writeonly mustprogress "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+x87,-aes,-avx,-avx2,-avx512bf16,-avx512bitalg,-avx512bw,-avx512cd,-avx512dq,-avx512er,-avx512f,-avx512ifma,-avx512pf,-avx512vbmi,-avx512vbmi2,-avx512vl,-avx512vnni,-avx512vp2intersect,-avx512vpopcntdq,-avxvnni,-f16c,-fma,-fma4,-gfni,-kl,-pclmul,-sha,-sse,-sse2,-sse3,-sse4.1,-sse4.2,-sse4a,-ssse3,-vaes,-vpclmulqdq,-widekl,-xop" "tune-cpu"="generic" }
+
+  !llvm.module.flags = !{!0, !1}
+  !llvm.ident = !{!2}
+  !0 = !{i32 1, !"wchar_size", i32 4}
+  !1 = !{i32 7, !"uwtable", i32 1}
+  !2 = !{!"clang version 13.0.0 (https://github.com/llvm/llvm-project.git c42dd5dbb015afaef99cf876195c474c63c2393e)"}
+  !3 = !{!4, !4, i64 0}
+  !4 = !{!"any pointer", !5, i64 0}
+  !5 = !{!"omnipotent char", !6, i64 0}
+  !6 = !{!"Simple C++ TBAA"}
+  !7 = !{!8, !8, i64 0}
+  !8 = !{!"int", !5, i64 0}
+  !9 = distinct !{!9, !10, !11}
+  !10 = !{!"llvm.loop.mustprogress"}
+  !11 = !{!"llvm.loop.unroll.disable"}
+
+...
+---
+name:            _Z3foov
+alignment:       16
+exposesReturnsTwice: false
+legalized:       false
+regBankSelected: false
+selected:        false
+failedISel:      false
+tracksRegLiveness: true
+hasWinCFI:       false
+registers:       []
+liveins:         []
+frameInfo:
+  isFrameAddressTaken: false
+  isReturnAddressTaken: false
+  hasStackMap:     false
+  hasPatchPoint:   false
+  stackSize:       0
+  offsetAdjustment: 0
+  maxAlignment:    1
+  adjustsStack:    false
+  hasCalls:        false
+  stackProtector:  ''
+  maxCallFrameSize: 4294967295
+  cvBytesOfCalleeSavedRegisters: 0
+  hasOpaqueSPAdjustment: false
+  hasVAStart:      false
+  hasMustTailInVarArgFunc: false
+  hasTailCall:     false
+  localFrameSize:  0
+  savePoint:       ''
+  restorePoint:    ''
+fixedStack:      []
+stack:           []
+callSites:       []
+debugValueSubstitutions: []
+constants:       []
+machineFunctionInfo: {}
+body:             |
+  ; CHECK-LABEL: name: _Z3foov
+  ; CHECK: bb.0 (%ir-block.0):
+  ; CHECK:   successors: %bb.2(0x80000000)
+  ; CHECK:   renamable $eax = MOV32r0 implicit-def dead $eflags
+  ; CHECK:   renamable $rcx = MOV64ri32 -4096
+  ; CHECK:   [[MOV64ri32_:%[0-9]+]]:gr64 = MOV64ri32 -4096
+  ; CHECK:   [[MOV64rm:%[0-9]+]]:gr64 = MOV64rm $rip, 1, $noreg, @y, $noreg :: (dereferenceable load 8 from @y, !tbaa !3)
+  ; CHECK:   JMP_1 %bb.2
+  ; CHECK: bb.1 (%ir-block.4):
+  ; CHECK:   RET 0
+  ; CHECK: bb.2 (%ir-block.5):
+  ; CHECK:   successors: %bb.1(0x04000000), %bb.2(0x7c000000)
+  ; CHECK:   liveins: $eax, $rcx
+  ; CHECK:   CMP64ri32 [[MOV64rm]], @x, implicit-def $eflags
+  ; CHECK:   renamable $al = SETCCr 4, implicit killed $eflags, implicit killed $eax, implicit-def $eax
+  ; CHECK:   MOV32mr renamable $rcx, 1, $noreg, @z + 4096, $noreg, renamable $eax :: (store 4 into %ir.scevgep, !tbaa !7)
+  ; CHECK:   renamable $rcx = ADD64ri8 killed renamable $rcx, 4, implicit-def $eflags
+  ; CHECK:   JCC_1 %bb.1, 4, implicit killed $eflags
+  ; CHECK:   JMP_1 %bb.2
+  bb.0 (%ir-block.0):
+    successors: %bb.2(0x80000000)
+    renamable $eax = MOV32r0 implicit-def dead $eflags
+    renamable $rcx = MOV64ri32 -4096
+    JMP_1 %bb.2
+  bb.1 (%ir-block.4):
+    RET 0
+  bb.2 (%ir-block.5):
+    successors: %bb.1(0x04000000), %bb.2(0x7c000000)
+    liveins: $eax, $rcx
+    %2:gr64 = MOV64ri32 -4096
+    CMP64mi32 $rip, 1, $noreg, @y, $noreg, @x, implicit-def $eflags :: (dereferenceable load 8 from @y, !tbaa !3)
+    renamable $al = SETCCr 4, implicit killed $eflags, implicit killed $eax, implicit-def $eax
+    MOV32mr renamable $rcx, 1, $noreg, @z + 4096, $noreg, renamable $eax :: (store 4 into %ir.scevgep, !tbaa !7)
+    renamable $rcx = ADD64ri8 killed renamable $rcx, 4, implicit-def $eflags
+    JCC_1 %bb.1, 4, implicit killed $eflags
+    JMP_1 %bb.2
+
+...