commit 31e75512174e1bdaa242ee5c7f30fe56e68c3748
Author: Stephen Tozer <Stephen.Tozer@Sony.com>
Date:   Mon Jul 26 09:59:20 2021 +0100

    [DebugInfo] Correctly update debug users of SSA values in tail duplication
    
    During tail duplication, SSA values may be updated and have their uses
    replaced with a virtual register, and any debug instructions that use
    that value are deleted. This patch fixes the implementation of the debug
    instruction deletion to work correctly for debug instructions that use
    the SSA value multiple times, by batching deletions so that we don't
    attempt to delete the same instruction twice.
    
    Differential Revision: https://reviews.llvm.org/D106557

diff --git a/llvm/lib/CodeGen/TailDuplicator.cpp b/llvm/lib/CodeGen/TailDuplicator.cpp
index 2b71490d67a8..af735f2a0216 100644
--- a/llvm/lib/CodeGen/TailDuplicator.cpp
+++ b/llvm/lib/CodeGen/TailDuplicator.cpp
@@ -216,6 +216,9 @@ bool TailDuplicator::tailDuplicateAndUpdate(
 
       // Rewrite uses that are outside of the original def's block.
       MachineRegisterInfo::use_iterator UI = MRI->use_begin(VReg);
+      // Only remove instructions after loop, as DBG_VALUE_LISTs with multiple
+      // uses of VReg may invalidate the use iterator when erased.
+      SmallPtrSet<MachineInstr *, 4> InstrsToRemove;
       while (UI != MRI->use_end()) {
         MachineOperand &UseMO = *UI;
         MachineInstr *UseMI = UseMO.getParent();
@@ -225,13 +228,15 @@ bool TailDuplicator::tailDuplicateAndUpdate(
           // a debug instruction that is a kill.
           // FIXME: Should it SSAUpdate job to delete debug instructions
           // instead of replacing the use with undef?
-          UseMI->eraseFromParent();
+          InstrsToRemove.insert(UseMI);
           continue;
         }
         if (UseMI->getParent() == DefBB && !UseMI->isPHI())
           continue;
         SSAUpdate.RewriteUse(UseMO);
       }
+      for (auto *MI : InstrsToRemove)
+        MI->eraseFromParent();
     }
 
     SSAUpdateVRs.clear();
diff --git a/llvm/test/CodeGen/X86/tail-dup-debugvalue.mir b/llvm/test/CodeGen/X86/tail-dup-debugvalue.mir
new file mode 100644
index 000000000000..835d276c9928
--- /dev/null
+++ b/llvm/test/CodeGen/X86/tail-dup-debugvalue.mir
@@ -0,0 +1,127 @@
+# RUN: llc -run-pass=early-tailduplication -mtriple=x86_64-unknown-linux-gnu %s -o - | FileCheck %s
+
+# Tail Duplication may update SSA values and invalidate any DBG_VALUEs that
+# use those values; those DBG_VALUEs should be deleted. This behaviour is tested
+# for DBG_VALUE users, and DBG_VALUE_LISTs that use the value multiple times.
+
+# CHECK-NOT: DBG_VALUE
+--- |
+  target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+  target triple = "x86_64-unknown-linux-gnu"
+  
+  define dso_local void @main() local_unnamed_addr #0 !dbg !15 {
+  entry:
+    br label %L.outer
+  
+  L:                                                ; preds = %L, %L.outer
+    %tobool2.not = icmp eq i32 undef, 0
+    br i1 %tobool2.not, label %if.end4, label %L
+  
+  if.end4:                                          ; preds = %L
+    call void @llvm.dbg.value(metadata !DIArgList(i32 %f.0.ph, i32 1, i32 %f.0.ph), metadata !19, metadata !DIExpression(DW_OP_LLVM_arg, 0, DW_OP_LLVM_arg, 1, DW_OP_LLVM_arg, 2, DW_OP_and, DW_OP_or, DW_OP_stack_value)), !dbg !21
+    %cmp = icmp slt i32 %f.0.ph, undef
+    br i1 %cmp, label %if.then5, label %if.end6
+  
+  if.then5:                                         ; preds = %if.end4
+    call void @h() #2
+    br label %L.outer
+  
+  L.outer:                                          ; preds = %if.then5, %entry
+    %f.0.ph = phi i32 [ 0, %if.then5 ], [ 1, %entry ]
+    br label %L
+  
+  if.end6:                                          ; preds = %if.end4
+    ret void
+  }
+  
+  declare dso_local void @h() local_unnamed_addr
+  
+  declare void @llvm.dbg.value(metadata, metadata, metadata)
+  
+  !llvm.dbg.cu = !{!0}
+  !llvm.module.flags = !{!13, !14}
+  
+  !0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 13.0.0", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, globals: !3, splitDebugInlining: false, nameTableKind: None)
+  !1 = !DIFile(filename: "tail-dup-debugvalue.c", directory: "/tmp")
+  !2 = !{}
+  !3 = !{!4, !7, !9, !11}
+  !4 = !DIGlobalVariableExpression(var: !5, expr: !DIExpression())
+  !5 = distinct !DIGlobalVariable(name: "a", scope: !0, file: !1, line: 2, type: !6, isLocal: false, isDefinition: true)
+  !6 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
+  !7 = !DIGlobalVariableExpression(var: !8, expr: !DIExpression())
+  !8 = distinct !DIGlobalVariable(name: "b", scope: !0, file: !1, line: 2, type: !6, isLocal: false, isDefinition: true)
+  !9 = !DIGlobalVariableExpression(var: !10, expr: !DIExpression())
+  !10 = distinct !DIGlobalVariable(name: "c", scope: !0, file: !1, line: 2, type: !6, isLocal: false, isDefinition: true)
+  !11 = !DIGlobalVariableExpression(var: !12, expr: !DIExpression())
+  !12 = distinct !DIGlobalVariable(name: "d", scope: !0, file: !1, line: 2, type: !6, isLocal: false, isDefinition: true)
+  !13 = !{i32 2, !"Debug Info Version", i32 3}
+  !14 = !{i32 7, !"uwtable", i32 1}
+  !15 = distinct !DISubprogram(name: "main", scope: !1, file: !1, line: 3, type: !16, scopeLine: 3, flags: DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !18)
+  !16 = !DISubroutineType(types: !17)
+  !17 = !{!6}
+  !18 = !{!19, !20}
+  !19 = !DILocalVariable(name: "j", scope: !15, file: !1, line: 14, type: !6)
+  !20 = !DILocalVariable(name: "k", scope: !15, file: !1, line: 14, type: !6)
+  !21 = !DILocation(line: 0, scope: !15)
+
+...
+---
+name:            main
+alignment:       16
+tracksRegLiveness: true
+registers:
+  - { id: 0, class: gr32 }
+  - { id: 1, class: gr32 }
+  - { id: 2, class: gr32 }
+  - { id: 3, class: gr8 }
+  - { id: 4, class: gr32 }
+  - { id: 5, class: gr32 }
+  - { id: 6, class: gr32 }
+frameInfo:
+  maxAlignment:    1
+  hasCalls:        true
+machineFunctionInfo: {}
+body:             |
+  bb.0.entry:
+    successors: %bb.4(0x80000000)
+  
+    %1:gr32 = MOV32ri 1
+    JMP_1 %bb.4
+  
+  bb.1.L:
+    successors: %bb.2(0x04000000), %bb.1(0x7c000000)
+  
+    %2:gr32 = MOV32r0 implicit-def dead $eflags
+    %3:gr8 = COPY %2.sub_8bit
+    TEST8rr %3, %3, implicit-def $eflags
+    JCC_1 %bb.1, 5, implicit $eflags
+    JMP_1 %bb.2
+  
+  bb.2.if.end4:
+    successors: %bb.3(0x783e0f0f), %bb.5(0x07c1f0f1)
+  
+    DBG_VALUE_LIST !19, !DIExpression(DW_OP_LLVM_arg, 0, DW_OP_LLVM_arg, 1, DW_OP_LLVM_arg, 2, DW_OP_and, DW_OP_or, DW_OP_stack_value), %0, 1, %0, debug-location !21
+    DBG_VALUE %0, $noreg, !20, !DIExpression(), debug-location !21
+    %5:gr32 = IMPLICIT_DEF
+    %4:gr32 = SUB32rr %0, killed %5, implicit-def $eflags
+    JCC_1 %bb.5, 13, implicit $eflags
+    JMP_1 %bb.3
+  
+  bb.3.if.then5:
+    successors: %bb.4(0x80000000)
+  
+    ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
+    CALL64pcrel32 @h, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp
+    ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
+    %6:gr32 = MOV32r0 implicit-def dead $eflags
+  
+  bb.4.L.outer:
+    successors: %bb.1(0x80000000)
+  
+    %0:gr32 = PHI %1, %bb.0, %6, %bb.3
+    JMP_1 %bb.1
+  
+  bb.5.if.end6:
+    RET 0
+
+...