release/20.x: [TailDuplicator] Determine if computed gotos using `blockaddress` (#132536) #133082

llvmbot · 2025-03-26T13:34:45Z

Backport 66f158d

Requested by: @dianqk

llvmbot · 2025-03-26T13:34:53Z

@arsenm What do you think about merging this PR to the release branch?

llvmbot · 2025-03-26T13:35:23Z

@llvm/pr-subscribers-backend-x86

Author: None (llvmbot)

Changes

Backport 66f158d

Requested by: @dianqk

Patch is 29.24 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/133082.diff

4 Files Affected:

(modified) llvm/include/llvm/CodeGen/MachineBasicBlock.h (+9)
(modified) llvm/include/llvm/CodeGen/MachineInstr.h (+5-11)
(modified) llvm/lib/CodeGen/TailDuplicator.cpp (+1-1)
(modified) llvm/test/CodeGen/X86/tail-dup-computed-goto.mir (+188-77)

diff --git a/llvm/include/llvm/CodeGen/MachineBasicBlock.h b/llvm/include/llvm/CodeGen/MachineBasicBlock.h
index 0b803a9724742..11efb2f656a7a 100644
--- a/llvm/include/llvm/CodeGen/MachineBasicBlock.h
+++ b/llvm/include/llvm/CodeGen/MachineBasicBlock.h
@@ -311,6 +311,15 @@ class MachineBasicBlock
   const MachineFunction *getParent() const { return xParent; }
   MachineFunction *getParent() { return xParent; }
 
+  /// Returns true if the original IR terminator is an `indirectbr`. This
+  /// typically corresponds to a `goto` in C, rather than jump tables.
+  bool terminatorIsComputedGoto() const {
+    return back().isIndirectBranch() &&
+           llvm::all_of(successors(), [](const MachineBasicBlock *Succ) {
+             return Succ->isIRBlockAddressTaken();
+           });
+  }
+
   using instr_iterator = Instructions::iterator;
   using const_instr_iterator = Instructions::const_iterator;
   using reverse_instr_iterator = Instructions::reverse_iterator;
diff --git a/llvm/include/llvm/CodeGen/MachineInstr.h b/llvm/include/llvm/CodeGen/MachineInstr.h
index b26cabe801ee8..997d6a5554e06 100644
--- a/llvm/include/llvm/CodeGen/MachineInstr.h
+++ b/llvm/include/llvm/CodeGen/MachineInstr.h
@@ -994,17 +994,8 @@ class MachineInstr
 
   /// Return true if this is an indirect branch, such as a
   /// branch through a register.
-  bool isIndirectBranch(QueryType Type = AnyInBundle,
-                        bool IncludeJumpTable = true) const {
-    return hasProperty(MCID::IndirectBranch, Type) &&
-           (IncludeJumpTable || !llvm::any_of(operands(), [](const auto &Op) {
-              return Op.isJTI();
-            }));
-  }
-
-  bool isComputedGoto(QueryType Type = AnyInBundle) const {
-    // Jump tables are not considered computed gotos.
-    return isIndirectBranch(Type, /*IncludeJumpTable=*/false);
+  bool isIndirectBranch(QueryType Type = AnyInBundle) const {
+    return hasProperty(MCID::IndirectBranch, Type);
   }
 
   /// Return true if this is a branch which may fall
@@ -2088,6 +2079,9 @@ class MachineInstr
                     MCSymbol *PreInstrSymbol, MCSymbol *PostInstrSymbol,
                     MDNode *HeapAllocMarker, MDNode *PCSections,
                     uint32_t CFIType, MDNode *MMRAs);
+
+  /// Returns true if all successors are IRBlockAddressTaken.
+  bool jumpToIRBlockAddressTaken() const;
 };
 
 /// Special DenseMapInfo traits to compare MachineInstr* by *value* of the
diff --git a/llvm/lib/CodeGen/TailDuplicator.cpp b/llvm/lib/CodeGen/TailDuplicator.cpp
index 21f75458c90f3..b0de3c322ddd0 100644
--- a/llvm/lib/CodeGen/TailDuplicator.cpp
+++ b/llvm/lib/CodeGen/TailDuplicator.cpp
@@ -604,7 +604,7 @@ bool TailDuplicator::shouldTailDuplicate(bool IsSimple,
   bool HasComputedGoto = false;
   if (!TailBB.empty()) {
     HasIndirectbr = TailBB.back().isIndirectBranch();
-    HasComputedGoto = TailBB.back().isComputedGoto();
+    HasComputedGoto = TailBB.terminatorIsComputedGoto();
   }
 
   if (HasIndirectbr && PreRegAlloc)
diff --git a/llvm/test/CodeGen/X86/tail-dup-computed-goto.mir b/llvm/test/CodeGen/X86/tail-dup-computed-goto.mir
index a472dc67d8d51..17de405928d37 100644
--- a/llvm/test/CodeGen/X86/tail-dup-computed-goto.mir
+++ b/llvm/test/CodeGen/X86/tail-dup-computed-goto.mir
@@ -2,15 +2,27 @@
 # RUN: llc -mtriple=x86_64-unknown-linux-gnu -run-pass=early-tailduplication -tail-dup-pred-size=1 -tail-dup-succ-size=1 %s -o - | FileCheck %s
 # Check that only the computed goto is not be restrict by tail-dup-pred-size and tail-dup-succ-size.
 --- |
+  @computed_goto.dispatch = constant [5 x ptr] [ptr null, ptr blockaddress(@computed_goto, %bb1), ptr blockaddress(@computed_goto, %bb2), ptr blockaddress(@computed_goto, %bb3), ptr blockaddress(@computed_goto, %bb4)]
   declare i64 @f0()
   declare i64 @f1()
   declare i64 @f2()
   declare i64 @f3()
   declare i64 @f4()
   declare i64 @f5()
-  @computed_goto.dispatch = external global [5 x ptr]
-  define void @computed_goto() { ret void }
+  define void @computed_goto() {
+    start:
+      ret void
+    bb1:
+      ret void
+    bb2:
+      ret void
+    bb3:
+      ret void
+    bb4:
+      ret void
+  }
   define void @jump_table() { ret void }
+  define void @jump_table_pic() { ret void }
 ...
 ---
 name:            computed_goto
@@ -23,98 +35,88 @@ body:             |
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f0, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:gr64 = COPY $rax
+  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:gr64_nosp = COPY $rax
   ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:gr64_nosp = COPY [[COPY]]
-  ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:gr64_nosp = COPY [[COPY1]]
-  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY1]], @computed_goto.dispatch, $noreg
+  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY]], @computed_goto.dispatch, $noreg
   ; CHECK-NEXT: {{  $}}
-  ; CHECK-NEXT: bb.1:
+  ; CHECK-NEXT: bb.1.bb1 (ir-block-address-taken %ir-block.bb1):
   ; CHECK-NEXT:   successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f1, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY3:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY4:%[0-9]+]]:gr64_nosp = COPY [[COPY3]]
-  ; CHECK-NEXT:   [[COPY5:%[0-9]+]]:gr64_nosp = COPY [[COPY4]]
-  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY4]], @computed_goto.dispatch, $noreg
+  ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:gr64_nosp = COPY $rax
+  ; CHECK-NEXT:   [[COPY3:%[0-9]+]]:gr64_nosp = COPY [[COPY2]]
+  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY2]], @computed_goto.dispatch, $noreg
   ; CHECK-NEXT: {{  $}}
-  ; CHECK-NEXT: bb.2:
+  ; CHECK-NEXT: bb.2.bb2 (ir-block-address-taken %ir-block.bb2):
   ; CHECK-NEXT:   successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f2, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY6:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY7:%[0-9]+]]:gr64_nosp = COPY [[COPY6]]
-  ; CHECK-NEXT:   [[COPY8:%[0-9]+]]:gr64_nosp = COPY [[COPY7]]
-  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY7]], @computed_goto.dispatch, $noreg
+  ; CHECK-NEXT:   [[COPY4:%[0-9]+]]:gr64_nosp = COPY $rax
+  ; CHECK-NEXT:   [[COPY5:%[0-9]+]]:gr64_nosp = COPY [[COPY4]]
+  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY4]], @computed_goto.dispatch, $noreg
   ; CHECK-NEXT: {{  $}}
-  ; CHECK-NEXT: bb.3:
+  ; CHECK-NEXT: bb.3.bb3 (ir-block-address-taken %ir-block.bb3):
   ; CHECK-NEXT:   successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f3, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY9:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY10:%[0-9]+]]:gr64_nosp = COPY [[COPY9]]
-  ; CHECK-NEXT:   [[COPY11:%[0-9]+]]:gr64_nosp = COPY [[COPY10]]
-  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY10]], @computed_goto.dispatch, $noreg
+  ; CHECK-NEXT:   [[COPY6:%[0-9]+]]:gr64_nosp = COPY $rax
+  ; CHECK-NEXT:   [[COPY7:%[0-9]+]]:gr64_nosp = COPY [[COPY6]]
+  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY6]], @computed_goto.dispatch, $noreg
   ; CHECK-NEXT: {{  $}}
-  ; CHECK-NEXT: bb.4:
+  ; CHECK-NEXT: bb.4.bb4 (ir-block-address-taken %ir-block.bb4):
   ; CHECK-NEXT:   successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f4, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY12:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY13:%[0-9]+]]:gr64_nosp = COPY [[COPY12]]
-  ; CHECK-NEXT:   [[COPY14:%[0-9]+]]:gr64_nosp = COPY [[COPY13]]
-  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY13]], @computed_goto.dispatch, $noreg
+  ; CHECK-NEXT:   [[COPY8:%[0-9]+]]:gr64_nosp = COPY $rax
+  ; CHECK-NEXT:   [[COPY9:%[0-9]+]]:gr64_nosp = COPY [[COPY8]]
+  ; CHECK-NEXT:   JMP64m $noreg, 8, [[COPY8]], @computed_goto.dispatch, $noreg
   bb.0:
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f0, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-    %6:gr64 = COPY $rax
-    %0:gr64 = COPY %6
+    %1:gr64 = COPY $rax
     JMP_1 %bb.5
 
-  bb.1:
+  bb.1.bb1 (ir-block-address-taken %ir-block.bb1):
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f1, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-    %10:gr64 = COPY $rax
-    %1:gr64 = COPY %10
+    %3:gr64 = COPY $rax
     JMP_1 %bb.5
 
-  bb.2:
+  bb.2.bb2 (ir-block-address-taken %ir-block.bb2):
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f2, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-    %9:gr64 = COPY $rax
-    %2:gr64 = COPY %9
+    %5:gr64 = COPY $rax
     JMP_1 %bb.5
 
-  bb.3:
+  bb.3.bb3 (ir-block-address-taken %ir-block.bb3):
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f3, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-    %8:gr64 = COPY $rax
-    %3:gr64 = COPY %8
+    %7:gr64 = COPY $rax
     JMP_1 %bb.5
 
-  bb.4:
+  bb.4.bb4 (ir-block-address-taken %ir-block.bb4):
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f4, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-    %7:gr64 = COPY $rax
-    %4:gr64 = COPY %7
+    %9:gr64 = COPY $rax
 
   bb.5:
     successors: %bb.1, %bb.2, %bb.3, %bb.4
 
-    %5:gr64_nosp = PHI %0, %bb.0, %4, %bb.4, %3, %bb.3, %2, %bb.2, %1, %bb.1
-    JMP64m $noreg, 8, %5, @computed_goto.dispatch, $noreg
+    %10:gr64_nosp = PHI %1, %bb.0, %9, %bb.4, %7, %bb.3, %5, %bb.2, %3, %bb.1
+    JMP64m $noreg, 8, %10, @computed_goto.dispatch, $noreg
 
 ...
 ---
@@ -124,7 +126,7 @@ jumpTable:
   kind:            block-address
   entries:
     - id:              0
-      blocks:          [ '%bb.2', '%bb.3', '%bb.4', '%bb.5', '%bb.6' ]
+      blocks:          [ '%bb.3', '%bb.4', '%bb.5', '%bb.6', '%bb.7' ]
 body:             |
   ; CHECK-LABEL: name: jump_table
   ; CHECK: bb.0:
@@ -134,12 +136,11 @@ body:             |
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f0, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   [[COPY:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:gr64 = COPY [[COPY]]
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT: bb.1:
   ; CHECK-NEXT:   successors: %bb.3(0x1999999a), %bb.4(0x1999999a), %bb.5(0x1999999a), %bb.6(0x1999999a), %bb.7(0x1999999a)
   ; CHECK-NEXT: {{  $}}
-  ; CHECK-NEXT:   [[PHI:%[0-9]+]]:gr64 = PHI [[COPY1]], %bb.0, %3, %bb.7, %4, %bb.6, %5, %bb.5, %6, %bb.4, %7, %bb.3
+  ; CHECK-NEXT:   [[PHI:%[0-9]+]]:gr64 = PHI [[COPY]], %bb.0, %2, %bb.7, %3, %bb.6, %4, %bb.5, %5, %bb.4, %6, %bb.3
   ; CHECK-NEXT:   [[DEC64r:%[0-9]+]]:gr64_nosp = DEC64r [[PHI]], implicit-def dead $eflags
   ; CHECK-NEXT:   JMP64m $noreg, 8, [[DEC64r]], %jump-table.0, $noreg :: (load (s64) from jump-table)
   ; CHECK-NEXT: {{  $}}
@@ -149,8 +150,7 @@ body:             |
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f1, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY3:%[0-9]+]]:gr64 = COPY [[COPY2]]
+  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:gr64 = COPY $rax
   ; CHECK-NEXT:   JMP_1 %bb.1
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT: bb.4:
@@ -159,8 +159,7 @@ body:             |
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f2, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY4:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY5:%[0-9]+]]:gr64 = COPY [[COPY4]]
+  ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:gr64 = COPY $rax
   ; CHECK-NEXT:   JMP_1 %bb.1
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT: bb.5:
@@ -169,8 +168,7 @@ body:             |
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f3, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY6:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY7:%[0-9]+]]:gr64 = COPY [[COPY6]]
+  ; CHECK-NEXT:   [[COPY3:%[0-9]+]]:gr64 = COPY $rax
   ; CHECK-NEXT:   JMP_1 %bb.1
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT: bb.6:
@@ -179,8 +177,7 @@ body:             |
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f4, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY8:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY9:%[0-9]+]]:gr64 = COPY [[COPY8]]
+  ; CHECK-NEXT:   [[COPY4:%[0-9]+]]:gr64 = COPY $rax
   ; CHECK-NEXT:   JMP_1 %bb.1
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT: bb.7:
@@ -189,67 +186,181 @@ body:             |
   ; CHECK-NEXT:   ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
   ; CHECK-NEXT:   CALL64pcrel32 target-flags(x86-plt) @f5, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
   ; CHECK-NEXT:   ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-  ; CHECK-NEXT:   [[COPY10:%[0-9]+]]:gr64 = COPY $rax
-  ; CHECK-NEXT:   [[COPY11:%[0-9]+]]:gr64 = COPY [[COPY10]]
+  ; CHECK-NEXT:   [[COPY5:%[0-9]+]]:gr64 = COPY $rax
   ; CHECK-NEXT:   JMP_1 %bb.1
-  ; CHECK-NEXT: {{  $}}
-  ; CHECK-NEXT: bb.8:
   bb.0:
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f0, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-    %7:gr64 = COPY $rax
-    %0:gr64 = COPY %7
+    %1:gr64 = COPY $rax
 
   bb.1:
-    %1:gr64 = PHI %0, %bb.0, %6, %bb.6, %5, %bb.5, %4, %bb.4, %3, %bb.3, %2, %bb.2
-    %8:gr64_nosp = DEC64r %1, implicit-def dead $eflags
+    %2:gr64 = PHI %1, %bb.0, %3, %bb.7, %4, %bb.6, %5, %bb.5, %6, %bb.4, %7, %bb.3
+    %8:gr64_nosp = DEC64r %2, implicit-def dead $eflags
 
-  bb.8:
-    successors: %bb.2(0x1999999a), %bb.3(0x1999999a), %bb.4(0x1999999a), %bb.5(0x1999999a), %bb.6(0x1999999a)
+  bb.2:
+    successors: %bb.3(0x1999999a), %bb.4(0x1999999a), %bb.5(0x1999999a), %bb.6(0x1999999a), %bb.7(0x1999999a)
 
     JMP64m $noreg, 8, %8, %jump-table.0, $noreg :: (load (s64) from jump-table)
 
-  bb.2:
+  bb.3:
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f1, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-    %13:gr64 = COPY $rax
-    %2:gr64 = COPY %13
+    %7:gr64 = COPY $rax
     JMP_1 %bb.1
 
-  bb.3:
+  bb.4:
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f2, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
     ADJCALLSTACKUP64 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
-    %12:gr64 = COPY $rax
-    %3:gr64 = COPY %12
+    %6:gr64 = COPY $rax
     JMP_1 %bb.1
 
-  bb.4:
+  bb.5:
     ADJCALLSTACKDOWN64 0, 0, 0, implicit-def dead $rsp, implicit-def dead $eflags, implicit-def dead $ssp, implicit $rsp, implicit $ssp
     CALL64pcrel32 target-flags(x86-plt) @f3, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
    ...
[truncated]

arsenm

I can't say I know much about the feature but this should strictly move in a more conservative direction

tstellar · 2025-03-31T22:26:20Z

This has some CI test failures.

tstellar · 2025-04-01T20:10:22Z

This introduces an ABI change, I think. It modifies isIndirectBranch and removes isComputedGoto

dianqk · 2025-04-02T11:45:32Z

This introduces an ABI change, I think. It modifies isIndirectBranch and removes isComputedGoto

Although I could add a deprecation warning, this might also be considered an "ABI" break, so I removed these changes.

…vm#132536) Using `blockaddress` should be more reliable than determining if an operand comes from a jump table index. Alternative: Add the `MachineInstr::MIFlag::ComputedGoto` flag when lowering `indirectbr`. But I don't think this approach is suitable to backport. (cherry picked from commit 66f158d)

github-actions · 2025-04-11T16:37:06Z

@dianqk (or anyone else). If you would like to add a note about this fix in the release notes (completely optional). Please reply to this comment with a one or two sentence description of the fix. When you are done, please add the release:note label to this PR.

llvmbot added this to the LLVM 20.X Release milestone Mar 26, 2025

github-project-automation bot added this to LLVM Release Status Mar 26, 2025

github-project-automation bot moved this to Needs Triage in LLVM Release Status Mar 26, 2025

llvmbot requested a review from arsenm March 26, 2025 13:34

llvmbot added the backend:X86 label Mar 26, 2025

llvmbot mentioned this pull request Mar 26, 2025

[TailDuplicator] Determine if computed gotos using blockaddress #132536

Merged

tstellar moved this from Needs Triage to Needs Review in LLVM Release Status Mar 28, 2025

dianqk force-pushed the issue132536 branch from 8241e69 to a78eb08 Compare March 30, 2025 13:17

arsenm approved these changes Mar 30, 2025

View reviewed changes

github-project-automation bot moved this from Needs Review to Needs Merge in LLVM Release Status Mar 30, 2025

arsenm added the llvm:codegen label Mar 30, 2025

dianqk force-pushed the issue132536 branch from a78eb08 to 16dbf24 Compare April 1, 2025 10:19

dianqk force-pushed the issue132536 branch from 16dbf24 to 6a8abf2 Compare April 2, 2025 11:42

llvmbot force-pushed the issue132536 branch from 6a8abf2 to cf7bb13 Compare April 11, 2025 16:36

tstellar merged commit cf7bb13 into llvm:release/20.x Apr 11, 2025
7 of 10 checks passed

github-project-automation bot moved this from Needs Merge to Done in LLVM Release Status Apr 11, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

release/20.x: [TailDuplicator] Determine if computed gotos using `blockaddress` (#132536) #133082

release/20.x: [TailDuplicator] Determine if computed gotos using `blockaddress` (#132536) #133082

Uh oh!

llvmbot commented Mar 26, 2025

Uh oh!

llvmbot commented Mar 26, 2025

Uh oh!

llvmbot commented Mar 26, 2025

Uh oh!

arsenm left a comment

Uh oh!

tstellar commented Mar 31, 2025 •

edited

Loading

Uh oh!

tstellar commented Apr 1, 2025

Uh oh!

dianqk commented Apr 2, 2025

Uh oh!

Uh oh!

github-actions bot commented Apr 11, 2025

Uh oh!

Uh oh!

release/20.x: [TailDuplicator] Determine if computed gotos using blockaddress (#132536) #133082

release/20.x: [TailDuplicator] Determine if computed gotos using blockaddress (#132536) #133082

Uh oh!

Conversation

llvmbot commented Mar 26, 2025

Uh oh!

llvmbot commented Mar 26, 2025

Uh oh!

llvmbot commented Mar 26, 2025

Uh oh!

arsenm left a comment

Choose a reason for hiding this comment

Uh oh!

tstellar commented Mar 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tstellar commented Apr 1, 2025

Uh oh!

dianqk commented Apr 2, 2025

Uh oh!

Uh oh!

github-actions bot commented Apr 11, 2025

Uh oh!

Uh oh!

release/20.x: [TailDuplicator] Determine if computed gotos using `blockaddress` (#132536) #133082

release/20.x: [TailDuplicator] Determine if computed gotos using `blockaddress` (#132536) #133082

tstellar commented Mar 31, 2025 •

edited

Loading