[TailDuplicator] Add maximum predecessors and successors to consider …

…tail duplicating blocks (#78582) Fixes #78578. Duplicating a BB which has both multiple predecessors and successors will result in a complex CFG and also may cause huge amount of PHI nodes. See #78578 (comment) for a detailed description of the limit.
llvm · Apr 17, 2024 · 86a7828 · 86a7828
1 parent 4536ad4
commit 86a7828
Show file tree

Hide file tree

Showing 2 changed files with 280 additions and 0 deletions.
diff --git a/llvm/lib/CodeGen/TailDuplicator.cpp b/llvm/lib/CodeGen/TailDuplicator.cpp
@@ -68,6 +68,18 @@ static cl::opt<unsigned> TailDupIndirectBranchSize(
              "end with indirect branches."), cl::init(20),
     cl::Hidden);
 
+static cl::opt<unsigned>
+    TailDupPredSize("tail-dup-pred-size",
+                    cl::desc("Maximum predecessors (maximum successors at the "
+                             "same time) to consider tail duplicating blocks."),
+                    cl::init(16), cl::Hidden);
+
+static cl::opt<unsigned>
+    TailDupSuccSize("tail-dup-succ-size",
+                    cl::desc("Maximum successors (maximum predecessors at the "
+                             "same time) to consider tail duplicating blocks."),
+                    cl::init(16), cl::Hidden);
+
 static cl::opt<bool>
     TailDupVerify("tail-dup-verify",
                   cl::desc("Verify sanity of PHI instructions during taildup"),
@@ -565,6 +577,14 @@ bool TailDuplicator::shouldTailDuplicate(bool IsSimple,
   if (TailBB.isSuccessor(&TailBB))
     return false;
 
+  // Duplicating a BB which has both multiple predecessors and successors will
+  // result in a complex CFG and also may cause huge amount of PHI nodes. If we
+  // want to remove this limitation, we have to address
+  // https://github.com/llvm/llvm-project/issues/78578.
+  if (TailBB.pred_size() > TailDupPredSize &&
+      TailBB.succ_size() > TailDupSuccSize)
+    return false;
+
   // Set the limit on the cost to duplicate. When optimizing for size,
   // duplicate only one, because one branch instruction can be eliminated to
   // compensate for the duplication.

diff --git a/llvm/test/CodeGen/X86/tail-dup-pred-succ-size.mir b/llvm/test/CodeGen/X86/tail-dup-pred-succ-size.mir
@@ -0,0 +1,260 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 4
+# RUN: llc -mtriple=x86_64-unknown-linux-gnu -run-pass=early-tailduplication -tail-dup-pred-size=3 -tail-dup-succ-size=3 %s -o - | FileCheck %s -check-prefix=LIMIT
+# RUN: llc -mtriple=x86_64-unknown-linux-gnu -run-pass=early-tailduplication -tail-dup-pred-size=4 -tail-dup-succ-size=4 %s -o - | FileCheck %s -check-prefix=NOLIMIT
+
+---
+name:            foo
+tracksRegLiveness: true
+jumpTable:
+  kind:            block-address
+  entries:
+    - id:              0
+      blocks:          [ '%bb.2', '%bb.3', '%bb.4', '%bb.5' ]
+    - id:              1
+      blocks:          [ '%bb.9', '%bb.10', '%bb.11', '%bb.12' ]
+body:             |
+  ; LIMIT-LABEL: name: foo
+  ; LIMIT: bb.0:
+  ; LIMIT-NEXT:   successors: %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000), %bb.5(0x20000000)
+  ; LIMIT-NEXT:   liveins: $rdi, $esi
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[COPY:%[0-9]+]]:gr32 = COPY $esi
+  ; LIMIT-NEXT:   [[COPY1:%[0-9]+]]:gr64 = COPY $rdi
+  ; LIMIT-NEXT:   [[SHR32ri:%[0-9]+]]:gr32 = SHR32ri [[COPY]], 1, implicit-def dead $eflags
+  ; LIMIT-NEXT:   [[AND32ri:%[0-9]+]]:gr32 = AND32ri [[SHR32ri]], 7, implicit-def dead $eflags
+  ; LIMIT-NEXT:   [[SUBREG_TO_REG:%[0-9]+]]:gr64_nosp = SUBREG_TO_REG 0, killed [[AND32ri]], %subreg.sub_32bit
+  ; LIMIT-NEXT:   JMP64m $noreg, 8, [[SUBREG_TO_REG]], %jump-table.0, $noreg
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.2:
+  ; LIMIT-NEXT:   successors: %bb.7(0x80000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[MOV32rm:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; LIMIT-NEXT:   JMP_1 %bb.7
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.3:
+  ; LIMIT-NEXT:   successors: %bb.7(0x80000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[MOV32rm1:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; LIMIT-NEXT:   [[SHR32ri1:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm1]], 1, implicit-def dead $eflags
+  ; LIMIT-NEXT:   JMP_1 %bb.7
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.4:
+  ; LIMIT-NEXT:   successors: %bb.7(0x80000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[MOV32rm2:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; LIMIT-NEXT:   [[SHR32ri2:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm2]], 2, implicit-def dead $eflags
+  ; LIMIT-NEXT:   JMP_1 %bb.7
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.5:
+  ; LIMIT-NEXT:   successors: %bb.7(0x80000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[MOV32rm3:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; LIMIT-NEXT:   [[SHR32ri3:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm3]], 3, implicit-def dead $eflags
+  ; LIMIT-NEXT:   JMP_1 %bb.7
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.6:
+  ; LIMIT-NEXT:   successors:
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.7:
+  ; LIMIT-NEXT:   successors: %bb.9(0x20000000), %bb.10(0x20000000), %bb.11(0x20000000), %bb.12(0x20000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[PHI:%[0-9]+]]:gr32 = PHI [[SHR32ri3]], %bb.5, [[SHR32ri2]], %bb.4, [[SHR32ri1]], %bb.3, [[MOV32rm]], %bb.2
+  ; LIMIT-NEXT:   [[SHR32ri4:%[0-9]+]]:gr32 = SHR32ri [[COPY]], 2, implicit-def dead $eflags
+  ; LIMIT-NEXT:   [[AND32ri1:%[0-9]+]]:gr32 = AND32ri [[SHR32ri4]], 7, implicit-def dead $eflags
+  ; LIMIT-NEXT:   [[SUBREG_TO_REG1:%[0-9]+]]:gr64_nosp = SUBREG_TO_REG 0, killed [[AND32ri1]], %subreg.sub_32bit
+  ; LIMIT-NEXT:   JMP64m $noreg, 8, [[SUBREG_TO_REG1]], %jump-table.1, $noreg
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.9:
+  ; LIMIT-NEXT:   successors: %bb.13(0x80000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[MOV32rm4:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; LIMIT-NEXT:   JMP_1 %bb.13
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.10:
+  ; LIMIT-NEXT:   successors: %bb.13(0x80000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[MOV32rm5:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; LIMIT-NEXT:   [[SHR32ri5:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm5]], 1, implicit-def dead $eflags
+  ; LIMIT-NEXT:   JMP_1 %bb.13
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.11:
+  ; LIMIT-NEXT:   successors: %bb.13(0x80000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[MOV32rm6:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; LIMIT-NEXT:   [[SHR32ri6:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm6]], 2, implicit-def dead $eflags
+  ; LIMIT-NEXT:   JMP_1 %bb.13
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.12:
+  ; LIMIT-NEXT:   successors: %bb.13(0x80000000)
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT:   [[MOV32rm7:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; LIMIT-NEXT:   [[SHR32ri7:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm7]], 6, implicit-def dead $eflags
+  ; LIMIT-NEXT: {{  $}}
+  ; LIMIT-NEXT: bb.13:
+  ; LIMIT-NEXT:   [[PHI1:%[0-9]+]]:gr32 = PHI [[SHR32ri7]], %bb.12, [[SHR32ri6]], %bb.11, [[SHR32ri5]], %bb.10, [[MOV32rm4]], %bb.9
+  ; LIMIT-NEXT:   [[OR32rr:%[0-9]+]]:gr32 = OR32rr [[PHI1]], [[PHI]], implicit-def dead $eflags
+  ; LIMIT-NEXT:   $eax = COPY [[OR32rr]]
+  ; LIMIT-NEXT:   RET 0, $eax
+  ;
+  ; NOLIMIT-LABEL: name: foo
+  ; NOLIMIT: bb.0:
+  ; NOLIMIT-NEXT:   successors: %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000), %bb.5(0x20000000)
+  ; NOLIMIT-NEXT:   liveins: $rdi, $esi
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[COPY:%[0-9]+]]:gr32 = COPY $esi
+  ; NOLIMIT-NEXT:   [[COPY1:%[0-9]+]]:gr64 = COPY $rdi
+  ; NOLIMIT-NEXT:   [[SHR32ri:%[0-9]+]]:gr32 = SHR32ri [[COPY]], 1, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[AND32ri:%[0-9]+]]:gr32 = AND32ri [[SHR32ri]], 7, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[SUBREG_TO_REG:%[0-9]+]]:gr64_nosp = SUBREG_TO_REG 0, killed [[AND32ri]], %subreg.sub_32bit
+  ; NOLIMIT-NEXT:   JMP64m $noreg, 8, [[SUBREG_TO_REG]], %jump-table.0, $noreg
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.2:
+  ; NOLIMIT-NEXT:   successors: %bb.9(0x20000000), %bb.10(0x20000000), %bb.11(0x20000000), %bb.12(0x20000000)
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[MOV32rm:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; NOLIMIT-NEXT:   [[SHR32ri1:%[0-9]+]]:gr32 = SHR32ri [[COPY]], 2, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[AND32ri1:%[0-9]+]]:gr32 = AND32ri [[SHR32ri1]], 7, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[SUBREG_TO_REG1:%[0-9]+]]:gr64_nosp = SUBREG_TO_REG 0, [[AND32ri1]], %subreg.sub_32bit
+  ; NOLIMIT-NEXT:   JMP64m $noreg, 8, [[SUBREG_TO_REG1]], %jump-table.1, $noreg
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.3:
+  ; NOLIMIT-NEXT:   successors: %bb.9(0x20000000), %bb.10(0x20000000), %bb.11(0x20000000), %bb.12(0x20000000)
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[MOV32rm1:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; NOLIMIT-NEXT:   [[SHR32ri2:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm1]], 1, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[SHR32ri3:%[0-9]+]]:gr32 = SHR32ri [[COPY]], 2, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[AND32ri2:%[0-9]+]]:gr32 = AND32ri [[SHR32ri3]], 7, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[SUBREG_TO_REG2:%[0-9]+]]:gr64_nosp = SUBREG_TO_REG 0, [[AND32ri2]], %subreg.sub_32bit
+  ; NOLIMIT-NEXT:   JMP64m $noreg, 8, [[SUBREG_TO_REG2]], %jump-table.1, $noreg
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.4:
+  ; NOLIMIT-NEXT:   successors: %bb.9(0x20000000), %bb.10(0x20000000), %bb.11(0x20000000), %bb.12(0x20000000)
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[MOV32rm2:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; NOLIMIT-NEXT:   [[SHR32ri4:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm2]], 2, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[SHR32ri5:%[0-9]+]]:gr32 = SHR32ri [[COPY]], 2, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[AND32ri3:%[0-9]+]]:gr32 = AND32ri [[SHR32ri5]], 7, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[SUBREG_TO_REG3:%[0-9]+]]:gr64_nosp = SUBREG_TO_REG 0, [[AND32ri3]], %subreg.sub_32bit
+  ; NOLIMIT-NEXT:   JMP64m $noreg, 8, [[SUBREG_TO_REG3]], %jump-table.1, $noreg
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.5:
+  ; NOLIMIT-NEXT:   successors: %bb.9(0x20000000), %bb.10(0x20000000), %bb.11(0x20000000), %bb.12(0x20000000)
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[MOV32rm3:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; NOLIMIT-NEXT:   [[SHR32ri6:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm3]], 3, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[SHR32ri7:%[0-9]+]]:gr32 = SHR32ri [[COPY]], 2, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[AND32ri4:%[0-9]+]]:gr32 = AND32ri [[SHR32ri7]], 7, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   [[SUBREG_TO_REG4:%[0-9]+]]:gr64_nosp = SUBREG_TO_REG 0, [[AND32ri4]], %subreg.sub_32bit
+  ; NOLIMIT-NEXT:   JMP64m $noreg, 8, [[SUBREG_TO_REG4]], %jump-table.1, $noreg
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.6:
+  ; NOLIMIT-NEXT:   successors:
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.9:
+  ; NOLIMIT-NEXT:   successors: %bb.13(0x80000000)
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[PHI:%[0-9]+]]:gr32 = PHI [[MOV32rm]], %bb.2, [[SHR32ri2]], %bb.3, [[SHR32ri4]], %bb.4, [[SHR32ri6]], %bb.5
+  ; NOLIMIT-NEXT:   [[MOV32rm4:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; NOLIMIT-NEXT:   JMP_1 %bb.13
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.10:
+  ; NOLIMIT-NEXT:   successors: %bb.13(0x80000000)
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[PHI1:%[0-9]+]]:gr32 = PHI [[MOV32rm]], %bb.2, [[SHR32ri2]], %bb.3, [[SHR32ri4]], %bb.4, [[SHR32ri6]], %bb.5
+  ; NOLIMIT-NEXT:   [[MOV32rm5:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; NOLIMIT-NEXT:   [[SHR32ri8:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm5]], 1, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   JMP_1 %bb.13
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.11:
+  ; NOLIMIT-NEXT:   successors: %bb.13(0x80000000)
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[PHI2:%[0-9]+]]:gr32 = PHI [[MOV32rm]], %bb.2, [[SHR32ri2]], %bb.3, [[SHR32ri4]], %bb.4, [[SHR32ri6]], %bb.5
+  ; NOLIMIT-NEXT:   [[MOV32rm6:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; NOLIMIT-NEXT:   [[SHR32ri9:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm6]], 2, implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   JMP_1 %bb.13
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.12:
+  ; NOLIMIT-NEXT:   successors: %bb.13(0x80000000)
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT:   [[PHI3:%[0-9]+]]:gr32 = PHI [[MOV32rm]], %bb.2, [[SHR32ri2]], %bb.3, [[SHR32ri4]], %bb.4, [[SHR32ri6]], %bb.5
+  ; NOLIMIT-NEXT:   [[MOV32rm7:%[0-9]+]]:gr32 = MOV32rm [[COPY1]], 1, $noreg, 0, $noreg
+  ; NOLIMIT-NEXT:   [[SHR32ri10:%[0-9]+]]:gr32 = SHR32ri [[MOV32rm7]], 6, implicit-def dead $eflags
+  ; NOLIMIT-NEXT: {{  $}}
+  ; NOLIMIT-NEXT: bb.13:
+  ; NOLIMIT-NEXT:   [[PHI4:%[0-9]+]]:gr32 = PHI [[PHI]], %bb.9, [[PHI1]], %bb.10, [[PHI2]], %bb.11, [[PHI3]], %bb.12
+  ; NOLIMIT-NEXT:   [[PHI5:%[0-9]+]]:gr32 = PHI [[SHR32ri10]], %bb.12, [[SHR32ri9]], %bb.11, [[SHR32ri8]], %bb.10, [[MOV32rm4]], %bb.9
+  ; NOLIMIT-NEXT:   [[OR32rr:%[0-9]+]]:gr32 = OR32rr [[PHI5]], [[PHI4]], implicit-def dead $eflags
+  ; NOLIMIT-NEXT:   $eax = COPY [[OR32rr]]
+  ; NOLIMIT-NEXT:   RET 0, $eax
+  bb.0:
+    liveins: $rdi, $esi
+
+    %11:gr32 = COPY $esi
+    %10:gr64 = COPY $rdi
+    %13:gr32 = SHR32ri %11, 1, implicit-def dead $eflags
+    %14:gr32 = AND32ri %13, 7, implicit-def dead $eflags
+    %12:gr64_nosp = SUBREG_TO_REG 0, killed %14, %subreg.sub_32bit
+
+  bb.1:
+    successors: %bb.2, %bb.3, %bb.4, %bb.5
+
+    JMP64m $noreg, 8, %12, %jump-table.0, $noreg
+
+  bb.2:
+    %0:gr32 = MOV32rm %10, 1, $noreg, 0, $noreg
+    JMP_1 %bb.7
+
+  bb.3:
+    %17:gr32 = MOV32rm %10, 1, $noreg, 0, $noreg
+    %1:gr32 = SHR32ri %17, 1, implicit-def dead $eflags
+    JMP_1 %bb.7
+
+  bb.4:
+    %16:gr32 = MOV32rm %10, 1, $noreg, 0, $noreg
+    %2:gr32 = SHR32ri %16, 2, implicit-def dead $eflags
+    JMP_1 %bb.7
+
+  bb.5:
+    %15:gr32 = MOV32rm %10, 1, $noreg, 0, $noreg
+    %3:gr32 = SHR32ri %15, 3, implicit-def dead $eflags
+    JMP_1 %bb.7
+
+  bb.6:
+    successors:
+
+  bb.7:
+    %4:gr32 = PHI %3, %bb.5, %2, %bb.4, %1, %bb.3, %0, %bb.2
+    %19:gr32 = SHR32ri %11, 2, implicit-def dead $eflags
+    %20:gr32 = AND32ri %19, 7, implicit-def dead $eflags
+    %18:gr64_nosp = SUBREG_TO_REG 0, killed %20, %subreg.sub_32bit
+
+  bb.8:
+    successors: %bb.9, %bb.10, %bb.11, %bb.12
+
+    JMP64m $noreg, 8, %18, %jump-table.1, $noreg
+
+  bb.9:
+    %5:gr32 = MOV32rm %10, 1, $noreg, 0, $noreg
+    JMP_1 %bb.13
+
+  bb.10:
+    %23:gr32 = MOV32rm %10, 1, $noreg, 0, $noreg
+    %6:gr32 = SHR32ri %23, 1, implicit-def dead $eflags
+    JMP_1 %bb.13
+
+  bb.11:
+    %22:gr32 = MOV32rm %10, 1, $noreg, 0, $noreg
+    %7:gr32 = SHR32ri %22, 2, implicit-def dead $eflags
+    JMP_1 %bb.13
+
+  bb.12:
+    %21:gr32 = MOV32rm %10, 1, $noreg, 0, $noreg
+    %8:gr32 = SHR32ri %21, 6, implicit-def dead $eflags
+
+  bb.13:
+    %9:gr32 = PHI %8, %bb.12, %7, %bb.11, %6, %bb.10, %5, %bb.9
+    %24:gr32 = OR32rr %9, %4, implicit-def dead $eflags
+    $eax = COPY %24
+    RET 0, $eax
+
+...