[X86][GISel] Add missing legalization for G_IMPLICIT_DEF (#161699)

bonsthie · web-flow · commit a13ff2cb027e · 2025-10-06T09:25:11.000Z
Legalize scalar and vector integer types for `G_IMPLICIT_DEF` at
SSE2/AVX2/AVX-512 widths. This is groundwork for upcoming `G_*_VECTOR`
legalization, since vector inserts/builds rely on undef bases.
diff --git a/llvm/lib/Target/X86/GISel/X86InstructionSelector.cpp b/llvm/lib/Target/X86/GISel/X86InstructionSelector.cpp
@@ -407,6 +407,7 @@ bool X86InstructionSelector::select(MachineInstr &I) {
   case TargetOpcode::G_TRUNC:
     return selectTruncOrPtrToInt(I, MRI, MF);
   case TargetOpcode::G_INTTOPTR:
+  case TargetOpcode::G_FREEZE:
     return selectCopy(I, MRI);
   case TargetOpcode::G_ZEXT:
     return selectZext(I, MRI, MF);
diff --git a/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp b/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp
@@ -89,9 +89,29 @@ X86LegalizerInfo::X86LegalizerInfo(const X86Subtarget &STI,
   // 32/64-bits needs support for s64/s128 to handle cases:
   // s64 = EXTEND (G_IMPLICIT_DEF s32) -> s64 = G_IMPLICIT_DEF
   // s128 = EXTEND (G_IMPLICIT_DEF s32/s64) -> s128 = G_IMPLICIT_DEF
-  getActionDefinitionsBuilder(G_IMPLICIT_DEF)
+  getActionDefinitionsBuilder(
+      {G_IMPLICIT_DEF, G_PHI, G_FREEZE, G_CONSTANT_FOLD_BARRIER})
       .legalFor({p0, s1, s8, s16, s32, s64})
-      .legalFor(Is64Bit, {s128});
+      .legalFor(UseX87, {s80})
+      .legalFor(Is64Bit, {s128})
+      .legalFor(HasSSE2, {v16s8, v8s16, v4s32, v2s64})
+      .legalFor(HasAVX, {v32s8, v16s16, v8s32, v4s64})
+      .legalFor(HasAVX512, {v64s8, v32s16, v16s32, v8s64})
+      .widenScalarOrEltToNextPow2(0, /*Min=*/8)
+      .clampScalarOrElt(0, s8, sMaxScalar)
+      .moreElementsToNextPow2(0)
+      .clampMinNumElements(0, s8, 16)
+      .clampMinNumElements(0, s16, 8)
+      .clampMinNumElements(0, s32, 4)
+      .clampMinNumElements(0, s64, 2)
+      .clampMaxNumElements(0, s8, HasAVX512 ? 64 : (HasAVX ? 32 : 16))
+      .clampMaxNumElements(0, s16, HasAVX512 ? 32 : (HasAVX ? 16 : 8))
+      .clampMaxNumElements(0, s32, HasAVX512 ? 16 : (HasAVX ? 8 : 4))
+      .clampMaxNumElements(0, s64, HasAVX512 ? 8 : (HasAVX ? 4 : 2))
+      .clampMaxNumElements(0, p0,
+                           Is64Bit ? s64MaxVector.getNumElements()
+                                   : s32MaxVector.getNumElements())
+      .scalarizeIf(scalarOrEltWiderThan(0, 64), 0);
 
   getActionDefinitionsBuilder(G_CONSTANT)
       .legalFor({p0, s8, s16, s32})
@@ -289,26 +309,6 @@ X86LegalizerInfo::X86LegalizerInfo(const X86Subtarget &STI,
       .clampScalar(1, s16, sMaxScalar)
       .scalarSameSizeAs(0, 1);
 
-  // control flow
-  getActionDefinitionsBuilder(G_PHI)
-      .legalFor({s8, s16, s32, p0})
-      .legalFor(UseX87, {s80})
-      .legalFor(Is64Bit, {s64})
-      .legalFor(HasSSE1, {v16s8, v8s16, v4s32, v2s64})
-      .legalFor(HasAVX, {v32s8, v16s16, v8s32, v4s64})
-      .legalFor(HasAVX512, {v64s8, v32s16, v16s32, v8s64})
-      .clampMinNumElements(0, s8, 16)
-      .clampMinNumElements(0, s16, 8)
-      .clampMinNumElements(0, s32, 4)
-      .clampMinNumElements(0, s64, 2)
-      .clampMaxNumElements(0, s8, HasAVX512 ? 64 : (HasAVX ? 32 : 16))
-      .clampMaxNumElements(0, s16, HasAVX512 ? 32 : (HasAVX ? 16 : 8))
-      .clampMaxNumElements(0, s32, HasAVX512 ? 16 : (HasAVX ? 8 : 4))
-      .clampMaxNumElements(0, s64, HasAVX512 ? 8 : (HasAVX ? 4 : 2))
-      .widenScalarToNextPow2(0, /*Min=*/32)
-      .clampScalar(0, s8, sMaxScalar)
-      .scalarize(0);
-
   getActionDefinitionsBuilder(G_BRCOND).legalFor({s1});
 
   // pointer handling
@@ -592,11 +592,6 @@ X86LegalizerInfo::X86LegalizerInfo(const X86Subtarget &STI,
       .minScalar(0, LLT::scalar(32))
       .libcall();
 
-  getActionDefinitionsBuilder({G_FREEZE, G_CONSTANT_FOLD_BARRIER})
-      .legalFor({s8, s16, s32, s64, p0})
-      .widenScalarToNextPow2(0, /*Min=*/8)
-      .clampScalar(0, s8, sMaxScalar);
-
   getLegacyLegalizerInfo().computeTables();
   verify(*STI.getInstrInfo());
 }
diff --git a/llvm/test/CodeGen/X86/GlobalISel/legalize-phi.mir b/llvm/test/CodeGen/X86/GlobalISel/legalize-phi.mir
@@ -148,21 +148,21 @@ body:             |
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $edi
   ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $esi
+  ; CHECK-NEXT:   [[TRUNC:%[0-9]+]]:_(s1) = G_TRUNC [[COPY1]](s32)
   ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $edx
+  ; CHECK-NEXT:   [[TRUNC2:%[0-9]+]]:_(s1) = G_TRUNC [[COPY2]](s32)
   ; CHECK-NEXT:   [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
   ; CHECK-NEXT:   [[ICMP:%[0-9]+]]:_(s8) = G_ICMP intpred(sgt), [[COPY]](s32), [[C]]
-  ; CHECK-NEXT:   [[TRUNC:%[0-9]+]]:_(s1) = G_TRUNC [[ICMP]](s8)
-  ; CHECK-NEXT:   [[TRUNC1:%[0-9]+]]:_(s8) = G_TRUNC [[COPY1]](s32)
-  ; CHECK-NEXT:   G_BRCOND [[TRUNC]](s1), %bb.2
+  ; CHECK-NEXT:   [[TRUNC1:%[0-9]+]]:_(s1) = G_TRUNC [[ICMP]](s8)
+  ; CHECK-NEXT:   G_BRCOND [[TRUNC1]](s1), %bb.2
   ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT: bb.1.cond.false:
   ; CHECK-NEXT:   successors: %bb.2(0x80000000)
   ; CHECK-NEXT: {{  $}}
-  ; CHECK-NEXT:   [[TRUNC2:%[0-9]+]]:_(s8) = G_TRUNC [[COPY2]](s32)
-  ; CHECK-NEXT: {{  $}}
   ; CHECK-NEXT: bb.2.cond.end:
-  ; CHECK-NEXT:   [[PHI:%[0-9]+]]:_(s8) = G_PHI [[TRUNC2]](s8), %bb.1, [[TRUNC1]](s8), %bb.0
-  ; CHECK-NEXT:   $al = COPY [[PHI]](s8)
+  ; CHECK-NEXT:   [[PHI:%[0-9]+]]:_(s1) = G_PHI [[TRUNC2]](s1), %bb.1, [[TRUNC]](s1), %bb.0
+  ; CHECK-NEXT:   [[EXT:%[0-9]+]]:_(s8) = G_ANYEXT [[PHI]](s1)
+  ; CHECK-NEXT:   $al = COPY [[EXT]](s8)
   ; CHECK-NEXT:   RET 0, implicit $al
   bb.1.entry:
     successors: %bb.3(0x40000000), %bb.2(0x40000000)
diff --git a/llvm/test/CodeGen/X86/GlobalISel/legalize-undef-vec-scaling.mir b/llvm/test/CodeGen/X86/GlobalISel/legalize-undef-vec-scaling.mir
@@ -0,0 +1,32 @@
+# RUN: llc -mtriple=x86_64-linux-gnu -mattr=avx2 -run-pass=legalizer -global-isel-abort=2 -pass-remarks-missed='gisel*' %s -o -  | FileCheck %s --check-prefixes=CHECK,AVX2
+# RUN: llc -mtriple=x86_64-linux-gnu -mattr=sse2 -run-pass=legalizer -global-isel-abort=2 -pass-remarks-missed='gisel*' %s -o -  | FileCheck %s --check-prefixes=CHECK,SSE2
+# RUN: llc -mtriple=x86_64-linux-gnu -mattr=avx512f -run-pass=legalizer -global-isel-abort=2 -pass-remarks-missed='gisel*' %s -o -  | FileCheck %s --check-prefixes=CHECK,AVX512F 
+
+
+---
+name: test_basic_g_implicit_def_v8i64
+body: |
+  bb.0:
+    ; CHECK-LABEL: name: test_basic_g_implicit_def_v8i64
+    ; AVX512F: {{%[0-9]+}}:_(<8 x s64>) = G_IMPLICIT_DEF
+    ; AVX2: [[DEF_AVX2:%[0-9]+]]:_(<4 x s64>) = G_IMPLICIT_DEF
+    ; AVX2-NEXT: {{%[0-9]+}}:_(<8 x s64>) = G_CONCAT_VECTORS [[DEF_AVX2]](<4 x s64>), [[DEF_AVX2]](<4 x s64>)
+    ; SSE2: [[DEF_SSE2:%[0-9]+]]:_(<2 x s64>) = G_IMPLICIT_DEF
+    ; SSE2-NEXT: {{%[0-9]+}}:_(<8 x s64>) = G_CONCAT_VECTORS [[DEF_SSE2]](<2 x s64>), [[DEF_SSE2]](<2 x s64>), [[DEF_SSE2]](<2 x s64>), [[DEF_SSE2]](<2 x s64>)
+    %0:_(<8 x s64>) = G_IMPLICIT_DEF
+    RET 0, implicit %0
+...
+
+---
+name: test_g_implicit_def_cample_size
+body: |
+  bb.1:
+   ; CHECK-LABEL: name: test_g_implicit_def_cample_size
+   ; AVX512: {{%[0-9]+}}:_(<8 x s64>) = G_IMPLICIT_DEF
+   ; AVX2: {{%[0-9]+}}:_(<4 x s64>) = G_IMPLICIT_DEF
+   ; SSE2: {{%[0-9]+}}:_(<2 x s64>) = G_IMPLICIT_DEF
+    %0:_(<5 x s63>) = G_IMPLICIT_DEF
+    RET 0, implicit %0
+...
+
+
diff --git a/llvm/test/CodeGen/X86/GlobalISel/select-constant-fold-barrier-vec256.mir b/llvm/test/CodeGen/X86/GlobalISel/select-constant-fold-barrier-vec256.mir
@@ -0,0 +1,23 @@
+# RUN: llc -mtriple=x86_64-linux-gnu -mattr=+avx -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+
+---
+name:            select_cfb_vec256
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: vecr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: vecr, preferred-register: '', flags: [  ] }
+body:             |
+  bb.0:
+    liveins: $ymm0
+
+    ; CHECK-LABEL: name:            select_cfb_vec256
+    ; CHECK: [[COPY:%[0-9]+]]:vr256 = COPY $ymm0
+    ; CHECK-NOT: G_CONSTANT_FOLD_BARRIER
+    ; CHECK-NEXT: $ymm1 = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $ymm1
+    %0:vecr(<8 x s32>) = COPY $ymm0
+    %1:vecr(<8 x s32>) = G_CONSTANT_FOLD_BARRIER %0
+    $ymm1 = COPY %1(<8 x s32>)
+    RET 0, implicit $ymm1
+...
diff --git a/llvm/test/CodeGen/X86/GlobalISel/select-constant-fold-barrier-vec512.mir b/llvm/test/CodeGen/X86/GlobalISel/select-constant-fold-barrier-vec512.mir
@@ -0,0 +1,23 @@
+# RUN: llc -mtriple=x86_64-linux-gnu -mattr=+avx512f -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+
+---
+name:            select_cfb_vec512
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: vecr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: vecr, preferred-register: '', flags: [  ] }
+body:             |
+  bb.0:
+    liveins: $zmm0
+
+    ; CHECK-LABEL: name:            select_cfb_vec512
+    ; CHECK: [[COPY:%[0-9]+]]:vr512 = COPY $zmm0
+    ; CHECK-NOT: G_CONSTANT_FOLD_BARRIER
+    ; CHECK-NEXT: $zmm1 = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $zmm1
+    %0:vecr(<8 x s64>) = COPY $zmm0
+    %1:vecr(<8 x s64>) = G_CONSTANT_FOLD_BARRIER %0
+    $zmm1 = COPY %1(<8 x s64>)
+    RET 0, implicit $zmm1
+...
diff --git a/llvm/test/CodeGen/X86/GlobalISel/select-constant-fold-barrier.mir b/llvm/test/CodeGen/X86/GlobalISel/select-constant-fold-barrier.mir
@@ -0,0 +1,77 @@
+# RUN: llc -mtriple=x86_64-linux-gnu -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+
+
+---
+name:            select_cfb_scalar_s32
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: gpr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: gpr, preferred-register: '', flags: [  ] }
+liveins:
+fixedStack:
+stack:
+constants:
+body:             |
+  bb.0:
+    liveins: $edi
+
+    ; CHECK-LABEL: name:            select_cfb_scalar_s32
+    ; CHECK: [[COPY:%[0-9]+]]:gr32 = COPY $edi
+    ; CHECK-NOT: G_CONSTANT_FOLD_BARRIER
+    ; CHECK-NEXT: $eax = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $eax
+    %0:gpr(s32) = COPY $edi
+    %1:gpr(s32) = G_CONSTANT_FOLD_BARRIER %0
+    $eax = COPY %1(s32)
+    RET 0, implicit $eax
+...
+
+---
+name:            select_cfb_scalar_s64
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: gpr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: gpr, preferred-register: '', flags: [  ] }
+liveins:
+fixedStack:
+stack:
+constants:
+body:             |
+  bb.0:
+    liveins: $rdi
+
+    ; CHECK-LABEL: name:            select_cfb_scalar_s64
+    ; CHECK: [[COPY:%[0-9]+]]:gr64 = COPY $rdi
+    ; CHECK-NOT: G_CONSTANT_FOLD_BARRIER
+    ; CHECK-NEXT: $rax = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $rax
+    %0:gpr(s64) = COPY $rdi
+    %1:gpr(s64) = G_CONSTANT_FOLD_BARRIER %0
+    $rax = COPY %1(s64)
+    RET 0, implicit $rax
+...
+
+
+---
+name:            select_cfb_vec128
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: vecr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: vecr, preferred-register: '', flags: [  ] }
+body:             |
+  bb.0:
+    liveins: $xmm0
+
+    ; CHECK-LABEL: name:            select_cfb_vec128
+    ; CHECK: [[COPY:%[0-9]+]]:vr128 = COPY $xmm0
+    ; CHECK-NOT: G_CONSTANT_FOLD_BARRIER
+    ; CHECK-NEXT: $xmm1 = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $xmm1
+    %0:vecr(<4 x s32>) = COPY $xmm0
+    %1:vecr(<4 x s32>) = G_CONSTANT_FOLD_BARRIER %0
+    $xmm1 = COPY %1(<4 x s32>)
+    RET 0, implicit $xmm1
+...
diff --git a/llvm/test/CodeGen/X86/GlobalISel/select-freeze-vec256.mir b/llvm/test/CodeGen/X86/GlobalISel/select-freeze-vec256.mir
@@ -0,0 +1,23 @@
+# RUN: llc -mtriple=x86_64-linux-gnu -mattr=+avx -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+
+---
+name:            select_freeze_vec256
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: vecr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: vecr, preferred-register: '', flags: [  ] }
+body:             |
+  bb.0:
+    liveins: $ymm0
+
+    ; CHECK-LABEL: name:            select_freeze_vec256
+    ; CHECK: [[COPY:%[0-9]+]]:vr256 = COPY $ymm0
+    ; CHECK-NOT: G_FREEZE
+    ; CHECK-NEXT: $ymm1 = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $ymm1
+    %0:vecr(<8 x s32>) = COPY $ymm0
+    %1:vecr(<8 x s32>) = G_FREEZE %0
+    $ymm1 = COPY %1(<8 x s32>)
+    RET 0, implicit $ymm1
+...
diff --git a/llvm/test/CodeGen/X86/GlobalISel/select-freeze-vec512.mir b/llvm/test/CodeGen/X86/GlobalISel/select-freeze-vec512.mir
@@ -0,0 +1,23 @@
+# RUN: llc -mtriple=x86_64-linux-gnu -mattr=+avx512f -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+
+---
+name:            select_freeze_vec512
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: vecr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: vecr, preferred-register: '', flags: [  ] }
+body:             |
+  bb.0:
+    liveins: $zmm0
+
+    ; CHECK-LABEL: name:            select_freeze_vec512
+    ; CHECK: [[COPY:%[0-9]+]]:vr512 = COPY $zmm0
+    ; CHECK-NOT: G_FREEZE
+    ; CHECK-NEXT: $zmm1 = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $zmm1
+    %0:vecr(<8 x s64>) = COPY $zmm0
+    %1:vecr(<8 x s64>) = G_FREEZE %0
+    $zmm1 = COPY %1(<8 x s64>)
+    RET 0, implicit $zmm1
+...
diff --git a/llvm/test/CodeGen/X86/GlobalISel/select-freeze.mir b/llvm/test/CodeGen/X86/GlobalISel/select-freeze.mir
@@ -0,0 +1,77 @@
+# RUN: llc -mtriple=x86_64-linux-gnu -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+
+
+---
+name:            select_freeze_scalar_s32
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: gpr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: gpr, preferred-register: '', flags: [  ] }
+liveins:
+fixedStack:
+stack:
+constants:
+body:             |
+  bb.0:
+    liveins: $edi
+
+    ; CHECK-LABEL: name:            select_freeze_scalar_s32
+    ; CHECK: [[COPY:%[0-9]+]]:gr32 = COPY $edi
+    ; CHECK-NOT: G_FREEZE
+    ; CHECK-NEXT: $eax = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $eax
+    %0:gpr(s32) = COPY $edi
+    %1:gpr(s32) = G_FREEZE %0
+    $eax = COPY %1(s32)
+    RET 0, implicit $eax
+...
+
+---
+name:            select_freeze_scalar_s64
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: gpr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: gpr, preferred-register: '', flags: [  ] }
+liveins:
+fixedStack:
+stack:
+constants:
+body:             |
+  bb.0:
+    liveins: $rdi
+
+    ; CHECK-LABEL: name:            select_freeze_scalar_s64
+    ; CHECK: [[COPY:%[0-9]+]]:gr64 = COPY $rdi
+    ; CHECK-NOT: G_FREEZE
+    ; CHECK-NEXT: $rax = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $rax
+    %0:gpr(s64) = COPY $rdi
+    %1:gpr(s64) = G_FREEZE %0
+    $rax = COPY %1(s64)
+    RET 0, implicit $rax
+...
+
+
+---
+name:            select_freeze_vec128
+legalized:       true
+regBankSelected: true
+registers:
+  - { id: 0, class: vecr, preferred-register: '', flags: [  ] }
+  - { id: 1, class: vecr, preferred-register: '', flags: [  ] }
+body:             |
+  bb.0:
+    liveins: $xmm0
+
+    ; CHECK-LABEL: name:            select_freeze_vec128
+    ; CHECK: [[COPY:%[0-9]+]]:vr128 = COPY $xmm0
+    ; CHECK-NOT: G_FREEZE
+    ; CHECK-NEXT: $xmm1 = COPY [[COPY]]
+    ; CHECK-NEXT: RET 0, implicit $xmm1
+    %0:vecr(<4 x s32>) = COPY $xmm0
+    %1:vecr(<4 x s32>) = G_FREEZE %0
+    $xmm1 = COPY %1(<4 x s32>)
+    RET 0, implicit $xmm1
+...
