[IntrinsicInst] Remove MemCpyInlineInst and MemSetInlineInst [nfc] (#138568)

preames · web-flow · commit c0a264e6a905 · 2025-05-05T14:07:31.000-07:00
I'm looking for ways to simplify the Mem*Inst class structure, and these
two seem to have fairly minimal justification, so let's remove them.
diff --git a/llvm/include/llvm/IR/InstVisitor.h b/llvm/include/llvm/IR/InstVisitor.h
@@ -207,12 +207,10 @@ class InstVisitor {
   RetTy visitDbgLabelInst(DbgLabelInst &I)        { DELEGATE(DbgInfoIntrinsic);}
   RetTy visitDbgInfoIntrinsic(DbgInfoIntrinsic &I){ DELEGATE(IntrinsicInst); }
   RetTy visitMemSetInst(MemSetInst &I)            { DELEGATE(MemIntrinsic); }
-  RetTy visitMemSetInlineInst(MemSetInlineInst &I){ DELEGATE(MemSetInst); }
   RetTy visitMemSetPatternInst(MemSetPatternInst &I) {
     DELEGATE(IntrinsicInst);
   }
   RetTy visitMemCpyInst(MemCpyInst &I)            { DELEGATE(MemTransferInst); }
-  RetTy visitMemCpyInlineInst(MemCpyInlineInst &I){ DELEGATE(MemCpyInst); }
   RetTy visitMemMoveInst(MemMoveInst &I)          { DELEGATE(MemTransferInst); }
   RetTy visitMemTransferInst(MemTransferInst &I)  { DELEGATE(MemIntrinsic); }
   RetTy visitMemIntrinsic(MemIntrinsic &I)        { DELEGATE(IntrinsicInst); }
@@ -291,13 +289,13 @@ class InstVisitor {
       case Intrinsic::dbg_declare: DELEGATE(DbgDeclareInst);
       case Intrinsic::dbg_value:   DELEGATE(DbgValueInst);
       case Intrinsic::dbg_label:   DELEGATE(DbgLabelInst);
-      case Intrinsic::memcpy:      DELEGATE(MemCpyInst);
+      case Intrinsic::memcpy:
       case Intrinsic::memcpy_inline:
-        DELEGATE(MemCpyInlineInst);
+        DELEGATE(MemCpyInst);
       case Intrinsic::memmove:     DELEGATE(MemMoveInst);
-      case Intrinsic::memset:      DELEGATE(MemSetInst);
+      case Intrinsic::memset:
       case Intrinsic::memset_inline:
-        DELEGATE(MemSetInlineInst);
+        DELEGATE(MemSetInst);
       case Intrinsic::experimental_memset_pattern:
         DELEGATE(MemSetPatternInst);
       case Intrinsic::vastart:     DELEGATE(VAStartInst);
diff --git a/llvm/include/llvm/IR/IntrinsicInst.h b/llvm/include/llvm/IR/IntrinsicInst.h
@@ -1215,6 +1215,16 @@ class MemIntrinsic : public MemIntrinsicBase<MemIntrinsic> {
 
   void setVolatile(Constant *V) { setArgOperand(ARG_VOLATILE, V); }
 
+  bool isForceInlined() const {
+    switch (getIntrinsicID()) {
+    case Intrinsic::memset_inline:
+    case Intrinsic::memcpy_inline:
+      return true;
+    default:
+      return false;
+    }
+  }
+
   // Methods for support type inquiry through isa, cast, and dyn_cast:
   static bool classof(const IntrinsicInst *I) {
     switch (I->getIntrinsicID()) {
@@ -1251,18 +1261,6 @@ class MemSetInst : public MemSetBase<MemIntrinsic> {
   }
 };
 
-/// This class wraps the llvm.memset.inline intrinsic.
-class MemSetInlineInst : public MemSetInst {
-public:
-  // Methods for support type inquiry through isa, cast, and dyn_cast:
-  static bool classof(const IntrinsicInst *I) {
-    return I->getIntrinsicID() == Intrinsic::memset_inline;
-  }
-  static bool classof(const Value *V) {
-    return isa<IntrinsicInst>(V) && classof(cast<IntrinsicInst>(V));
-  }
-};
-
 /// This is the base class for llvm.experimental.memset.pattern
 class MemSetPatternIntrinsic : public MemIntrinsicBase<MemIntrinsic> {
 private:
@@ -1342,18 +1340,6 @@ class MemMoveInst : public MemTransferInst {
   }
 };
 
-/// This class wraps the llvm.memcpy.inline intrinsic.
-class MemCpyInlineInst : public MemCpyInst {
-public:
-  // Methods for support type inquiry through isa, cast, and dyn_cast:
-  static bool classof(const IntrinsicInst *I) {
-    return I->getIntrinsicID() == Intrinsic::memcpy_inline;
-  }
-  static bool classof(const Value *V) {
-    return isa<IntrinsicInst>(V) && classof(cast<IntrinsicInst>(V));
-  }
-};
-
 // The common base class for any memset/memmove/memcpy intrinsics;
 // whether they be atomic or non-atomic.
 // i.e. llvm.element.unordered.atomic.memset/memcpy/memmove
diff --git a/llvm/lib/Analysis/Lint.cpp b/llvm/lib/Analysis/Lint.cpp
@@ -344,19 +344,13 @@ void Lint::visitCallBase(CallBase &I) {
                            MMI->getSourceAlign(), nullptr, MemRef::Read);
       break;
     }
-    case Intrinsic::memset: {
+    case Intrinsic::memset:
+    case Intrinsic::memset_inline: {
       MemSetInst *MSI = cast<MemSetInst>(&I);
       visitMemoryReference(I, MemoryLocation::getForDest(MSI),
                            MSI->getDestAlign(), nullptr, MemRef::Write);
       break;
     }
-    case Intrinsic::memset_inline: {
-      MemSetInlineInst *MSII = cast<MemSetInlineInst>(&I);
-      visitMemoryReference(I, MemoryLocation::getForDest(MSII),
-                           MSII->getDestAlign(), nullptr, MemRef::Write);
-      break;
-    }
-
     case Intrinsic::vastart:
       // vastart in non-varargs function is rejected by the verifier
       visitMemoryReference(I, MemoryLocation::getForArgument(&I, 0, TLI),
diff --git a/llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp b/llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp
@@ -1734,10 +1734,6 @@ bool IRTranslator::translateMemFunc(const CallInst &CI,
     DstAlign = MCI->getDestAlign().valueOrOne();
     SrcAlign = MCI->getSourceAlign().valueOrOne();
     CopySize = dyn_cast<ConstantInt>(MCI->getArgOperand(2));
-  } else if (auto *MCI = dyn_cast<MemCpyInlineInst>(&CI)) {
-    DstAlign = MCI->getDestAlign().valueOrOne();
-    SrcAlign = MCI->getSourceAlign().valueOrOne();
-    CopySize = dyn_cast<ConstantInt>(MCI->getArgOperand(2));
   } else if (auto *MMI = dyn_cast<MemMoveInst>(&CI)) {
     DstAlign = MMI->getDestAlign().valueOrOne();
     SrcAlign = MMI->getSourceAlign().valueOrOne();
diff --git a/llvm/lib/CodeGen/PreISelIntrinsicLowering.cpp b/llvm/lib/CodeGen/PreISelIntrinsicLowering.cpp
@@ -319,7 +319,7 @@ bool PreISelIntrinsicLowering::expandMemIntrinsicUses(Function &F) const {
       // Only expand llvm.memcpy.inline with non-constant length in this
       // codepath, leaving the current SelectionDAG expansion for constant
       // length memcpy intrinsics undisturbed.
-      auto *Memcpy = cast<MemCpyInlineInst>(Inst);
+      auto *Memcpy = cast<MemCpyInst>(Inst);
       if (isa<ConstantInt>(Memcpy->getLength()))
         break;
 
@@ -367,7 +367,7 @@ bool PreISelIntrinsicLowering::expandMemIntrinsicUses(Function &F) const {
       // Only expand llvm.memset.inline with non-constant length in this
       // codepath, leaving the current SelectionDAG expansion for constant
       // length memset intrinsics undisturbed.
-      auto *Memset = cast<MemSetInlineInst>(Inst);
+      auto *Memset = cast<MemSetInst>(Inst);
       if (isa<ConstantInt>(Memset->getLength()))
         break;
 
diff --git a/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp b/llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
@@ -6483,7 +6483,7 @@ void SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I,
     return;
   }
   case Intrinsic::memcpy_inline: {
-    const auto &MCI = cast<MemCpyInlineInst>(I);
+    const auto &MCI = cast<MemCpyInst>(I);
     SDValue Dst = getValue(I.getArgOperand(0));
     SDValue Src = getValue(I.getArgOperand(1));
     SDValue Size = getValue(I.getArgOperand(2));
@@ -6503,35 +6503,21 @@ void SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I,
     updateDAGForMaybeTailCall(MC);
     return;
   }
-  case Intrinsic::memset: {
-    const auto &MSI = cast<MemSetInst>(I);
-    SDValue Op1 = getValue(I.getArgOperand(0));
-    SDValue Op2 = getValue(I.getArgOperand(1));
-    SDValue Op3 = getValue(I.getArgOperand(2));
-    // @llvm.memset defines 0 and 1 to both mean no alignment.
-    Align Alignment = MSI.getDestAlign().valueOrOne();
-    bool isVol = MSI.isVolatile();
-    SDValue Root = isVol ? getRoot() : getMemoryRoot();
-    SDValue MS = DAG.getMemset(
-        Root, sdl, Op1, Op2, Op3, Alignment, isVol, /* AlwaysInline */ false,
-        &I, MachinePointerInfo(I.getArgOperand(0)), I.getAAMetadata());
-    updateDAGForMaybeTailCall(MS);
-    return;
-  }
+  case Intrinsic::memset:
   case Intrinsic::memset_inline: {
-    const auto &MSII = cast<MemSetInlineInst>(I);
+    const auto &MSII = cast<MemSetInst>(I);
     SDValue Dst = getValue(I.getArgOperand(0));
     SDValue Value = getValue(I.getArgOperand(1));
     SDValue Size = getValue(I.getArgOperand(2));
-    assert(isa<ConstantSDNode>(Size) && "memset_inline needs constant size");
+    assert((!MSII.isForceInlined() || isa<ConstantSDNode>(Size)) &&
+           "memset_inline needs constant size");
     // @llvm.memset defines 0 and 1 to both mean no alignment.
     Align DstAlign = MSII.getDestAlign().valueOrOne();
     bool isVol = MSII.isVolatile();
     SDValue Root = isVol ? getRoot() : getMemoryRoot();
-    SDValue MC = DAG.getMemset(Root, sdl, Dst, Value, Size, DstAlign, isVol,
-                               /* AlwaysInline */ true, &I,
-                               MachinePointerInfo(I.getArgOperand(0)),
-                               I.getAAMetadata());
+    SDValue MC = DAG.getMemset(
+        Root, sdl, Dst, Value, Size, DstAlign, isVol, MSII.isForceInlined(),
+        &I, MachinePointerInfo(I.getArgOperand(0)), I.getAAMetadata());
     updateDAGForMaybeTailCall(MC);
     return;
   }
diff --git a/llvm/lib/IR/IRBuilder.cpp b/llvm/lib/IR/IRBuilder.cpp
@@ -203,7 +203,7 @@ CallInst *IRBuilderBase::CreateMemSetInline(Value *Dst, MaybeAlign DstAlign,
   CallInst *CI = CreateIntrinsic(Intrinsic::memset_inline, Tys, Ops);
 
   if (DstAlign)
-    cast<MemSetInlineInst>(CI)->setDestAlignment(*DstAlign);
+    cast<MemSetInst>(CI)->setDestAlignment(*DstAlign);
 
   // Set the TBAA info if present.
   if (TBAATag)
diff --git a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
@@ -1142,18 +1142,18 @@ static bool handleMemIntrinsicPtrUse(MemIntrinsic *MI, Value *OldV,
     if (Dest == OldV)
       Dest = NewV;
 
-    if (isa<MemCpyInlineInst>(MTI)) {
+    if (auto *MCI = dyn_cast<MemCpyInst>(MTI)) {
       MDNode *TBAAStruct = MTI->getMetadata(LLVMContext::MD_tbaa_struct);
-      B.CreateMemCpyInline(Dest, MTI->getDestAlign(), Src,
-                           MTI->getSourceAlign(), MTI->getLength(),
-                           false, // isVolatile
-                           TBAA, TBAAStruct, ScopeMD, NoAliasMD);
-    } else if (isa<MemCpyInst>(MTI)) {
-      MDNode *TBAAStruct = MTI->getMetadata(LLVMContext::MD_tbaa_struct);
-      B.CreateMemCpy(Dest, MTI->getDestAlign(), Src, MTI->getSourceAlign(),
-                     MTI->getLength(),
-                     false, // isVolatile
-                     TBAA, TBAAStruct, ScopeMD, NoAliasMD);
+      if (MCI->isForceInlined())
+        B.CreateMemCpyInline(Dest, MTI->getDestAlign(), Src,
+                             MTI->getSourceAlign(), MTI->getLength(),
+                             false, // isVolatile
+                             TBAA, TBAAStruct, ScopeMD, NoAliasMD);
+      else
+        B.CreateMemCpy(Dest, MTI->getDestAlign(), Src, MTI->getSourceAlign(),
+                       MTI->getLength(),
+                       false, // isVolatile
+                       TBAA, TBAAStruct, ScopeMD, NoAliasMD);
     } else {
       assert(isa<MemMoveInst>(MTI));
       B.CreateMemMove(Dest, MTI->getDestAlign(), Src, MTI->getSourceAlign(),
diff --git a/llvm/lib/Transforms/Scalar/LoopIdiomRecognize.cpp b/llvm/lib/Transforms/Scalar/LoopIdiomRecognize.cpp
@@ -780,7 +780,7 @@ bool LoopIdiomRecognize::processLoopMemCpy(MemCpyInst *MCI,
     return false;
 
   // If we're not allowed to hack on memcpy, we fail.
-  if ((!HasMemcpy && !isa<MemCpyInlineInst>(MCI)) || DisableLIRP::Memcpy)
+  if ((!HasMemcpy && !MCI->isForceInlined()) || DisableLIRP::Memcpy)
     return false;
 
   Value *Dest = MCI->getDest();
@@ -1267,7 +1267,7 @@ bool LoopIdiomRecognize::processLoopStoreOfLoopLoad(
   // FIXME: until llvm.memcpy.inline supports dynamic sizes, we need to
   // conservatively bail here, since otherwise we may have to transform
   // llvm.memcpy.inline into llvm.memcpy which is illegal.
-  if (isa<MemCpyInlineInst>(TheStore))
+  if (auto *MCI = dyn_cast<MemCpyInst>(TheStore); MCI && MCI->isForceInlined())
     return false;
 
   // The trip count of the loop and the base pointer of the addrec SCEV is
diff --git a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
@@ -1212,7 +1212,7 @@ bool MemCpyOptPass::processMemCpyMemCpyDependence(MemCpyInst *M,
     // Don't convert llvm.memcpy.inline into memmove because memmove can be
     // lowered as a call, and that is not allowed for llvm.memcpy.inline (and
     // there is no inline version of llvm.memmove)
-    if (isa<MemCpyInlineInst>(M))
+    if (M->isForceInlined())
       return false;
     UseMemMove = true;
   }
@@ -1229,17 +1229,18 @@ bool MemCpyOptPass::processMemCpyMemCpyDependence(MemCpyInst *M,
     NewM =
         Builder.CreateMemMove(M->getDest(), M->getDestAlign(), CopySource,
                               CopySourceAlign, M->getLength(), M->isVolatile());
-  else if (isa<MemCpyInlineInst>(M)) {
+  else if (M->isForceInlined())
     // llvm.memcpy may be promoted to llvm.memcpy.inline, but the converse is
     // never allowed since that would allow the latter to be lowered as a call
     // to an external function.
     NewM = Builder.CreateMemCpyInline(M->getDest(), M->getDestAlign(),
                                       CopySource, CopySourceAlign,
                                       M->getLength(), M->isVolatile());
-  } else
-    NewM =
-        Builder.CreateMemCpy(M->getDest(), M->getDestAlign(), CopySource,
-                             CopySourceAlign, M->getLength(), M->isVolatile());
+  else
+    NewM = Builder.CreateMemCpy(M->getDest(), M->getDestAlign(), CopySource,
+                                CopySourceAlign, M->getLength(),
+                                M->isVolatile());
+
   NewM->copyMetadata(*M, LLVMContext::MD_DIAssignID);
 
   assert(isa<MemoryDef>(MSSA->getMemoryAccess(M)));
