1 files changed, 101 insertions, 1 deletions

diff --git a/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp b/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp
index dd8ecf6ef7fc..bab7fe9d25e4 100644
--- a/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp
+++ b/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp
@@ -315,7 +315,8 @@ X86LegalizerInfo::X86LegalizerInfo(const X86Subtarget &STI,
   getActionDefinitionsBuilder(G_PHI)
       .legalIf([=](const LegalityQuery &Query) -> bool {
         return typeInSet(0, {s8, s16, s32, p0})(Query) ||
-               (Is64Bit && typeInSet(0, {s64})(Query)) ||
+               (UseX87 && typeIs(0, s80)(Query)) ||
+               (Is64Bit && typeIs(0, s64)(Query)) ||
                (HasSSE1 && typeInSet(0, {v16s8, v8s16, v4s32, v2s64})(Query)) ||
                (HasAVX && typeInSet(0, {v32s8, v16s16, v8s32, v4s64})(Query)) ||
                (HasAVX512 &&
@@ -497,6 +498,62 @@ X86LegalizerInfo::X86LegalizerInfo(const X86Subtarget &STI,
       .clampScalar(0, s32, sMaxScalar)
       .widenScalarToNextPow2(1);
 
+  // For G_UITOFP and G_FPTOUI without AVX512, we have to custom legalize types
+  // <= s32 manually. Otherwise, in custom handler there is no way to
+  // understand whether s32 is an original type and we need to promote it to
+  // s64 or s32 is obtained after widening and we shouldn't widen it to s64.
+  //
+  // For AVX512 we simply widen types as there is direct mapping from opcodes
+  // to asm instructions.
+  getActionDefinitionsBuilder(G_UITOFP)
+      .legalIf([=](const LegalityQuery &Query) {
+        return HasAVX512 && typeInSet(0, {s32, s64})(Query) &&
+               typeInSet(1, {s32, s64})(Query);
+      })
+      .customIf([=](const LegalityQuery &Query) {
+        return !HasAVX512 &&
+               ((HasSSE1 && typeIs(0, s32)(Query)) ||
+                (HasSSE2 && typeIs(0, s64)(Query))) &&
+               scalarNarrowerThan(1, Is64Bit ? 64 : 32)(Query);
+      })
+      .lowerIf([=](const LegalityQuery &Query) {
+        // Lower conversions from s64
+        return !HasAVX512 &&
+               ((HasSSE1 && typeIs(0, s32)(Query)) ||
+                (HasSSE2 && typeIs(0, s64)(Query))) &&
+               (Is64Bit && typeIs(1, s64)(Query));
+      })
+      .clampScalar(0, s32, HasSSE2 ? s64 : s32)
+      .widenScalarToNextPow2(0)
+      .clampScalar(1, s32, sMaxScalar)
+      .widenScalarToNextPow2(1);
+
+  getActionDefinitionsBuilder(G_FPTOUI)
+      .legalIf([=](const LegalityQuery &Query) {
+        return HasAVX512 && typeInSet(0, {s32, s64})(Query) &&
+               typeInSet(1, {s32, s64})(Query);
+      })
+      .customIf([=](const LegalityQuery &Query) {
+        return !HasAVX512 &&
+               ((HasSSE1 && typeIs(1, s32)(Query)) ||
+                (HasSSE2 && typeIs(1, s64)(Query))) &&
+               scalarNarrowerThan(0, Is64Bit ? 64 : 32)(Query);
+      })
+      // TODO: replace with customized legalization using
+      // specifics of cvttsd2si. The selection of this node requires
+      // a vector type. Either G_SCALAR_TO_VECTOR is needed or more advanced
+      // support of G_BUILD_VECTOR/G_INSERT_VECTOR_ELT is required beforehand.
+      .lowerIf([=](const LegalityQuery &Query) {
+        return !HasAVX512 &&
+               ((HasSSE1 && typeIs(1, s32)(Query)) ||
+                (HasSSE2 && typeIs(1, s64)(Query))) &&
+               (Is64Bit && typeIs(0, s64)(Query));
+      })
+      .clampScalar(0, s32, sMaxScalar)
+      .widenScalarToNextPow2(0)
+      .clampScalar(1, s32, HasSSE2 ? s64 : s32)
+      .widenScalarToNextPow2(1);
+
   // vector ops
   getActionDefinitionsBuilder(G_BUILD_VECTOR)
       .customIf([=](const LegalityQuery &Query) {
@@ -589,6 +646,10 @@ bool X86LegalizerInfo::legalizeCustom(LegalizerHelper &Helper, MachineInstr &MI,
     return false;
   case TargetOpcode::G_BUILD_VECTOR:
     return legalizeBuildVector(MI, MRI, Helper);
+  case TargetOpcode::G_FPTOUI:
+    return legalizeFPTOUI(MI, MRI, Helper);
+  case TargetOpcode::G_UITOFP:
+    return legalizeUITOFP(MI, MRI, Helper);
   }
   llvm_unreachable("expected switch to return");
 }
@@ -644,6 +705,45 @@ bool X86LegalizerInfo::legalizeBuildVector(MachineInstr &MI,
   return true;
 }
 
+bool X86LegalizerInfo::legalizeFPTOUI(MachineInstr &MI,
+                                      MachineRegisterInfo &MRI,
+                                      LegalizerHelper &Helper) const {
+  MachineIRBuilder &MIRBuilder = Helper.MIRBuilder;
+  auto [Dst, DstTy, Src, SrcTy] = MI.getFirst2RegLLTs();
+  unsigned DstSizeInBits = DstTy.getScalarSizeInBits();
+  const LLT s32 = LLT::scalar(32);
+  const LLT s64 = LLT::scalar(64);
+
+  // Simply reuse FPTOSI when it is possible to widen the type
+  if (DstSizeInBits <= 32) {
+    auto Casted = MIRBuilder.buildFPTOSI(DstTy == s32 ? s64 : s32, Src);
+    MIRBuilder.buildTrunc(Dst, Casted);
+    MI.eraseFromParent();
+    return true;
+  }
+
+  return false;
+}
+
+bool X86LegalizerInfo::legalizeUITOFP(MachineInstr &MI,
+                                      MachineRegisterInfo &MRI,
+                                      LegalizerHelper &Helper) const {
+  MachineIRBuilder &MIRBuilder = Helper.MIRBuilder;
+  auto [Dst, DstTy, Src, SrcTy] = MI.getFirst2RegLLTs();
+  const LLT s32 = LLT::scalar(32);
+  const LLT s64 = LLT::scalar(64);
+
+  // Simply reuse SITOFP when it is possible to widen the type
+  if (SrcTy.getSizeInBits() <= 32) {
+    auto Ext = MIRBuilder.buildZExt(SrcTy == s32 ? s64 : s32, Src);
+    MIRBuilder.buildSITOFP(Dst, Ext);
+    MI.eraseFromParent();
+    return true;
+  }
+
+  return false;
+}
+
 bool X86LegalizerInfo::legalizeIntrinsic(LegalizerHelper &Helper,
                                          MachineInstr &MI) const {
   return true;