Merge branch 'main' into remove-POST_BUILD

xingxue-ibm · web-flow · commit f54ffd8d6e5d · 2025-10-15T14:55:27.000-04:00
diff --git a/flang/lib/Semantics/check-omp-atomic.cpp b/flang/lib/Semantics/check-omp-atomic.cpp
@@ -519,8 +519,8 @@ struct AtomicAnalysis {
 ///   function references with scalar data pointer result of non-character
 ///   intrinsic type or variables that are non-polymorphic scalar pointers
 ///   and any length type parameter must be constant.
-void OmpStructureChecker::CheckAtomicType(
-    SymbolRef sym, parser::CharBlock source, std::string_view name) {
+void OmpStructureChecker::CheckAtomicType(SymbolRef sym,
+    parser::CharBlock source, std::string_view name, bool checkTypeOnPointer) {
   const DeclTypeSpec *typeSpec{sym->GetType()};
   if (!typeSpec) {
     return;
@@ -547,6 +547,22 @@ void OmpStructureChecker::CheckAtomicType(
     return;
   }
 
+  // Apply pointer-to-non-intrinsic rule only for intrinsic-assignment paths.
+  if (checkTypeOnPointer) {
+    using Category = DeclTypeSpec::Category;
+    Category cat{typeSpec->category()};
+    if (cat != Category::Numeric && cat != Category::Logical) {
+      std::string details = " has the POINTER attribute";
+      if (const auto *derived{typeSpec->AsDerived()}) {
+        details += " and derived type '"s + derived->name().ToString() + "'";
+      }
+      context_.Say(source,
+          "ATOMIC operation requires an intrinsic scalar variable; '%s'%s"_err_en_US,
+          sym->name(), details);
+      return;
+    }
+  }
+
   // Go over all length parameters, if any, and check if they are
   // explicit.
   if (const DerivedTypeSpec *derived{typeSpec->AsDerived()}) {
@@ -562,7 +578,7 @@ void OmpStructureChecker::CheckAtomicType(
 }
 
 void OmpStructureChecker::CheckAtomicVariable(
-    const SomeExpr &atom, parser::CharBlock source) {
+    const SomeExpr &atom, parser::CharBlock source, bool checkTypeOnPointer) {
   if (atom.Rank() != 0) {
     context_.Say(source, "Atomic variable %s should be a scalar"_err_en_US,
         atom.AsFortran());
@@ -572,7 +588,7 @@ void OmpStructureChecker::CheckAtomicVariable(
   assert(dsgs.size() == 1 && "Should have a single top-level designator");
   evaluate::SymbolVector syms{evaluate::GetSymbolVector(dsgs.front())};
 
-  CheckAtomicType(syms.back(), source, atom.AsFortran());
+  CheckAtomicType(syms.back(), source, atom.AsFortran(), checkTypeOnPointer);
 
   if (IsAllocatable(syms.back()) && !IsArrayElement(atom)) {
     context_.Say(source, "Atomic variable %s cannot be ALLOCATABLE"_err_en_US,
@@ -789,7 +805,8 @@ void OmpStructureChecker::CheckAtomicCaptureAssignment(
   if (!IsVarOrFunctionRef(atom)) {
     ErrorShouldBeVariable(atom, rsrc);
   } else {
-    CheckAtomicVariable(atom, rsrc);
+    CheckAtomicVariable(
+        atom, rsrc, /*checkTypeOnPointer=*/!IsPointerAssignment(capture));
     // This part should have been checked prior to calling this function.
     assert(*GetConvertInput(capture.rhs) == atom &&
         "This cannot be a capture assignment");
@@ -808,7 +825,8 @@ void OmpStructureChecker::CheckAtomicReadAssignment(
     if (!IsVarOrFunctionRef(atom)) {
       ErrorShouldBeVariable(atom, rsrc);
     } else {
-      CheckAtomicVariable(atom, rsrc);
+      CheckAtomicVariable(
+          atom, rsrc, /*checkTypeOnPointer=*/!IsPointerAssignment(read));
       CheckStorageOverlap(atom, {read.lhs}, source);
     }
   } else {
@@ -829,7 +847,8 @@ void OmpStructureChecker::CheckAtomicWriteAssignment(
   if (!IsVarOrFunctionRef(atom)) {
     ErrorShouldBeVariable(atom, rsrc);
   } else {
-    CheckAtomicVariable(atom, lsrc);
+    CheckAtomicVariable(
+        atom, lsrc, /*checkTypeOnPointer=*/!IsPointerAssignment(write));
     CheckStorageOverlap(atom, {write.rhs}, source);
   }
 }
@@ -854,7 +873,8 @@ OmpStructureChecker::CheckAtomicUpdateAssignment(
     return std::nullopt;
   }
 
-  CheckAtomicVariable(atom, lsrc);
+  CheckAtomicVariable(
+      atom, lsrc, /*checkTypeOnPointer=*/!IsPointerAssignment(update));
 
   auto [hasErrors, tryReassoc]{CheckAtomicUpdateAssignmentRhs(
       atom, update.rhs, source, /*suppressDiagnostics=*/true)};
@@ -1017,7 +1037,8 @@ void OmpStructureChecker::CheckAtomicConditionalUpdateAssignment(
     return;
   }
 
-  CheckAtomicVariable(atom, alsrc);
+  CheckAtomicVariable(
+      atom, alsrc, /*checkTypeOnPointer=*/!IsPointerAssignment(assign));
 
   auto top{GetTopLevelOperationIgnoreResizing(cond)};
   // Missing arguments to operations would have been diagnosed by now.
diff --git a/flang/lib/Semantics/check-omp-structure.h b/flang/lib/Semantics/check-omp-structure.h
@@ -262,10 +262,10 @@ class OmpStructureChecker
   void CheckStorageOverlap(const evaluate::Expr<evaluate::SomeType> &,
       llvm::ArrayRef<evaluate::Expr<evaluate::SomeType>>, parser::CharBlock);
   void ErrorShouldBeVariable(const MaybeExpr &expr, parser::CharBlock source);
-  void CheckAtomicType(
-      SymbolRef sym, parser::CharBlock source, std::string_view name);
-  void CheckAtomicVariable(
-      const evaluate::Expr<evaluate::SomeType> &, parser::CharBlock);
+  void CheckAtomicType(SymbolRef sym, parser::CharBlock source,
+      std::string_view name, bool checkTypeOnPointer = true);
+  void CheckAtomicVariable(const evaluate::Expr<evaluate::SomeType> &,
+      parser::CharBlock, bool checkTypeOnPointer = true);
   std::pair<const parser::ExecutionPartConstruct *,
       const parser::ExecutionPartConstruct *>
   CheckUpdateCapture(const parser::ExecutionPartConstruct *ec1,
diff --git a/flang/test/Semantics/OpenMP/omp-atomic-write-pointer-derived.f90 b/flang/test/Semantics/OpenMP/omp-atomic-write-pointer-derived.f90
@@ -0,0 +1,8 @@
+! RUN: not %flang_fc1 -fopenmp -fsyntax-only %s 2>&1 | FileCheck %s
+type t
+end type
+type(t), pointer :: a1, a2
+!$omp atomic write
+a1 = a2
+! CHECK: error: ATOMIC operation requires an intrinsic scalar variable; 'a1' has the POINTER attribute and derived type 't'
+end
diff --git a/llvm/include/llvm/ADT/StringSwitch.h b/llvm/include/llvm/ADT/StringSwitch.h
@@ -66,7 +66,7 @@ class StringSwitch {
 
   // Case-sensitive case matchers
   StringSwitch &Case(StringLiteral S, T Value) {
-    CaseImpl(Value, S);
+    CaseImpl(S, Value);
     return *this;
   }
 
@@ -86,68 +86,68 @@ class StringSwitch {
 
   StringSwitch &Cases(std::initializer_list<StringLiteral> CaseStrings,
                       T Value) {
-    return CasesImpl(Value, CaseStrings);
+    return CasesImpl(CaseStrings, Value);
   }
 
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, T Value) {
-    return CasesImpl(Value, {S0, S1});
+    return CasesImpl({S0, S1}, Value);
   }
 
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                       T Value) {
-    return CasesImpl(Value, {S0, S1, S2});
+    return CasesImpl({S0, S1, S2}, Value);
   }
 
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                       StringLiteral S3, T Value) {
-    return CasesImpl(Value, {S0, S1, S2, S3});
+    return CasesImpl({S0, S1, S2, S3}, Value);
   }
 
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                       StringLiteral S3, StringLiteral S4, T Value) {
-    return CasesImpl(Value, {S0, S1, S2, S3, S4});
+    return CasesImpl({S0, S1, S2, S3, S4}, Value);
   }
 
   [[deprecated("Pass cases in std::initializer_list instead")]]
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                       StringLiteral S3, StringLiteral S4, StringLiteral S5,
                       T Value) {
-    return CasesImpl(Value, {S0, S1, S2, S3, S4, S5});
+    return CasesImpl({S0, S1, S2, S3, S4, S5}, Value);
   }
 
   [[deprecated("Pass cases in std::initializer_list instead")]]
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                       StringLiteral S3, StringLiteral S4, StringLiteral S5,
                       StringLiteral S6, T Value) {
-    return CasesImpl(Value, {S0, S1, S2, S3, S4, S5, S6});
+    return CasesImpl({S0, S1, S2, S3, S4, S5, S6}, Value);
   }
 
   [[deprecated("Pass cases in std::initializer_list instead")]]
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                       StringLiteral S3, StringLiteral S4, StringLiteral S5,
                       StringLiteral S6, StringLiteral S7, T Value) {
-    return CasesImpl(Value, {S0, S1, S2, S3, S4, S5, S6, S7});
+    return CasesImpl({S0, S1, S2, S3, S4, S5, S6, S7}, Value);
   }
 
   [[deprecated("Pass cases in std::initializer_list instead")]]
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                       StringLiteral S3, StringLiteral S4, StringLiteral S5,
                       StringLiteral S6, StringLiteral S7, StringLiteral S8,
                       T Value) {
-    return CasesImpl(Value, {S0, S1, S2, S3, S4, S5, S6, S7, S8});
+    return CasesImpl({S0, S1, S2, S3, S4, S5, S6, S7, S8}, Value);
   }
 
   [[deprecated("Pass cases in std::initializer_list instead")]]
   StringSwitch &Cases(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                       StringLiteral S3, StringLiteral S4, StringLiteral S5,
                       StringLiteral S6, StringLiteral S7, StringLiteral S8,
                       StringLiteral S9, T Value) {
-    return CasesImpl(Value, {S0, S1, S2, S3, S4, S5, S6, S7, S8, S9});
+    return CasesImpl({S0, S1, S2, S3, S4, S5, S6, S7, S8, S9}, Value);
   }
 
   // Case-insensitive case matchers.
   StringSwitch &CaseLower(StringLiteral S, T Value) {
-    CaseLowerImpl(Value, S);
+    CaseLowerImpl(S, Value);
     return *this;
   }
 
@@ -167,26 +167,26 @@ class StringSwitch {
 
   StringSwitch &CasesLower(std::initializer_list<StringLiteral> CaseStrings,
                            T Value) {
-    return CasesLowerImpl(Value, CaseStrings);
+    return CasesLowerImpl(CaseStrings, Value);
   }
 
   StringSwitch &CasesLower(StringLiteral S0, StringLiteral S1, T Value) {
-    return CasesLowerImpl(Value, {S0, S1});
+    return CasesLowerImpl({S0, S1}, Value);
   }
 
   StringSwitch &CasesLower(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                            T Value) {
-    return CasesLowerImpl(Value, {S0, S1, S2});
+    return CasesLowerImpl({S0, S1, S2}, Value);
   }
 
   StringSwitch &CasesLower(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                            StringLiteral S3, T Value) {
-    return CasesLowerImpl(Value, {S0, S1, S2, S3});
+    return CasesLowerImpl({S0, S1, S2, S3}, Value);
   }
 
   StringSwitch &CasesLower(StringLiteral S0, StringLiteral S1, StringLiteral S2,
                            StringLiteral S3, StringLiteral S4, T Value) {
-    return CasesLowerImpl(Value, {S0, S1, S2, S3, S4});
+    return CasesLowerImpl({S0, S1, S2, S3, S4}, Value);
   }
 
   [[nodiscard]] R Default(T Value) {
@@ -207,7 +207,7 @@ class StringSwitch {
 
 private:
   // Returns true when `Str` matches the `S` argument, and stores the result.
-  bool CaseImpl(T &Value, StringLiteral S) {
+  bool CaseImpl(StringLiteral S, T &Value) {
     if (!Result && Str == S) {
       Result = std::move(Value);
       return true;
@@ -217,28 +217,28 @@ class StringSwitch {
 
   // Returns true when `Str` matches the `S` argument (case-insensitive), and
   // stores the result.
-  bool CaseLowerImpl(T &Value, StringLiteral S) {
+  bool CaseLowerImpl(StringLiteral S, T &Value) {
     if (!Result && Str.equals_insensitive(S)) {
       Result = std::move(Value);
       return true;
     }
     return false;
   }
 
-  StringSwitch &CasesImpl(T &Value,
-                          std::initializer_list<StringLiteral> Cases) {
+  StringSwitch &CasesImpl(std::initializer_list<StringLiteral> Cases,
+                          T &Value) {
     // Stop matching after the string is found.
     for (StringLiteral S : Cases)
-      if (CaseImpl(Value, S))
+      if (CaseImpl(S, Value))
         break;
     return *this;
   }
 
-  StringSwitch &CasesLowerImpl(T &Value,
-                               std::initializer_list<StringLiteral> Cases) {
+  StringSwitch &CasesLowerImpl(std::initializer_list<StringLiteral> Cases,
+                               T &Value) {
     // Stop matching after the string is found.
     for (StringLiteral S : Cases)
-      if (CaseLowerImpl(Value, S))
+      if (CaseLowerImpl(S, Value))
         break;
     return *this;
   }
diff --git a/llvm/lib/Target/AArch64/AArch64InstrGISel.td b/llvm/lib/Target/AArch64/AArch64InstrGISel.td
@@ -233,6 +233,12 @@ def G_SDOT : AArch64GenericInstruction {
   let hasSideEffects = 0;
 }
 
+def G_USDOT : AArch64GenericInstruction {
+  let OutOperandList = (outs type0:$dst);
+  let InOperandList = (ins type0:$src1, type0:$src2, type0:$src3);
+  let hasSideEffects = 0;
+}
+
 // Generic instruction for the BSP pseudo. It is expanded into BSP, which
 // expands into BSL/BIT/BIF after register allocation.
 def G_BSP : AArch64GenericInstruction {
@@ -278,6 +284,7 @@ def : GINodeEquiv<G_UADDLV, AArch64uaddlv>;
 
 def : GINodeEquiv<G_UDOT, AArch64udot>;
 def : GINodeEquiv<G_SDOT, AArch64sdot>;
+def : GINodeEquiv<G_USDOT, AArch64usdot>;
 
 def : GINodeEquiv<G_EXTRACT_VECTOR_ELT, vector_extract>;
 
diff --git a/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp b/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp
@@ -1855,6 +1855,8 @@ bool AArch64LegalizerInfo::legalizeIntrinsic(LegalizerHelper &Helper,
     return LowerTriOp(AArch64::G_UDOT);
   case Intrinsic::aarch64_neon_sdot:
     return LowerTriOp(AArch64::G_SDOT);
+  case Intrinsic::aarch64_neon_usdot:
+    return LowerTriOp(AArch64::G_USDOT);
   case Intrinsic::aarch64_neon_sqxtn:
     return LowerUnaryOp(TargetOpcode::G_TRUNC_SSAT_S);
   case Intrinsic::aarch64_neon_sqxtun:
diff --git a/llvm/lib/Target/X86/X86InstrCompiler.td b/llvm/lib/Target/X86/X86InstrCompiler.td
@@ -1256,8 +1256,17 @@ def : Pat<(i64 (X86Wrapper tconstpool  :$dst)),
           (MOV64ri32 tconstpool  :$dst)>, Requires<[KernelCode]>;
 def : Pat<(i64 (X86Wrapper tjumptable  :$dst)),
           (MOV64ri32 tjumptable  :$dst)>, Requires<[KernelCode]>;
-def : Pat<(i64 (X86Wrapper tglobaladdr :$dst)),
-          (MOV64ri32 tglobaladdr :$dst)>, Requires<[KernelCode]>;
+
+// If the globaladdr is an absolute_symbol, don't bother using the sign extending
+// instruction since there's no benefit to using it with absolute symbols.
+def globalAddrNoAbsSym : PatLeaf<(tglobaladdr:$dst), [{
+  auto *GA = cast<GlobalAddressSDNode>(N);
+  return !GA->getGlobal()->getAbsoluteSymbolRange();
+}]>;
+def : Pat<(i64 (X86Wrapper globalAddrNoAbsSym:$dst)),
+          (MOV64ri32 tglobaladdr:$dst)>,
+      Requires<[KernelCode]>;
+
 def : Pat<(i64 (X86Wrapper texternalsym:$dst)),
           (MOV64ri32 texternalsym:$dst)>, Requires<[KernelCode]>;
 def : Pat<(i64 (X86Wrapper mcsym:$dst)),
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -4267,12 +4267,14 @@ class VPlan {
       BackedgeTakenCount = new VPValue();
     return BackedgeTakenCount;
   }
+  VPValue *getBackedgeTakenCount() const { return BackedgeTakenCount; }
 
   /// The vector trip count.
   VPValue &getVectorTripCount() { return VectorTripCount; }
 
   /// Returns the VF of the vector loop region.
   VPValue &getVF() { return VF; };
+  const VPValue &getVF() const { return VF; };
 
   /// Returns VF * UF of the vector loop region.
   VPValue &getVFxUF() { return VFxUF; }
diff --git a/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h b/llvm/lib/Transforms/Vectorize/VPlanPatternMatch.h
@@ -400,6 +400,12 @@ m_AnyOf(const Op0_t &Op0) {
   return m_VPInstruction<VPInstruction::AnyOf>(Op0);
 }
 
+template <typename Op0_t>
+inline VPInstruction_match<VPInstruction::FirstActiveLane, Op0_t>
+m_FirstActiveLane(const Op0_t &Op0) {
+  return m_VPInstruction<VPInstruction::FirstActiveLane>(Op0);
+}
+
 template <unsigned Opcode, typename Op0_t>
 inline AllRecipe_match<Opcode, Op0_t> m_Unary(const Op0_t &Op0) {
   return AllRecipe_match<Opcode, Op0_t>(Op0);
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -788,9 +788,7 @@ static VPValue *optimizeEarlyExitInductionUser(VPlan &Plan,
                                                ScalarEvolution &SE) {
   VPValue *Incoming, *Mask;
   if (!match(Op, m_VPInstruction<VPInstruction::ExtractLane>(
-                     m_VPInstruction<VPInstruction::FirstActiveLane>(
-                         m_VPValue(Mask)),
-                     m_VPValue(Incoming))))
+                     m_FirstActiveLane(m_VPValue(Mask)), m_VPValue(Incoming))))
     return nullptr;
 
   auto *WideIV = getOptimizableIVOf(Incoming, SE);
diff --git a/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp b/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp
@@ -352,8 +352,7 @@ void UnrollState::unrollBlock(VPBlockBase *VPB) {
     // Compute*Result which combine all parts to compute the final value.
     VPValue *Op1;
     if (match(&R, m_VPInstruction<VPInstruction::AnyOf>(m_VPValue(Op1))) ||
-        match(&R, m_VPInstruction<VPInstruction::FirstActiveLane>(
-                      m_VPValue(Op1))) ||
+        match(&R, m_FirstActiveLane(m_VPValue(Op1))) ||
         match(&R, m_VPInstruction<VPInstruction::ComputeAnyOfResult>(
                       m_VPValue(), m_VPValue(), m_VPValue(Op1))) ||
         match(&R, m_VPInstruction<VPInstruction::ComputeReductionResult>(
diff --git a/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp b/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp
diff --git a/llvm/lib/Transforms/Vectorize/VPlanUtils.h b/llvm/lib/Transforms/Vectorize/VPlanUtils.h
diff --git a/llvm/test/CodeGen/AArch64/aarch64-matmul.ll b/llvm/test/CodeGen/AArch64/aarch64-matmul.ll
diff --git a/llvm/test/CodeGen/X86/absolute-symbol-kernel-code-model.ll b/llvm/test/CodeGen/X86/absolute-symbol-kernel-code-model.ll
diff --git a/offload/test/offloading/interop-print.c b/offload/test/offloading/interop-print.c
