rebase

Created using spr 1.3.7

Author: boomanaiden154

clang/include/clang/Basic/LangOptions.def

@@ -454,7 +454,7 @@ LANGOPT(BranchTargetEnforcement, 1, 0, NotCompatible, "Branch-target enforcement
 LANGOPT(BranchProtectionPAuthLR, 1, 0, NotCompatible, "Use PC as a diversifier using PAuthLR NOP instructions.")
 LANGOPT(GuardedControlStack, 1, 0, NotCompatible, "Guarded control stack enabled")
 
-LANGOPT(SpeculativeLoadHardening, 1, 0, NotCompatible, "Speculative load hardening enabled")
+LANGOPT(SpeculativeLoadHardening, 1, 0, Benign, "Speculative load hardening enabled")
 
 LANGOPT(RelativeCXXABIVTables, 1, 0, NotCompatible,
         "Use an ABI-incompatible v-table layout that uses relative references")

clang/test/ClangScanDeps/strip-codegen-args.m

@@ -16,6 +16,7 @@
 // CHECK-NOT:          "-flto"
 // CHECK-NOT:          "-fno-autolink"
 // CHECK-NOT:          "-mrelax-relocations=no"
+// CHECK-NOT:          "-mspeculative-load-hardening"
 // CHECK:            ]
 // CHECK:            "name": "A"
 // CHECK:          }
@@ -39,6 +40,11 @@
     "command": "clang -Imodules/A -fmodules -fmodules-cache-path=DIR/module-cache -fimplicit-modules -O2 -flto=full -fsyntax-only DIR/t3.m",
     "file": "DIR/t2.m"
   }
+  {
+    "directory": "DIR",
+    "command": "clang -Imodules/A -fmodules -fmodules-cache-path=DIR/module-cache -fimplicit-modules -O2 -mspeculative-load-hardening -fsyntax-only DIR/t3.m",
+    "file": "DIR/t3.m"
+  }
 ]
 
 //--- modules/A/module.modulemap

flang/lib/Optimizer/Builder/CUDAIntrinsicCall.cpp

@@ -1122,13 +1122,7 @@ CUDAIntrinsicLibrary::genSyncThreadsOr(mlir::Type resultType,
 void CUDAIntrinsicLibrary::genSyncWarp(
     llvm::ArrayRef<fir::ExtendedValue> args) {
   assert(args.size() == 1);
-  constexpr llvm::StringLiteral funcName = "llvm.nvvm.bar.warp.sync";
-  mlir::Value mask = fir::getBase(args[0]);
-  mlir::FunctionType funcType =
-      mlir::FunctionType::get(builder.getContext(), {mask.getType()}, {});
-  auto funcOp = builder.createFunction(loc, funcName, funcType);
-  llvm::SmallVector<mlir::Value> argsList{mask};
-  fir::CallOp::create(builder, loc, funcOp, argsList);
+  mlir::NVVM::SyncWarpOp::create(builder, loc, fir::getBase(args[0]));
 }
 
 // THIS_GRID

flang/test/Lower/CUDA/cuda-device-proc.cuf

@@ -105,7 +105,7 @@ end
 
 ! CHECK-LABEL: func.func @_QPdevsub() attributes {cuf.proc_attr = #cuf.cuda_proc<global>}
 ! CHECK: nvvm.barrier0
-! CHECK: fir.call @llvm.nvvm.bar.warp.sync(%c1{{.*}}) fastmath<contract> : (i32) -> ()
+! CHECK: nvvm.bar.warp.sync %c1{{.*}} : i32 
 ! CHECK: fir.call @llvm.nvvm.membar.gl() fastmath<contract> : () -> ()
 ! CHECK: fir.call @llvm.nvvm.membar.cta() fastmath<contract> : () -> ()
 ! CHECK: fir.call @llvm.nvvm.membar.sys() fastmath<contract> : () -> ()
@@ -219,7 +219,7 @@ end
 ! CHECK-LABEL: func.func @_QPhost1()
 ! CHECK: cuf.kernel
 ! CHECK: nvvm.barrier0
-! CHECK: fir.call @llvm.nvvm.bar.warp.sync(%c1{{.*}}) fastmath<contract> : (i32) -> ()
+! CHECK: nvvm.bar.warp.sync %c1{{.*}} : i32 
 ! CHECK: fir.call @llvm.nvvm.barrier0.and(%c1{{.*}}) fastmath<contract> : (i32) -> i32
 ! CHECK: fir.call @llvm.nvvm.barrier0.popc(%c1{{.*}}) fastmath<contract> : (i32) -> i32
 ! CHECK: fir.call @llvm.nvvm.barrier0.or(%c1{{.*}}) fastmath<contract> : (i32) -> i32

libcxx/utils/ci/buildkite-pipeline.yml

@@ -103,7 +103,6 @@ steps:
       queue: libcxx-builders
       os: aix
     <<: *common
-    skip: "https://github.com/llvm/llvm-project/issues/162516"
 
   - label: AIX (64-bit)
     command: libcxx/utils/ci/run-buildbot aix
@@ -115,7 +114,6 @@ steps:
       queue: libcxx-builders
       os: aix
     <<: *common
-    skip: "https://github.com/llvm/llvm-project/issues/162516"
 
 - group: ':freebsd: FreeBSD'
   steps:

llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/LibraryResolver.h

@@ -211,6 +211,21 @@ class LibraryManager {
     return FilteredView(Libraries.begin(), Libraries.end(), S, K);
   }
 
+  using LibraryFilterFn = std::function<bool(const LibraryInfo &)>;
+  void getLibraries(LibState S, PathType K,
+                    std::vector<std::shared_ptr<LibraryInfo>> &Outs,
+                    LibraryFilterFn Filter = nullptr) const {
+    std::shared_lock<std::shared_mutex> Lock(Mtx);
+    for (const auto &[_, Entry] : Libraries) {
+      const auto &Info = *Entry;
+      if (Info.getKind() != K || Info.getState() != S)
+        continue;
+      if (Filter && !Filter(Info))
+        continue;
+      Outs.push_back(Entry);
+    }
+  }
+
   void forEachLibrary(const LibraryVisitor &visitor) const {
     std::unique_lock<std::shared_mutex> Lock(Mtx);
     for (const auto &[_, entry] : Libraries) {
@@ -220,14 +235,14 @@ class LibraryManager {
   }
 
   bool isLoaded(StringRef Path) const {
-    std::unique_lock<std::shared_mutex> Lock(Mtx);
+    std::shared_lock<std::shared_mutex> Lock(Mtx);
     if (auto It = Libraries.find(Path.str()); It != Libraries.end())
       return It->second->getState() == LibState::Loaded;
     return false;
   }
 
   bool isQueried(StringRef Path) const {
-    std::unique_lock<std::shared_mutex> Lock(Mtx);
+    std::shared_lock<std::shared_mutex> Lock(Mtx);
     if (auto It = Libraries.find(Path.str()); It != Libraries.end())
       return It->second->getState() == LibState::Queried;
     return false;

llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

@@ -4046,6 +4046,8 @@ static SDValue foldSubCtlzNot(SDNode *N, SelectionDAG &DAG) {
                                     m_ConstInt(AndMask)))) {
     // Type Legalisation Pattern:
     // (sub (ctlz (and (xor Op XorMask) AndMask)) BitWidthDiff)
+    if (BitWidthDiff.getZExtValue() >= BitWidth)
+      return SDValue();
     unsigned AndMaskWidth = BitWidth - BitWidthDiff.getZExtValue();
     if (!(AndMask.isMask(AndMaskWidth) && XorMask.countr_one() >= AndMaskWidth))
       return SDValue();

llvm/lib/ExecutionEngine/Orc/TargetProcess/LibraryResolver.cpp

@@ -184,17 +184,17 @@ class SymbolSearchContext {
 public:
   SymbolSearchContext(SymbolQuery &Q) : Q(Q) {}
 
-  bool hasSearched(LibraryInfo *Lib) const { return Searched.count(Lib); }
+  bool hasSearched(const LibraryInfo *Lib) const { return Searched.count(Lib); }
 
-  void markSearched(LibraryInfo *Lib) { Searched.insert(Lib); }
+  void markSearched(const LibraryInfo *Lib) { Searched.insert(Lib); }
 
   inline bool allResolved() const { return Q.allResolved(); }
 
   SymbolQuery &query() { return Q; }
 
 private:
   SymbolQuery &Q;
-  DenseSet<LibraryInfo *> Searched;
+  DenseSet<const LibraryInfo *> Searched;
 };
 
 void LibraryResolver::resolveSymbolsInLibrary(
@@ -226,19 +226,18 @@ void LibraryResolver::resolveSymbolsInLibrary(
           return EnumerateResult::Continue;
         },
         Opts);
+  };
 
+  if (!Lib.hasFilter()) {
+    LLVM_DEBUG(dbgs() << "Building filter for library: " << Lib.getFullPath()
+                      << "\n";);
+    enumerateSymbolsIfNeeded();
     if (DiscoveredSymbols.empty()) {
       LLVM_DEBUG(dbgs() << "  No symbols and remove library : "
                         << Lib.getFullPath() << "\n";);
       LibMgr.removeLibrary(Lib.getFullPath());
       return;
     }
-  };
-
-  if (!Lib.hasFilter()) {
-    LLVM_DEBUG(dbgs() << "Building filter for library: " << Lib.getFullPath()
-                      << "\n";);
-    enumerateSymbolsIfNeeded();
     SmallVector<StringRef> SymbolVec;
     SymbolVec.reserve(DiscoveredSymbols.size());
     for (const auto &KV : DiscoveredSymbols)
@@ -288,24 +287,22 @@ void LibraryResolver::searchSymbolsInLibraries(
 
     SymbolSearchContext Ctx(Q);
     while (!Ctx.allResolved()) {
+      std::vector<std::shared_ptr<LibraryInfo>> Libs;
+      LibMgr.getLibraries(S, K, Libs, [&](const LibraryInfo &Lib) {
+        return !Ctx.hasSearched(&Lib);
+      });
 
-      for (auto &Lib : LibMgr.getView(S, K)) {
-        if (Ctx.hasSearched(Lib.get()))
-          continue;
+      if (Libs.empty() && !scanLibrariesIfNeeded(K, scanBatchSize))
+        break; // no more new libs to scan
 
+      for (auto &Lib : Libs) {
         // can use Async here?
         resolveSymbolsInLibrary(*Lib, Ctx.query(), Config.Options);
         Ctx.markSearched(Lib.get());
 
         if (Ctx.allResolved())
           return;
       }
-
-      if (Ctx.allResolved())
-        return;
-
-      if (!scanLibrariesIfNeeded(K, scanBatchSize))
-        break; // no more new libs to scan
     }
   };
 

llvm/lib/ExecutionEngine/Orc/TargetProcess/LibraryScanner.cpp

@@ -50,7 +50,7 @@ void handleError(Error Err, StringRef context = "") {
 }
 
 bool ObjectFileLoader::isArchitectureCompatible(const object::ObjectFile &Obj) {
-  Triple HostTriple(sys::getDefaultTargetTriple());
+  Triple HostTriple(sys::getProcessTriple());
   Triple ObjTriple = Obj.makeTriple();
 
   LLVM_DEBUG({

llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h

@@ -312,7 +312,7 @@ class AArch64TTIImpl final : public BasicTTIImplBase<AArch64TTIImpl> {
   }
 
   bool isLegalMaskedLoadStore(Type *DataType, Align Alignment) const {
-    if (!ST->hasSVE())
+    if (!ST->isSVEorStreamingSVEAvailable())
       return false;
 
     // For fixed vectors, avoid scalarization if using SVE for them.