Merge branch 'main' into add-tests-for-assign

Author: ldionne

bolt/include/bolt/Passes/ADRRelaxationPass.h

@@ -25,7 +25,8 @@ namespace bolt {
 
 class ADRRelaxationPass : public BinaryFunctionPass {
 public:
-  explicit ADRRelaxationPass() : BinaryFunctionPass(false) {}
+  explicit ADRRelaxationPass(const cl::opt<bool> &PrintPass)
+      : BinaryFunctionPass(PrintPass) {}
 
   const char *getName() const override { return "adr-relaxation"; }
 

bolt/lib/Passes/LongJmp.cpp

@@ -138,9 +138,9 @@ BinaryBasicBlock *LongJmpPass::lookupStubFromGroup(
           const std::pair<uint64_t, BinaryBasicBlock *> &RHS) {
         return LHS.first < RHS.first;
       });
-  if (Cand == Candidates.end())
-    return nullptr;
-  if (Cand != Candidates.begin()) {
+  if (Cand == Candidates.end()) {
+    Cand = std::prev(Cand);
+  } else if (Cand != Candidates.begin()) {
     const StubTy *LeftCand = std::prev(Cand);
     if (Cand->first - DotAddress > DotAddress - LeftCand->first)
       Cand = LeftCand;

bolt/lib/Rewrite/BinaryPassManager.cpp

@@ -126,6 +126,11 @@ static cl::opt<bool> PrintJTFootprintReduction(
     cl::desc("print function after jt-footprint-reduction pass"), cl::Hidden,
     cl::cat(BoltOptCategory));
 
+static cl::opt<bool>
+    PrintAdrRelaxation("print-adr-relaxation",
+                       cl::desc("print functions after ADR Relaxation pass"),
+                       cl::Hidden, cl::cat(BoltOptCategory));
+
 static cl::opt<bool>
     PrintLongJmp("print-longjmp",
                  cl::desc("print functions after longjmp pass"), cl::Hidden,
@@ -493,7 +498,8 @@ Error BinaryFunctionPassManager::runAllPasses(BinaryContext &BC) {
   Manager.registerPass(std::make_unique<ReorderData>());
 
   if (BC.isAArch64()) {
-    Manager.registerPass(std::make_unique<ADRRelaxationPass>());
+    Manager.registerPass(
+        std::make_unique<ADRRelaxationPass>(PrintAdrRelaxation));
 
     // Tighten branches according to offset differences between branch and
     // targets. No extra instructions after this pass, otherwise we may have

bolt/test/AArch64/long-jmp-one-stub.s

@@ -0,0 +1,32 @@
+## This test verifies that no unnecessary stubs are inserted when each DotAddress increases during a lookup.
+
+# REQUIRES: system-linux, asserts
+
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %s -o %t.o
+# RUN: %clang %cflags -O0 %t.o -o %t.exe -Wl,-q
+# RUN: link_fdata %s %t.o %t.fdata
+# RUN: llvm-bolt %t.exe -o %t.bolt  \
+# RUN:   --data %t.fdata  | FileCheck %s
+
+# CHECK: BOLT-INFO: Inserted 1 stubs in the hot area and 0 stubs in the cold area. 
+
+  .section .text
+  .global _start
+  .global far_away_func
+
+  .align 4
+  .global _start
+  .type _start, %function
+_start:
+# FDATA: 0 [unknown] 0 1 _start 0 0 100
+    bl far_away_func
+    bl far_away_func
+    ret  
+  .space 0x8000000
+  .global far_away_func
+  .type far_away_func, %function
+far_away_func:
+    add x0, x0, #1
+    ret
+
+.reloc 0, R_AARCH64_NONE

clang/include/clang/AST/OpenACCClause.h

@@ -766,6 +766,29 @@ class OpenACCDetachClause final
          ArrayRef<Expr *> VarList, SourceLocation EndLoc);
 };
 
+class OpenACCDeleteClause final
+    : public OpenACCClauseWithVarList,
+      public llvm::TrailingObjects<OpenACCDeleteClause, Expr *> {
+
+  OpenACCDeleteClause(SourceLocation BeginLoc, SourceLocation LParenLoc,
+                      ArrayRef<Expr *> VarList, SourceLocation EndLoc)
+      : OpenACCClauseWithVarList(OpenACCClauseKind::Delete, BeginLoc, LParenLoc,
+                                 EndLoc) {
+    std::uninitialized_copy(VarList.begin(), VarList.end(),
+                            getTrailingObjects<Expr *>());
+    setExprs(MutableArrayRef(getTrailingObjects<Expr *>(), VarList.size()));
+  }
+
+public:
+  static bool classof(const OpenACCClause *C) {
+    return C->getClauseKind() == OpenACCClauseKind::Delete;
+  }
+  static OpenACCDeleteClause *
+  Create(const ASTContext &C, SourceLocation BeginLoc, SourceLocation LParenLoc,
+         ArrayRef<Expr *> VarList, SourceLocation EndLoc);
+};
+
+
 class OpenACCNoCreateClause final
     : public OpenACCClauseWithVarList,
       public llvm::TrailingObjects<OpenACCNoCreateClause, Expr *> {

clang/include/clang/Basic/OpenACCClauses.def

@@ -38,6 +38,7 @@ VISIT_CLAUSE(Create)
 CLAUSE_ALIAS(PCreate, Create, true)
 CLAUSE_ALIAS(PresentOrCreate, Create, true)
 VISIT_CLAUSE(Default)
+VISIT_CLAUSE(Delete)
 VISIT_CLAUSE(Detach)
 VISIT_CLAUSE(DevicePtr)
 VISIT_CLAUSE(DeviceType)

clang/include/clang/Basic/arm_sme.td

@@ -748,11 +748,28 @@ let SMETargetGuard = "sme2" in {
 let SMETargetGuard = "sme-f8f32" in {
   def SVDOT_LANE_FP8_ZA32_VG1x2 : Inst<"svdot_lane_za32[_mf8]_vg1x2_fpm", "vm2di>", "m", MergeNone, "aarch64_sme_fp8_fdot_lane_za32_vg1x2", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_3>]>;
   def SVDOT_LANE_FP8_ZA32_VG1x4 : Inst<"svdot_lane_za32[_mf8]_vg1x4_fpm", "vm4di>", "m", MergeNone, "aarch64_sme_fp8_fdot_lane_za32_vg1x4", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_3>]>;
+
+  def SVVDOTB_LANE_FP8_ZA32_VG1x4 : Inst<"svvdotb_lane_za32[_mf8]_vg1x4_fpm", "vm2di>", "m", MergeNone, "aarch64_sme_fp8_fvdotb_lane_za32_vg1x4", [IsOverloadNone, IsStreaming, IsInOutZA, SetsFPMR], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVVDOTT_LANE_FP8_ZA32_VG1x4 : Inst<"svvdott_lane_za32[_mf8]_vg1x4_fpm", "vm2di>", "m", MergeNone, "aarch64_sme_fp8_fvdott_lane_za32_vg1x4", [IsOverloadNone, IsStreaming, IsInOutZA, SetsFPMR], [ImmCheck<3, ImmCheck0_3>]>;
+
+  def SVDOT_SINGLE_FP8_ZA32_VG1x2 : Inst<"svdot[_single]_za32[_mf8]_vg1x2_fpm", "vm2d>", "m", MergeNone, "aarch64_sme_fp8_fdot_single_za32_vg1x2", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
+  def SVDOT_SINGLE_FP8_ZA32_VG1x4 : Inst<"svdot[_single]_za32[_mf8]_vg1x4_fpm", "vm4d>", "m", MergeNone, "aarch64_sme_fp8_fdot_single_za32_vg1x4", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
+
+  def SVDOT_MULTI_FP8_ZA32_VG1x2 : Inst<"svdot_za32[_mf8]_vg1x2_fpm", "vm22>", "m", MergeNone, "aarch64_sme_fp8_fdot_multi_za32_vg1x2", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
+  def SVDOT_MULTI_FP8_ZA32_VG1x4 : Inst<"svdot_za32[_mf8]_vg1x4_fpm", "vm44>", "m", MergeNone, "aarch64_sme_fp8_fdot_multi_za32_vg1x4", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
 }
 
 let SMETargetGuard = "sme-f8f16" in {
   def SVDOT_LANE_FP8_ZA16_VG1x2 : Inst<"svdot_lane_za16[_mf8]_vg1x2_fpm", "vm2di>", "m", MergeNone, "aarch64_sme_fp8_fdot_lane_za16_vg1x2", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_7>]>;
   def SVDOT_LANE_FP8_ZA16_VG1x4 : Inst<"svdot_lane_za16[_mf8]_vg1x4_fpm", "vm4di>", "m", MergeNone, "aarch64_sme_fp8_fdot_lane_za16_vg1x4", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_7>]>;
+
+  def SVVDOT_LANE_FP8_ZA16_VG1x2 : Inst<"svvdot_lane_za16[_mf8]_vg1x2_fpm", "vm2di>", "m", MergeNone, "aarch64_sme_fp8_fvdot_lane_za16_vg1x2", [IsOverloadNone, IsStreaming, IsInOutZA, SetsFPMR], [ImmCheck<3, ImmCheck0_7>]>;
+
+  def SVDOT_SINGLE_FP8_ZA16_VG1x2 : Inst<"svdot[_single]_za16[_mf8]_vg1x2_fpm", "vm2d>", "m", MergeNone, "aarch64_sme_fp8_fdot_single_za16_vg1x2", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
+  def SVDOT_SINGLE_FP8_ZA16_VG1x4 : Inst<"svdot[_single]_za16[_mf8]_vg1x4_fpm", "vm4d>", "m", MergeNone, "aarch64_sme_fp8_fdot_single_za16_vg1x4", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
+
+  def SVDOT_MULTI_FP8_ZA16_VG1x2 : Inst<"svdot_za16[_mf8]_vg1x2_fpm", "vm22>", "m", MergeNone, "aarch64_sme_fp8_fdot_multi_za16_vg1x2", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
+  def SVDOT_MULTI_FP8_ZA16_VG1x4 : Inst<"svdot_za16[_mf8]_vg1x4_fpm", "vm44>", "m", MergeNone, "aarch64_sme_fp8_fdot_multi_za16_vg1x4", [IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
 }
 
 ////////////////////////////////////////////////////////////////////////////////

clang/include/clang/Sema/SemaOpenACC.h

@@ -399,6 +399,7 @@ class SemaOpenACC : public SemaBase {
               ClauseKind == OpenACCClauseKind::PCreate ||
               ClauseKind == OpenACCClauseKind::PresentOrCreate ||
               ClauseKind == OpenACCClauseKind::Attach ||
+              ClauseKind == OpenACCClauseKind::Delete ||
               ClauseKind == OpenACCClauseKind::Detach ||
               ClauseKind == OpenACCClauseKind::DevicePtr ||
               ClauseKind == OpenACCClauseKind::Reduction ||
@@ -536,6 +537,7 @@ class SemaOpenACC : public SemaBase {
               ClauseKind == OpenACCClauseKind::PCreate ||
               ClauseKind == OpenACCClauseKind::PresentOrCreate ||
               ClauseKind == OpenACCClauseKind::Attach ||
+              ClauseKind == OpenACCClauseKind::Delete ||
               ClauseKind == OpenACCClauseKind::Detach ||
               ClauseKind == OpenACCClauseKind::DevicePtr ||
               ClauseKind == OpenACCClauseKind::FirstPrivate) &&
@@ -573,6 +575,7 @@ class SemaOpenACC : public SemaBase {
               ClauseKind == OpenACCClauseKind::PCreate ||
               ClauseKind == OpenACCClauseKind::PresentOrCreate ||
               ClauseKind == OpenACCClauseKind::Attach ||
+              ClauseKind == OpenACCClauseKind::Delete ||
               ClauseKind == OpenACCClauseKind::Detach ||
               ClauseKind == OpenACCClauseKind::DevicePtr ||
               ClauseKind == OpenACCClauseKind::FirstPrivate) &&

clang/include/clang/Serialization/ASTBitCodes.h

@@ -724,11 +724,9 @@ enum ASTRecordTypes {
   /// Record code for vtables to emit.
   VTABLES_TO_EMIT = 70,
 
-  /// Record code for the FunctionDecl to lambdas mapping. These lambdas have to
-  /// be loaded right after the function they belong to. It is required to have
-  /// canonical declaration for the lambda class from the same module as
-  /// enclosing function.
-  FUNCTION_DECL_TO_LAMBDAS_MAP = 71,
+  /// Record code for related declarations that have to be deserialized together
+  /// from the same module.
+  RELATED_DECLS_MAP = 71,
 
   /// Record code for Sema's vector of functions/blocks with effects to
   /// be verified.

clang/include/clang/Serialization/ASTReader.h

@@ -537,17 +537,14 @@ class ASTReader
   /// namespace as if it is not delayed.
   DelayedNamespaceOffsetMapTy DelayedNamespaceOffsetMap;
 
-  /// Mapping from FunctionDecl IDs to the corresponding lambda IDs.
-  ///
-  /// These lambdas have to be loaded right after the function they belong to.
-  /// It is required to have canonical declaration for lambda class from the
-  /// same module as enclosing function. This is required to correctly resolve
-  /// captured variables in the lambda. Without this, due to lazy
-  /// deserialization, canonical declarations for the function and lambdas can
-  /// be selected from different modules and DeclRefExprs may refer to the AST
-  /// nodes that don't exist in the function.
-  llvm::DenseMap<GlobalDeclID, SmallVector<GlobalDeclID, 4>>
-      FunctionToLambdasMap;
+  /// Mapping from main decl ID to the related decls IDs.
+  ///
+  /// These related decls have to be loaded right after the main decl.
+  /// It is required to have canonical declaration for related decls from the
+  /// same module as the enclosing main decl. Without this, due to lazy
+  /// deserialization, canonical declarations for the main decl and related can
+  /// be selected from different modules.
+  llvm::DenseMap<GlobalDeclID, SmallVector<GlobalDeclID, 4>> RelatedDeclsMap;
 
   struct PendingUpdateRecord {
     Decl *D;